{"meta":{"query_hash":"dc115bc2231e","filters":{"topic":"Advanced Clustering Algorithms Research"},"cohort_total":421,"direct_labels_cover":0,"predictions_cover":421,"exported":421,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/dc115bc2231e","api":"https://metacan.xera.ac/api/v1/cohort?topic=Advanced+Clustering+Algorithms+Research"},"results":[{"id":"W115127865","doi":"","title":"A New Cluster Validity Index for Fuzzy Clustering Based on Similarity Measure.","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster (spacecraft); Cluster analysis; Measure (data warehouse); Index (typography); Fuzzy clustering; Fuzzy logic; Similarity (geometry); Data mining; Mathematics; Computer science; Artificial intelligence; Pattern recognition (psychology); Statistics","score_opus":0.06919300065919722,"score_gpt":0.33919787417475383,"score_spread":0.27000487351555663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W115127865","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027180323,0.0031365212,0.95658743,0.0003629101,0.00042774822,0.00046512735,0.0006864746,0.0005563186,0.010597093],"genre_scores_gemma":[0.29704,0.0011457169,0.69607556,0.00024710433,0.00023499469,0.00094477076,0.0019747694,0.00024337173,0.0020936672],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98643714,0.0032342728,0.0011192878,0.0013308208,0.0075004995,0.0003780795],"domain_scores_gemma":[0.98580325,0.0070036533,0.0012590061,0.0011672059,0.004485324,0.00028167406],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009896526,0.0010186508,0.00173019,0.011164378,0.0021447325,0.003991404,0.0022118946,0.0016934236,0.0016759982],"category_scores_gemma":[0.03821846,0.0003966527,0.0014460201,0.009094245,0.002296689,0.0060589723,0.002410237,0.002008398,0.00082834554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005703561,0.00023724226,0.02780275,0.0017695865,0.0011636945,0.0002736682,0.0015656473,0.07360561,0.013264014,0.15552433,0.020282336,0.7039407],"study_design_scores_gemma":[0.000118161784,0.0006092889,0.027544271,0.000678058,0.0005232378,0.0015430594,0.001722725,0.7263691,0.03257446,0.15448225,0.053337667,0.00049765245],"about_ca_topic_score_codex":0.0033219322,"about_ca_topic_score_gemma":0.0028206366,"teacher_disagreement_score":0.011164378,"about_ca_system_score_codex":0.0031936713,"about_ca_system_score_gemma":0.0023707468,"threshold_uncertainty_score":0.05233848},"labels":[],"label_agreement":null},{"id":"W118196362","doi":"","title":"Type-II Fuzzy Possibilistic C-Mean Clustering","year":2009,"lang":"en","type":"article","venue":"European Society for Fuzzy Logic and Technology Conference","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Mahalanobis distance; Fuzzy logic; Cluster analysis; Fuzzy clustering; Mathematics; Data mining; Fuzzy number; Fuzzy classification; Fuzzy set operations; Euclidean distance; Artificial intelligence; Fuzzy set; Pattern recognition (psychology); Computer science","score_opus":0.034835140925025716,"score_gpt":0.28989406581776006,"score_spread":0.25505892489273435,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W118196362","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060095284,0.00035062092,0.9907821,0.000110755725,0.000044775825,0.00011286729,0.00008557894,0.00021626575,0.0022874654],"genre_scores_gemma":[0.19513355,0.00062983885,0.80000466,0.00011681766,0.00007235273,0.00042262682,0.00037914855,0.000092746406,0.0031482782],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99662787,0.0006017101,0.0002108678,0.00063642586,0.0017474812,0.00017559467],"domain_scores_gemma":[0.9975206,0.0007555301,0.00020941145,0.00030021215,0.00114373,0.00007047865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022665164,0.0012835457,0.0015604204,0.0033458124,0.0021477062,0.002596996,0.002915725,0.0025482653,0.002385359],"category_scores_gemma":[0.007445004,0.00062990893,0.0019515677,0.0034260612,0.0014496323,0.0018265019,0.0012393792,0.0013585603,0.001088218],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036541885,0.00011630393,0.0021983776,0.0008461162,0.00029708512,0.00032765244,0.0005027985,0.5741365,0.01650617,0.06530795,0.006581389,0.33281425],"study_design_scores_gemma":[0.0000146968,0.00004911124,0.0005899757,0.00004221357,0.000035013338,0.0002615396,0.00006508249,0.96193224,0.0070278333,0.024679752,0.0052230605,0.00007946212],"about_ca_topic_score_codex":0.0073311883,"about_ca_topic_score_gemma":0.0050946604,"teacher_disagreement_score":0.0073311883,"about_ca_system_score_codex":0.0018358992,"about_ca_system_score_gemma":0.0025858078,"threshold_uncertainty_score":0.014577031},"labels":[],"label_agreement":null},{"id":"W129927049","doi":"10.1007/978-3-642-20841-6_17","title":"Spatial Entropy-Based Clustering for Mining Data with Spatial Correlation","year":2011,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Spatial analysis; Cluster analysis; Spatial correlation; Computer science; Data mining; Entropy (arrow of time); Similarity (geometry); Spatial dependence; Pattern recognition (psychology); Spatial relation; Correlation; Spatial database; Artificial intelligence; Mathematics; Statistics","score_opus":0.045589914004622825,"score_gpt":0.2901130431364683,"score_spread":0.24452312913184548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W129927049","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02443705,0.00080072676,0.97241706,0.000099722914,0.000051818773,0.00006413335,0.00038052857,0.0005079478,0.0012409902],"genre_scores_gemma":[0.36051196,0.00089513336,0.63351375,0.00009663956,0.00018315649,0.00021583312,0.0022320116,0.00015373321,0.0021978319],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986129,0.00024924244,0.0001352977,0.00023301312,0.0007007759,0.00006866906],"domain_scores_gemma":[0.9984034,0.0008230019,0.0001504212,0.00021784536,0.0003619739,0.000043372936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013716768,0.0006093633,0.0011737109,0.0030252882,0.00069445104,0.00117137,0.0012817913,0.0006616827,0.0010482151],"category_scores_gemma":[0.00426442,0.00040361428,0.0011310988,0.004393162,0.00070352684,0.0015286766,0.0013207934,0.00079613226,0.00056898413],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051365595,0.00025110404,0.012454237,0.0006915059,0.00068091234,0.00030236808,0.0004714806,0.23062825,0.023350922,0.038528923,0.007301982,0.68482476],"study_design_scores_gemma":[0.000017892125,0.000075942575,0.0058080642,0.000037542704,0.000100568235,0.00028821762,0.000087125074,0.9535387,0.008151526,0.029421365,0.0024300222,0.000043095333],"about_ca_topic_score_codex":0.0019508861,"about_ca_topic_score_gemma":0.003035348,"teacher_disagreement_score":0.0030252882,"about_ca_system_score_codex":0.0007191467,"about_ca_system_score_gemma":0.00087149354,"threshold_uncertainty_score":0.007254243},"labels":[],"label_agreement":null},{"id":"W1453103518","doi":"10.1016/j.neucom.2015.08.018","title":"Applying subclustering and Lp distance in Weighted K-Means with distributed centroids","year":2015,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Categorical variable; Silhouette; Cluster analysis; Centroid; Pattern recognition (psychology); Feature (linguistics); Mathematics; k-means clustering; Computer science; Scaling; Metric (unit); Data mining; Multidimensional scaling; Fuzzy clustering; Minkowski space; Artificial intelligence; Algorithm; Statistics","score_opus":0.022034390108879854,"score_gpt":0.2616475139197036,"score_spread":0.23961312381082372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1453103518","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006484979,0.000073830495,0.9928181,0.000039089213,0.000034190307,0.000028764887,0.00001667156,0.00015790407,0.00034653378],"genre_scores_gemma":[0.117916286,0.00013444584,0.87962806,0.00004425511,0.00006628732,0.00014983422,0.00014496596,0.00014516464,0.0017708021],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99692327,0.0010761005,0.00024178297,0.00058194227,0.0010258636,0.00015110362],"domain_scores_gemma":[0.9967366,0.0009688223,0.00018256786,0.0005497484,0.001436524,0.00012569205],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027287873,0.00076339487,0.0012279745,0.001803259,0.0013210356,0.001650648,0.0023635444,0.0013833354,0.0020920378],"category_scores_gemma":[0.013695436,0.00054092903,0.0010668397,0.0029680056,0.0009561996,0.0028636886,0.0026817545,0.0014307015,0.0010900286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004545108,0.00017519837,0.0017256913,0.00028545753,0.00015232708,0.000088134424,0.000578504,0.26479867,0.016054796,0.035557862,0.0023681934,0.67776066],"study_design_scores_gemma":[0.000025479949,0.00009544814,0.00041455246,0.000015264755,0.0000357457,0.000067422654,0.00009370095,0.97262776,0.0074409805,0.016628744,0.002527952,0.000027092528],"about_ca_topic_score_codex":0.0061033093,"about_ca_topic_score_gemma":0.0059672683,"teacher_disagreement_score":0.0061033093,"about_ca_system_score_codex":0.0010309921,"about_ca_system_score_gemma":0.0020472577,"threshold_uncertainty_score":0.014431417},"labels":[],"label_agreement":null},{"id":"W1497689567","doi":"10.1007/978-3-540-45231-7_29","title":"Refined Shared Nearest Neighbors Graph for Combining Multiple Data Clusterings","year":2003,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; k-nearest neighbors algorithm; Graph; Theoretical computer science; Data mining; Artificial intelligence","score_opus":0.0592828658569979,"score_gpt":0.30986915558395806,"score_spread":0.2505862897269602,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1497689567","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00455432,0.0001721416,0.9940122,0.000053053314,0.00003227164,0.000049909468,0.00012845412,0.00052692735,0.00047073804],"genre_scores_gemma":[0.11971974,0.0002814393,0.87577516,0.000095468225,0.00006170974,0.00020376248,0.0012091711,0.00029310418,0.0023604715],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961541,0.0013121892,0.00017234893,0.0007838504,0.0014303137,0.00014712207],"domain_scores_gemma":[0.9943453,0.0018170405,0.00029749825,0.0021959327,0.0011814781,0.00016280089],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023970169,0.0012964063,0.002492103,0.0040811235,0.0013178977,0.0018427961,0.004517547,0.002029028,0.0026716855],"category_scores_gemma":[0.010299712,0.0011724164,0.001838948,0.0060525397,0.0011374973,0.004142532,0.0040934,0.0015231234,0.0011532789],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047870332,0.00017174076,0.0014666975,0.0003617087,0.00039190063,0.00030930646,0.00047793242,0.40619913,0.009096521,0.057398513,0.0083748745,0.515273],"study_design_scores_gemma":[0.000027903368,0.000052191095,0.0003185902,0.000017029193,0.00006142012,0.00009400482,0.00008347784,0.9345092,0.0022499806,0.059597306,0.0029608945,0.000028019793],"about_ca_topic_score_codex":0.008318147,"about_ca_topic_score_gemma":0.013786707,"teacher_disagreement_score":0.008318147,"about_ca_system_score_codex":0.0012370304,"about_ca_system_score_gemma":0.0013472218,"threshold_uncertainty_score":0.016539454},"labels":[],"label_agreement":null},{"id":"W1503257002","doi":"10.1007/978-3-642-04843-2_17","title":"A Point Symmetry-Based Automatic Clustering Approach Using Differential Evolution","year":2009,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Cluster analysis; Computer science; Point (geometry); Cluster (spacecraft); Symmetry (geometry); Measure (data warehouse); Differential evolution; Representation (politics); Data mining; Algorithm; Artificial intelligence; Pattern recognition (psychology); Mathematics; Geometry","score_opus":0.026122202958597254,"score_gpt":0.27955413089078135,"score_spread":0.2534319279321841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1503257002","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005505902,0.00004395448,0.9928269,0.000029649445,0.000026309244,0.000037571597,0.000020537356,0.00035157226,0.0011575301],"genre_scores_gemma":[0.102523744,0.00008224874,0.89316636,0.000050853523,0.000028367915,0.00010752424,0.00014533704,0.0002579488,0.0036375122],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99945325,0.00007974294,0.00002524431,0.000121468875,0.00027379152,0.000046439185],"domain_scores_gemma":[0.9993401,0.00016549873,0.000052143627,0.00014815699,0.00025118658,0.000042883872],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00071326905,0.0005458546,0.0014746661,0.0017326833,0.00085734786,0.0010904726,0.0025070307,0.001154053,0.0027489632],"category_scores_gemma":[0.0013420116,0.0005494593,0.0015132827,0.0014387908,0.00072012015,0.0010308169,0.0017160478,0.00090601697,0.0012593234],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001631865,0.00015802405,0.0008627849,0.00024662478,0.00016733396,0.00020265974,0.00036498276,0.24618185,0.07794875,0.0705992,0.0050948,0.59800977],"study_design_scores_gemma":[0.00000904793,0.000022629438,0.00018284582,0.000004486438,0.000012978298,0.00007962607,0.000019551639,0.98663914,0.003992744,0.0076181665,0.0014021777,0.000016573586],"about_ca_topic_score_codex":0.0025791533,"about_ca_topic_score_gemma":0.002666631,"teacher_disagreement_score":0.0027489632,"about_ca_system_score_codex":0.0007978114,"about_ca_system_score_gemma":0.00076860416,"threshold_uncertainty_score":0.009196222},"labels":[],"label_agreement":null},{"id":"W151377110","doi":"10.1007/978-3-642-37456-2_14","title":"Density-Based Clustering Based on Hierarchical Density Estimates","year":2013,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2412,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Cluster analysis; Hierarchical clustering; Data mining; Artificial intelligence","score_opus":0.021558810818848025,"score_gpt":0.2774625965106832,"score_spread":0.2559037856918352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W151377110","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019377252,0.00024707397,0.99643743,0.00005330732,0.000029789328,0.000035853853,0.000050554994,0.00027923397,0.00092909305],"genre_scores_gemma":[0.14318849,0.0010184951,0.8486137,0.00013298237,0.00015346048,0.00026841715,0.00090742024,0.0004453143,0.0052716946],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99799407,0.0006995166,0.000106930944,0.00034709254,0.00073526875,0.000117125135],"domain_scores_gemma":[0.994879,0.0025980812,0.0002801873,0.0008395088,0.0012733701,0.00012981296],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023546899,0.001062691,0.0021304556,0.0028733653,0.0013402656,0.0019567758,0.0040763356,0.0019019056,0.0034579474],"category_scores_gemma":[0.013792427,0.0014000189,0.0018553176,0.0038616979,0.0011402343,0.0026602764,0.0030547744,0.002179965,0.0026145333],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017354921,0.00008608483,0.0012444669,0.00036243827,0.00021758505,0.000112129834,0.00041564179,0.59596145,0.005542593,0.11221188,0.009358329,0.27431384],"study_design_scores_gemma":[0.0000073953347,0.00001371658,0.0002926247,0.000019007877,0.000020442187,0.00006084052,0.000022948241,0.97012097,0.000895777,0.026969763,0.0015502952,0.000026166681],"about_ca_topic_score_codex":0.008361296,"about_ca_topic_score_gemma":0.008475352,"teacher_disagreement_score":0.008361296,"about_ca_system_score_codex":0.0015421081,"about_ca_system_score_gemma":0.001348675,"threshold_uncertainty_score":0.016625285},"labels":[],"label_agreement":null},{"id":"W1516498690","doi":"10.1007/3-540-44938-8_17","title":"Finding Natural Clusters Using Multi-clusterer Combiner Based on Shared Nearest Neighbors","year":2003,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Graph; Computer science; k-nearest neighbors algorithm; Set (abstract data type); Pattern recognition (psychology); Data mining; Artificial intelligence; Theoretical computer science; Mathematics","score_opus":0.04392408286800724,"score_gpt":0.3073721529406045,"score_spread":0.26344807007259724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1516498690","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05451648,0.00022882676,0.94067913,0.00007684666,0.000053383912,0.00016030125,0.00014838159,0.0019161789,0.002220484],"genre_scores_gemma":[0.18879572,0.00009087661,0.8065794,0.000055436707,0.000039927578,0.00016842457,0.00050789845,0.0002989019,0.0034634164],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983191,0.00017346535,0.000085958505,0.0005132315,0.00072397996,0.0001843311],"domain_scores_gemma":[0.9980077,0.00047055443,0.00016575004,0.00044431438,0.00077041495,0.00014127414],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012212592,0.0010462252,0.0025347208,0.0037449952,0.002130649,0.002081462,0.0031912464,0.0016832872,0.0040645283],"category_scores_gemma":[0.002756516,0.0007760948,0.0017184052,0.004458682,0.00097246334,0.0033036312,0.0032627783,0.0008765737,0.0015357214],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012502684,0.00045883682,0.007624807,0.00036327488,0.00045692027,0.0005066408,0.00096353004,0.045784436,0.09395356,0.010713917,0.0051931944,0.83273065],"study_design_scores_gemma":[0.0001140703,0.0003269617,0.004646148,0.000022596747,0.00026014488,0.0005894152,0.00075072463,0.9307965,0.042057116,0.016266478,0.004057369,0.000112431684],"about_ca_topic_score_codex":0.003024099,"about_ca_topic_score_gemma":0.0072874143,"teacher_disagreement_score":0.0040645283,"about_ca_system_score_codex":0.0008891238,"about_ca_system_score_gemma":0.00082935137,"threshold_uncertainty_score":0.01359719},"labels":[],"label_agreement":null},{"id":"W1521374378","doi":"","title":"Cluster Assessment: A Complimentary Aspect of Cluster Learning","year":2010,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Cluster (spacecraft); Computer science; Artificial intelligence; Computer network","score_opus":0.014442518119782487,"score_gpt":0.3244753965582688,"score_spread":0.3100328784384863,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1521374378","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0052329795,0.00007802948,0.98986816,0.00043379943,0.00007209874,0.00013335292,0.00009069006,0.0010357659,0.0030551348],"genre_scores_gemma":[0.15377833,0.00016649957,0.8395248,0.0003337407,0.00026638698,0.00022798806,0.00039967863,0.0006946403,0.0046079983],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9886635,0.00338449,0.0006565474,0.0017402847,0.0052307993,0.00032435948],"domain_scores_gemma":[0.9577859,0.0174618,0.0014142917,0.011693924,0.010620372,0.0010236647],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009097907,0.0011614332,0.0027923512,0.0033760476,0.0034124074,0.0071139713,0.005694921,0.0025472376,0.008953871],"category_scores_gemma":[0.06515795,0.0008021806,0.0011602538,0.0046885842,0.0026916745,0.008557975,0.008031182,0.0037846847,0.0031636083],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033348953,0.00041443307,0.012583796,0.00058308477,0.00033025278,0.00015679923,0.0013042263,0.042037856,0.011589177,0.12852193,0.0140418075,0.78810316],"study_design_scores_gemma":[0.000072459676,0.0002987239,0.0064596925,0.000107202395,0.0001788245,0.00056610745,0.0008622058,0.67093426,0.01980513,0.2728005,0.027746433,0.00016842142],"about_ca_topic_score_codex":0.003373071,"about_ca_topic_score_gemma":0.006045923,"teacher_disagreement_score":0.009097907,"about_ca_system_score_codex":0.0011176793,"about_ca_system_score_gemma":0.0034516712,"threshold_uncertainty_score":0.048114896},"labels":[],"label_agreement":null},{"id":"W153300101","doi":"10.1007/978-3-540-72432-2_82","title":"New Cluster Validity Index with Fuzzy Functions","year":2007,"lang":"en","type":"book-chapter","venue":"Advances in soft computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Mathematics; Fuzzy logic; Measure (data warehouse); Cluster (spacecraft); Fuzzy clustering; Compact space; Cluster analysis; Fuzzy set; Index (typography); FLAME clustering; Data mining; Pattern recognition (psychology); Statistics; Artificial intelligence; Computer science; Mathematical analysis","score_opus":0.04084506286960581,"score_gpt":0.3208448999270594,"score_spread":0.27999983705745357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W153300101","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016812943,0.0029051933,0.9698083,0.00031842664,0.0004012971,0.00014228675,0.0005674225,0.00044124873,0.008602863],"genre_scores_gemma":[0.21712336,0.0013668453,0.773654,0.00017779441,0.0005750363,0.00039284327,0.0015261223,0.0002898062,0.0048942217],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9916535,0.0016948119,0.00071506907,0.0010190472,0.0046082204,0.0003093801],"domain_scores_gemma":[0.99144214,0.0037752083,0.0004324213,0.00083543337,0.0032565673,0.0002581389],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0072972733,0.0009495712,0.0022444811,0.009846202,0.0017972447,0.004370446,0.0026367665,0.0015320969,0.0025384221],"category_scores_gemma":[0.01883527,0.0005843265,0.0017729553,0.008836757,0.0019045051,0.006894037,0.0024426093,0.0019220371,0.0008815508],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007506954,0.0001788767,0.0063081197,0.00091640564,0.0005979885,0.000188269,0.0005206013,0.040196855,0.012052732,0.19092782,0.023014152,0.7243475],"study_design_scores_gemma":[0.000073259966,0.00022777698,0.008944501,0.00024271534,0.00032226086,0.0007362933,0.00031542423,0.73273605,0.017359046,0.21516258,0.023660451,0.00021965617],"about_ca_topic_score_codex":0.001677233,"about_ca_topic_score_gemma":0.0015909161,"teacher_disagreement_score":0.009846202,"about_ca_system_score_codex":0.002907306,"about_ca_system_score_gemma":0.0014682813,"threshold_uncertainty_score":0.03859216},"labels":[],"label_agreement":null},{"id":"W1536355413","doi":"10.1109/ccece.2015.7129370","title":"A tweet grouping methodology utilizing inter and intra cosine similarity","year":2015,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Popularity; Cosine similarity; Computer science; Similarity (geometry); Publication; Cluster analysis; Subject (documents); Information retrieval; Microblogging; Social media; Feeling; World Wide Web; Artificial intelligence; Advertising; Psychology","score_opus":0.2558543091809517,"score_gpt":0.40551641913127173,"score_spread":0.14966210995032003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1536355413","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024142683,0.00035894066,0.9678889,0.0001726808,0.00034526046,0.0012566036,0.00089696626,0.0018054498,0.0031325282],"genre_scores_gemma":[0.16653122,0.00041000854,0.8224592,0.000077513956,0.00031194944,0.0008319338,0.0025076398,0.0002670354,0.006603532],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99582195,0.0006397559,0.0004988382,0.0010845198,0.0017486977,0.00020624089],"domain_scores_gemma":[0.9956299,0.00086682284,0.00065101404,0.0006065415,0.0020493388,0.00019632063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002641212,0.0013879661,0.0015552094,0.012459571,0.0018468307,0.002028195,0.001648793,0.0010824611,0.0025152746],"category_scores_gemma":[0.0076912134,0.00044773155,0.0015793775,0.011285317,0.00094593223,0.0023641752,0.0013875682,0.0012108022,0.0030822235],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005952796,0.0006994996,0.013811606,0.0007730205,0.0005670177,0.0002655832,0.0012424194,0.013947397,0.061615996,0.018241933,0.016945079,0.8712953],"study_design_scores_gemma":[0.00021303968,0.0016605944,0.040479835,0.00014931189,0.000560938,0.0022701228,0.0021567543,0.7771358,0.08926575,0.028017256,0.05752166,0.0005689521],"about_ca_topic_score_codex":0.0043943073,"about_ca_topic_score_gemma":0.0048422217,"teacher_disagreement_score":0.012459571,"about_ca_system_score_codex":0.0010015831,"about_ca_system_score_gemma":0.001979713,"threshold_uncertainty_score":0.013968229},"labels":[],"label_agreement":null},{"id":"W1556829956","doi":"10.1007/978-3-540-25966-4_14","title":"A Probabilistic Model Using Information Theoretic Measures for Cluster Ensembles","year":2004,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Probabilistic logic; Computer science; Conditional probability; Cluster analysis; Probability distribution; Entropy (arrow of time); Conditional probability distribution; Data mining; Divergence (linguistics); Expectation–maximization algorithm; Statistical model; Artificial intelligence; Mathematics; Statistics","score_opus":0.03497427916053233,"score_gpt":0.29213932538523485,"score_spread":0.25716504622470254,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1556829956","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038813266,0.00018209862,0.9934809,0.00040990306,0.00004795713,0.000024624916,0.00010317313,0.00009101689,0.0017790782],"genre_scores_gemma":[0.43286678,0.0015152543,0.5458333,0.0007599997,0.00096642977,0.00084647955,0.0012473216,0.0005560414,0.015408434],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99484056,0.0023097822,0.00028518031,0.0009168094,0.001318803,0.00032889636],"domain_scores_gemma":[0.98877245,0.0074304887,0.00087030587,0.0014374079,0.001066137,0.00042319624],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055294847,0.0011502227,0.0023225746,0.0023512836,0.0018624973,0.005102461,0.006008395,0.0032965934,0.0047569717],"category_scores_gemma":[0.019787647,0.0014736545,0.0026586612,0.0034293383,0.0032259137,0.009740058,0.0039046102,0.0047112894,0.0014380798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000034621065,0.00004307585,0.00035786926,0.000072078095,0.00005897408,0.0000666274,0.00017592964,0.15797311,0.0004894873,0.82531005,0.0023425776,0.013075597],"study_design_scores_gemma":[0.000005281578,0.000011296922,0.00008328528,0.000014163543,0.000015927748,0.00004825698,0.000020958327,0.5396363,0.00013078684,0.45913437,0.00087770296,0.000021716407],"about_ca_topic_score_codex":0.002205008,"about_ca_topic_score_gemma":0.002559955,"teacher_disagreement_score":0.006008395,"about_ca_system_score_codex":0.0022693153,"about_ca_system_score_gemma":0.0013075324,"threshold_uncertainty_score":0.029243052},"labels":[],"label_agreement":null},{"id":"W1559950581","doi":"10.1007/11494683_24","title":"Cluster-Based Cumulative Ensembles","year":2005,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":21,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Jaccard index; Cluster (spacecraft); Computer science; Cluster analysis; Granularity; Similarity (geometry); Data mining; Hierarchical clustering; Computation; Measure (data warehouse); Single-linkage clustering; Pattern recognition (psychology); Algorithm; Artificial intelligence; Fuzzy clustering; CURE data clustering algorithm","score_opus":0.02871685949556528,"score_gpt":0.3041407387586494,"score_spread":0.2754238792630841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1559950581","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01768848,0.0005944172,0.9640229,0.00016630469,0.00018949098,0.00002861612,0.00008818009,0.00032040145,0.016901176],"genre_scores_gemma":[0.59013665,0.0015191761,0.3662959,0.00025648472,0.00062228,0.00018653645,0.0006884616,0.0006263632,0.039668128],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989705,0.00035855453,0.0000363565,0.00017232895,0.0003903849,0.000071877],"domain_scores_gemma":[0.99661094,0.0014665716,0.0001434914,0.0007875067,0.0007938527,0.00019758822],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015872031,0.0006382625,0.0013148478,0.0014253238,0.000991795,0.0015317036,0.0015566954,0.00095723144,0.005678073],"category_scores_gemma":[0.006462424,0.00055989047,0.00085573556,0.0015322785,0.0009816296,0.0016976115,0.0025638628,0.0016933478,0.0012864377],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010101675,0.000048160207,0.00082686124,0.000118213444,0.00009773563,0.00006860238,0.00016481873,0.26665586,0.0039922805,0.58109486,0.008331903,0.13849977],"study_design_scores_gemma":[0.000006515274,0.000024753455,0.00045287848,0.00001885843,0.000021690526,0.000103991275,0.000027714865,0.73662925,0.0013235005,0.25607607,0.005291846,0.00002289181],"about_ca_topic_score_codex":0.0007174029,"about_ca_topic_score_gemma":0.001495248,"teacher_disagreement_score":0.005678073,"about_ca_system_score_codex":0.00066727534,"about_ca_system_score_gemma":0.00039116282,"threshold_uncertainty_score":0.018995106},"labels":[],"label_agreement":null},{"id":"W1560541823","doi":"10.1007/978-3-540-24741-8_9","title":"LIMBO: Scalable Clustering of Categorical Data","year":2004,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":260,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Data mining; Tuple; Data stream clustering; Hierarchical clustering; Bottleneck; Scalability; Measure (data warehouse); CURE data clustering algorithm; Correlation clustering; Clustering high-dimensional data; Artificial intelligence; Machine learning; Database; Mathematics","score_opus":0.047484930229229186,"score_gpt":0.3097387021538155,"score_spread":0.2622537719245863,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1560541823","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0049286853,0.00055552984,0.9234459,0.00026835297,0.0002638773,0.00029759482,0.005706437,0.060987864,0.003545901],"genre_scores_gemma":[0.030527797,0.0003678072,0.9360351,0.00023914345,0.000099734134,0.00060860824,0.02082241,0.0039336355,0.0073657925],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983224,0.0003099332,0.00012155138,0.00027268898,0.00081742194,0.000155941],"domain_scores_gemma":[0.9978982,0.00044532676,0.00009649377,0.0009617906,0.00042930502,0.00016890652],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014816469,0.0014468796,0.002048129,0.0032568611,0.001583504,0.002625802,0.004145035,0.0009237397,0.01694855],"category_scores_gemma":[0.006111883,0.0013795405,0.0023952527,0.006372483,0.000685823,0.0029283338,0.004924891,0.0021473833,0.011295347],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00068073923,0.00022011917,0.0014646152,0.0006858487,0.00025457703,0.00019312384,0.00023665233,0.022781406,0.010761541,0.015642282,0.24374706,0.70333195],"study_design_scores_gemma":[0.0004898231,0.0002014556,0.0028556162,0.00015214818,0.00018688219,0.0005725932,0.0002563499,0.7180532,0.018633453,0.14802624,0.11042363,0.0001486605],"about_ca_topic_score_codex":0.0073229,"about_ca_topic_score_gemma":0.0141377775,"teacher_disagreement_score":0.01694855,"about_ca_system_score_codex":0.001009359,"about_ca_system_score_gemma":0.002159623,"threshold_uncertainty_score":0.05669856},"labels":[],"label_agreement":null},{"id":"W1561886765","doi":"10.1007/978-3-540-30076-2_22","title":"Novel Clustering Approach that Employs Genetic Algorithm with New Representation Scheme and Multiple Objectives","year":2004,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Computer science; Cluster analysis; Partition (number theory); Scheme (mathematics); Encoding (memory); Algorithm; Graph partition; Graph; Set (abstract data type); Theoretical computer science; Artificial intelligence; Mathematics","score_opus":0.03641534816727335,"score_gpt":0.2810127521842175,"score_spread":0.24459740401694416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1561886765","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0087873135,0.00021303612,0.987665,0.00008741735,0.000136,0.000047929767,0.000038635702,0.00053470413,0.002490026],"genre_scores_gemma":[0.09956707,0.00019230963,0.89397854,0.00010360742,0.00007088534,0.00014495237,0.00021686447,0.00012028898,0.0056055197],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99933213,0.00009348009,0.00003025904,0.00017400109,0.00031820397,0.00005194874],"domain_scores_gemma":[0.9996755,0.00004817812,0.000029593777,0.000045140787,0.00017640683,0.00002520116],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005374034,0.0008373927,0.0012451108,0.0014739921,0.00096322654,0.0009932054,0.0020465895,0.0014514896,0.0013086507],"category_scores_gemma":[0.0007096535,0.00039289743,0.0010731422,0.0021490024,0.00037836708,0.0009734283,0.0009264577,0.00080944074,0.0005169991],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014610619,0.00033897828,0.0014430432,0.00024617746,0.0003201617,0.00022063017,0.00018352439,0.33811224,0.06638648,0.032047376,0.008879343,0.55167603],"study_design_scores_gemma":[0.000023356448,0.00006539773,0.00051014055,0.000009482207,0.00005274899,0.00018700141,0.00003125751,0.98266184,0.0076265293,0.0045952364,0.004198275,0.00003877393],"about_ca_topic_score_codex":0.004681225,"about_ca_topic_score_gemma":0.005676563,"teacher_disagreement_score":0.004681225,"about_ca_system_score_codex":0.0007951894,"about_ca_system_score_gemma":0.0010868354,"threshold_uncertainty_score":0.009307981},"labels":[],"label_agreement":null},{"id":"W1567786505","doi":"10.1002/0471708607.ch10","title":"Fuzzy Clustering of Heterogeneous Patterns","year":2005,"lang":"en","type":"other","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Data mining; Representation (politics); Fuzzy logic; Computer science; Artificial intelligence; Mathematics; Pattern recognition (psychology); Theoretical computer science","score_opus":0.01856599385683777,"score_gpt":0.29215851760196254,"score_spread":0.27359252374512477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1567786505","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011478489,0.003155653,0.9748123,0.0009528111,0.0005246412,0.000068138055,0.0002020083,0.00019419039,0.008611769],"genre_scores_gemma":[0.3272975,0.008709008,0.64135116,0.00058055425,0.001003614,0.00021549327,0.0010221273,0.00020321461,0.019617278],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99859196,0.00032084348,0.00007622257,0.0004422498,0.0004514765,0.0001171753],"domain_scores_gemma":[0.9990338,0.00025597145,0.00011492108,0.00025013203,0.00029900277,0.00004611361],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010612346,0.0009988695,0.0014590682,0.0032225363,0.0014941677,0.003075111,0.0010829025,0.0019342235,0.002754414],"category_scores_gemma":[0.0028308895,0.0005274862,0.0016312243,0.0043789363,0.001319619,0.002454264,0.0012127339,0.0011284332,0.0012872706],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001641522,0.00006321884,0.0023891933,0.0008014561,0.00029358556,0.0008378591,0.0011672906,0.0756934,0.0222809,0.45739216,0.013888059,0.42502874],"study_design_scores_gemma":[0.000020746436,0.00011592431,0.0022947227,0.00021192327,0.0002009375,0.0015789563,0.00076767616,0.21936055,0.018374372,0.67758405,0.07937366,0.00011643731],"about_ca_topic_score_codex":0.0014276896,"about_ca_topic_score_gemma":0.0009823416,"teacher_disagreement_score":0.0032225363,"about_ca_system_score_codex":0.0010004424,"about_ca_system_score_gemma":0.00077758747,"threshold_uncertainty_score":0.009214461},"labels":[],"label_agreement":null},{"id":"W1568806324","doi":"10.1007/3-540-45683-x_46","title":"Efficiently Clustering Documents with Committees","year":2002,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Similarity (geometry); Set (abstract data type); Data mining; Task (project management); Group (periodic table); Cluster (spacecraft); Space (punctuation); Document clustering; Correlation clustering; Single-linkage clustering; CURE data clustering algorithm; Information retrieval; Artificial intelligence; Image (mathematics)","score_opus":0.018519313587672916,"score_gpt":0.269153790340296,"score_spread":0.2506344767526231,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1568806324","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010480861,0.0008877034,0.9777827,0.0002442684,0.00013065596,0.00020884957,0.00069334375,0.0064889584,0.00308279],"genre_scores_gemma":[0.06185454,0.0005124663,0.9210721,0.00015636733,0.00021298735,0.00028106503,0.00489845,0.00093529304,0.01007676],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963154,0.0012754695,0.00028796747,0.00072039256,0.0011567936,0.00024388773],"domain_scores_gemma":[0.9955687,0.0014312738,0.00021689729,0.0015393007,0.0010286226,0.00021520774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026857085,0.0017153273,0.0027241171,0.0037891583,0.0019459483,0.0031556468,0.0030639018,0.001808059,0.0068686786],"category_scores_gemma":[0.009371942,0.0015782635,0.0023427901,0.0063566905,0.0008368842,0.0037295371,0.0033813126,0.0020917093,0.0084011555],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006312073,0.00022958321,0.0009432637,0.00033837234,0.00024460597,0.00010537261,0.00016568904,0.048817456,0.018056806,0.012518083,0.04411806,0.87383145],"study_design_scores_gemma":[0.00017580552,0.0002090296,0.00085337163,0.000058988193,0.00017735663,0.0004228434,0.00015661388,0.8676834,0.030533655,0.07318313,0.026482513,0.00006330356],"about_ca_topic_score_codex":0.0029278968,"about_ca_topic_score_gemma":0.0065086894,"teacher_disagreement_score":0.0068686786,"about_ca_system_score_codex":0.00096733886,"about_ca_system_score_gemma":0.0018428242,"threshold_uncertainty_score":0.022978067},"labels":[],"label_agreement":null},{"id":"W1584777468","doi":"10.1007/11811305_30","title":"A Fuzzy Subspace Algorithm for Clustering High Dimensional Data","year":2006,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":57,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Computer science; FLAME clustering; Subspace topology; Dimension (graph theory); k-medians clustering; Cluster (spacecraft); Fuzzy logic; Clustering high-dimensional data; Data mining; Correlation clustering; Pattern recognition (psychology); Single-linkage clustering; Algorithm; CURE data clustering algorithm; Artificial intelligence; Mathematics; Combinatorics","score_opus":0.03486772328265359,"score_gpt":0.3000635860296849,"score_spread":0.26519586274703133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1584777468","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017675045,0.00021281575,0.9972517,0.000031750777,0.000039985716,0.000024714753,0.00004288311,0.0002628761,0.0003658338],"genre_scores_gemma":[0.019739883,0.00026662086,0.9778641,0.000041304556,0.00005801071,0.000096121825,0.00020819854,0.00006894937,0.0016568094],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984861,0.00035233962,0.000086810534,0.00022781927,0.00077047123,0.00007650796],"domain_scores_gemma":[0.99913114,0.00023630525,0.000040117047,0.00017590365,0.00036983573,0.000046755675],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015168152,0.0008224557,0.0017832291,0.0015478912,0.0012437257,0.0012760594,0.0017732735,0.0011730108,0.0037064971],"category_scores_gemma":[0.0026183734,0.0004793171,0.0012618036,0.0035805237,0.0008781232,0.0018425129,0.0016357306,0.0014005277,0.0019539273],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002025711,0.00008207828,0.00032351853,0.00016243396,0.0001358277,0.00005614708,0.00014321027,0.07365577,0.01387736,0.025153048,0.005893388,0.88031465],"study_design_scores_gemma":[0.000037587262,0.00014092904,0.0005516906,0.000023788743,0.000047111997,0.00027652446,0.00008000887,0.94546306,0.007377995,0.034610163,0.011317792,0.000073357965],"about_ca_topic_score_codex":0.004351097,"about_ca_topic_score_gemma":0.0043022446,"teacher_disagreement_score":0.004351097,"about_ca_system_score_codex":0.0004441468,"about_ca_system_score_gemma":0.0012042027,"threshold_uncertainty_score":0.012399435},"labels":[],"label_agreement":null},{"id":"W1586523483","doi":"10.1007/11533719_67","title":"On Approximate Balanced Bi-clustering","year":2005,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Cluster analysis; Computer science; Heuristic; Bounded function; Mathematical optimization; Cluster (spacecraft); Set (abstract data type); Quadratic programming; Quadratic equation; Constrained clustering; Heuristics; Algorithm; Correlation clustering; Theoretical computer science; Mathematics; CURE data clustering algorithm; Artificial intelligence","score_opus":0.021576937168322402,"score_gpt":0.28273569928771697,"score_spread":0.2611587621193946,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1586523483","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006847898,0.0027552655,0.975459,0.00073412625,0.00036147799,0.000116327916,0.00049857993,0.0008303783,0.0123969205],"genre_scores_gemma":[0.11567022,0.0031359387,0.8499073,0.0007492708,0.00055403623,0.00047049107,0.0027887241,0.0009422178,0.025781743],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99748373,0.0008483875,0.00012298193,0.00035425313,0.0009912925,0.00019924658],"domain_scores_gemma":[0.9957802,0.001511471,0.00016555937,0.0015007085,0.0008766672,0.00016536495],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028164606,0.0018653371,0.0029124927,0.0023743596,0.0019655423,0.0024854324,0.004219633,0.0020248909,0.012657778],"category_scores_gemma":[0.012119804,0.00096735405,0.0010803337,0.0081394045,0.0012548412,0.0066671884,0.0049754777,0.0030108641,0.005000431],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058925786,0.00016916775,0.00080202246,0.0005185973,0.00013342983,0.00008272853,0.0002542191,0.15501676,0.0038014809,0.2808426,0.054229546,0.5035602],"study_design_scores_gemma":[0.000044816104,0.00008162953,0.00044483689,0.00007464697,0.000050356484,0.00024212596,0.0001351834,0.55049527,0.002101779,0.41646355,0.029829765,0.000036047273],"about_ca_topic_score_codex":0.0041500176,"about_ca_topic_score_gemma":0.0064614015,"teacher_disagreement_score":0.012657778,"about_ca_system_score_codex":0.002396927,"about_ca_system_score_gemma":0.001537458,"threshold_uncertainty_score":0.04234451},"labels":[],"label_agreement":null},{"id":"W1633233002","doi":"10.48550/arxiv.1311.2334","title":"Embed and Conquer: Scalable Embeddings for Kernel k-Means on MapReduce","year":2013,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Kernel (algebra); Computer science; Scalability; Embedding; Divide and conquer algorithms; Benchmark (surveying); Cluster analysis; Parallel computing; Computation; Scaling; Theoretical computer science; Algorithm; Mathematics; Artificial intelligence; Discrete mathematics","score_opus":0.07124242219138041,"score_gpt":0.2299106971668386,"score_spread":0.15866827497545818,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1633233002","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017761622,0.00023766859,0.9726638,0.00022232898,0.00011702078,0.000117884614,0.0002343432,0.0068999287,0.0017454422],"genre_scores_gemma":[0.24100812,0.00022391233,0.7521549,0.00016031589,0.0000831309,0.00033148064,0.0014730317,0.00094665383,0.00361855],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987476,0.00025160713,0.00007710277,0.0002873485,0.0004949836,0.00014137093],"domain_scores_gemma":[0.9987435,0.0002582957,0.00006288329,0.0005365652,0.00029814208,0.00010064844],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010374448,0.0012799643,0.0011355117,0.0007134373,0.0008591724,0.0014591016,0.0023858913,0.000895149,0.0024993455],"category_scores_gemma":[0.004614611,0.0004916476,0.0010494491,0.001286787,0.00077304523,0.002850072,0.0029286277,0.0017808734,0.001714875],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076363015,0.00068286335,0.0023841367,0.00041313493,0.000235571,0.0003135208,0.00068257586,0.28527352,0.02086028,0.05081484,0.04022171,0.5973542],"study_design_scores_gemma":[0.000054171982,0.000056385365,0.00035910567,0.000007622172,0.000014047506,0.000078619254,0.00016385158,0.9539624,0.007194556,0.03146478,0.006619218,0.000025201758],"about_ca_topic_score_codex":0.0054999785,"about_ca_topic_score_gemma":0.0084217535,"teacher_disagreement_score":0.0054999785,"about_ca_system_score_codex":0.0008702418,"about_ca_system_score_gemma":0.0016737435,"threshold_uncertainty_score":0.010935962},"labels":[],"label_agreement":null},{"id":"W1698234622","doi":"10.1109/wiri.2005.7","title":"An Efficient Technique for Mining Usage Profiles Using Relational Fuzzy Subtractive Clustering","year":2005,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Data mining; Computer science; Subtractive color; Fuzzy clustering; Correlation clustering; CURE data clustering algorithm; Data stream clustering; Web mining; Machine learning; Web page","score_opus":0.05623948391364995,"score_gpt":0.35776641058514536,"score_spread":0.3015269266714954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1698234622","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006019624,0.000079388694,0.9924688,0.00006459434,0.000014482541,0.00007408222,0.000069440466,0.000641903,0.00056772813],"genre_scores_gemma":[0.09482382,0.00013786704,0.9030117,0.00006163033,0.000030920182,0.00019095709,0.0005376898,0.0001029437,0.0011026188],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970515,0.00045308282,0.00020662081,0.0004369743,0.0017258269,0.0001261158],"domain_scores_gemma":[0.99736315,0.0007716724,0.00026144084,0.00042963578,0.0011030441,0.00007112641],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018287218,0.0010660504,0.0012042772,0.0045968844,0.0012917301,0.0012635299,0.002369447,0.0008290968,0.0012997444],"category_scores_gemma":[0.0064328886,0.00057574763,0.0015749856,0.00495415,0.0006131939,0.0017730548,0.0015017071,0.0016260225,0.001462557],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021253472,0.0003194095,0.005079765,0.00028823048,0.00030233763,0.00028734974,0.001077411,0.078069076,0.036962695,0.017816767,0.0038454325,0.85573906],"study_design_scores_gemma":[0.000021873617,0.00009714124,0.002224386,0.00003205551,0.0000776119,0.00053676864,0.00032278392,0.95118856,0.026729293,0.012663893,0.0060251583,0.00008051042],"about_ca_topic_score_codex":0.0055126688,"about_ca_topic_score_gemma":0.0070464453,"teacher_disagreement_score":0.0055126688,"about_ca_system_score_codex":0.0007115585,"about_ca_system_score_gemma":0.0014272892,"threshold_uncertainty_score":0.010961115},"labels":[],"label_agreement":null},{"id":"W1712808800","doi":"10.1007/11527503_23","title":"A Genetic k-Modes Algorithm for Clustering Categorical Data","year":2005,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Computer science; Cluster analysis; Categorical variable; Algorithm; Genetic algorithm; Data mining; Artificial intelligence; Machine learning","score_opus":0.05120452901881751,"score_gpt":0.32311428098578854,"score_spread":0.27190975196697104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1712808800","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004053739,0.00006760198,0.9945116,0.000055959357,0.00003976663,0.000051427007,0.000037999358,0.0006468408,0.00053503254],"genre_scores_gemma":[0.036913544,0.00005415676,0.96069586,0.0000654598,0.000027779874,0.00020598398,0.00015312612,0.00015503046,0.0017289924],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985702,0.0004507196,0.000089157846,0.00034562405,0.0004645516,0.000079672485],"domain_scores_gemma":[0.99834585,0.00075370946,0.00008132516,0.00026762404,0.00047674874,0.00007476914],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019667803,0.00083510263,0.0012602906,0.001633145,0.0013722557,0.0010970696,0.0033046422,0.0017155078,0.0038029437],"category_scores_gemma":[0.005718213,0.00085344206,0.0015900392,0.002142035,0.0009976812,0.0010662418,0.0017472918,0.0019051714,0.001588815],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023834631,0.00017669878,0.0014917187,0.00010954811,0.00017823195,0.00007624039,0.00031351674,0.22833009,0.006028495,0.018846208,0.0055291546,0.7386817],"study_design_scores_gemma":[0.000047832422,0.00006012325,0.00037487302,0.000018175893,0.000031599866,0.00008139037,0.000048441554,0.97890395,0.0017426001,0.016518127,0.0021388354,0.000033983342],"about_ca_topic_score_codex":0.008496757,"about_ca_topic_score_gemma":0.009517913,"teacher_disagreement_score":0.008496757,"about_ca_system_score_codex":0.00096236385,"about_ca_system_score_gemma":0.0017122135,"threshold_uncertainty_score":0.016894639},"labels":[],"label_agreement":null},{"id":"W1724273493","doi":"","title":"Analyse automatique de donnees par support vector machines non supervises","year":2012,"lang":"fr","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Humanities; Support vector machine; Computer science; Artificial intelligence; Philosophy","score_opus":0.03550654522456509,"score_gpt":0.334112047573668,"score_spread":0.29860550234910294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1724273493","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014364515,0.00071298453,0.9822036,0.00018285793,0.000067401445,0.000112412854,0.00020505794,0.0014761313,0.00067513133],"genre_scores_gemma":[0.1513288,0.0013556775,0.83974975,0.00015691279,0.0001234013,0.00032742103,0.0016206151,0.00043804586,0.0048993886],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99395835,0.00092035095,0.00059613236,0.0011768548,0.0030427172,0.00030560698],"domain_scores_gemma":[0.98812276,0.006271887,0.00074125023,0.0009828365,0.0037151834,0.00016604786],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043605804,0.0017328786,0.0019547234,0.0030335146,0.0007766452,0.0038644152,0.0019254897,0.0014757295,0.0032852339],"category_scores_gemma":[0.018055355,0.0010111781,0.0029588805,0.0022560388,0.0009220936,0.0029303979,0.001577782,0.0024825884,0.0023047908],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029232557,0.00008371753,0.0037606882,0.0006700397,0.00024655095,0.00021209937,0.00058709335,0.08606622,0.032602735,0.006090631,0.0024504391,0.8669375],"study_design_scores_gemma":[0.000023177343,0.00018472994,0.0044329376,0.00014140953,0.00011797127,0.00038552468,0.00044119253,0.9377896,0.03503687,0.008996853,0.012387797,0.00006185601],"about_ca_topic_score_codex":0.01184292,"about_ca_topic_score_gemma":0.009596406,"teacher_disagreement_score":0.01184292,"about_ca_system_score_codex":0.001261908,"about_ca_system_score_gemma":0.0023792733,"threshold_uncertainty_score":0.023547947},"labels":[],"label_agreement":null},{"id":"W182707955","doi":"10.1137/1.9781611972764.29","title":"Density-Based Clustering over an Evolving Data Stream with Noise","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1000,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Cluster analysis; Noise (video); Data mining; Artificial intelligence","score_opus":0.030329829383991817,"score_gpt":0.3020425877552507,"score_spread":0.2717127583712589,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W182707955","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034375653,0.0001806806,0.9645568,0.000105918945,0.000017830902,0.000041417385,0.00005160315,0.0003962387,0.00027374295],"genre_scores_gemma":[0.4814362,0.00058428885,0.51587343,0.0000833756,0.000069207475,0.00013252116,0.00047315727,0.00011087562,0.0012368935],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989274,0.00022550063,0.00007843198,0.00023656459,0.00045887852,0.000073262694],"domain_scores_gemma":[0.9957184,0.0022720082,0.00041940217,0.00046478075,0.0010000826,0.00012541967],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002345751,0.00063001644,0.0013026983,0.0021545906,0.00081802165,0.0011067505,0.0019578424,0.0009710949,0.00035526368],"category_scores_gemma":[0.011223438,0.0004607879,0.00064243196,0.0022959854,0.0007807803,0.0019133595,0.0013257312,0.00091312913,0.00020490351],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000219837,0.000047422276,0.004229107,0.000105889376,0.000074908,0.00014968839,0.00031256748,0.8321775,0.0059493994,0.012269755,0.0011599837,0.14330395],"study_design_scores_gemma":[0.000003513441,0.00000842825,0.00027718378,0.0000026127486,0.000003519457,0.000022763832,0.00001333284,0.9961361,0.0010005097,0.002316302,0.00021034367,0.000005294796],"about_ca_topic_score_codex":0.006717699,"about_ca_topic_score_gemma":0.005122504,"teacher_disagreement_score":0.006717699,"about_ca_system_score_codex":0.0013940777,"about_ca_system_score_gemma":0.0008709253,"threshold_uncertainty_score":0.013357222},"labels":[],"label_agreement":null},{"id":"W1875624538","doi":"10.1007/s11634-015-0219-5","title":"Factor probabilistic distance clustering (FPDC): a new clustering method","year":2015,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Università degli Studi di Napoli Federico II","keywords":"Cluster analysis; Correlation clustering; CURE data clustering algorithm; Single-linkage clustering; Probabilistic logic; Canopy clustering algorithm; Computer science; Data stream clustering; Fuzzy clustering; Determining the number of clusters in a data set; Clustering high-dimensional data; k-medians clustering; Transformation (genetics); Mathematics; Pattern recognition (psychology); Data mining; Artificial intelligence","score_opus":0.12235253318764872,"score_gpt":0.4083881885969034,"score_spread":0.2860356554092547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1875624538","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011951176,0.0007192656,0.9967601,0.00012680679,0.0001219096,0.000054091328,0.00011744802,0.00035847185,0.0005468645],"genre_scores_gemma":[0.03915475,0.0012054665,0.9558839,0.00018155322,0.000273096,0.0002045026,0.00063115614,0.0002979931,0.0021675325],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9926645,0.0017187134,0.00042034817,0.001765367,0.0031921363,0.00023891119],"domain_scores_gemma":[0.9955818,0.0010924345,0.00035465005,0.0010324306,0.0017795999,0.0001590322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037781529,0.0016513546,0.002357854,0.0045523457,0.0022730345,0.0026633928,0.0037815142,0.0026331637,0.0022695842],"category_scores_gemma":[0.010060121,0.0008283834,0.0022515785,0.007889587,0.0019187598,0.0033432723,0.0034210489,0.003131452,0.0020724712],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027776673,0.00011612515,0.002149991,0.00069898576,0.0003766307,0.00009979348,0.00040757566,0.08116129,0.008370257,0.06646979,0.019165006,0.8207068],"study_design_scores_gemma":[0.0000643426,0.00017484362,0.0019198905,0.00012762296,0.00020361092,0.0009411274,0.0001695635,0.82802284,0.009874055,0.09945981,0.058808766,0.00023353526],"about_ca_topic_score_codex":0.00695526,"about_ca_topic_score_gemma":0.0063181594,"teacher_disagreement_score":0.00695526,"about_ca_system_score_codex":0.0015881434,"about_ca_system_score_gemma":0.0033170464,"threshold_uncertainty_score":0.019981086},"labels":[],"label_agreement":null},{"id":"W1891514807","doi":"10.1007/3-540-47887-6_4","title":"On Data Clustering Analysis: Scalability, Constraints, and Validation","year":2002,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":88,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Scalability; Data mining; Correlation clustering; Similarity (geometry); Consensus clustering; CURE data clustering algorithm; Artificial intelligence; Machine learning; Database","score_opus":0.050158553012031065,"score_gpt":0.31330475141825476,"score_spread":0.26314619840622366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1891514807","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015461093,0.0020902723,0.97208655,0.002543507,0.00017005546,0.000554643,0.0003631346,0.0028498396,0.0038808973],"genre_scores_gemma":[0.16109584,0.0013421993,0.8317899,0.0008916941,0.00041177895,0.0007241092,0.0010388279,0.0010372143,0.0016683635],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9064413,0.05417688,0.006022006,0.006326301,0.02502504,0.0020085599],"domain_scores_gemma":[0.53571093,0.35108078,0.008944332,0.06862209,0.033270262,0.00237163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09286975,0.0031803937,0.0049676797,0.0050252946,0.00366681,0.007899754,0.008711525,0.005999687,0.0036038307],"category_scores_gemma":[0.34986407,0.0026246533,0.003295851,0.009786966,0.00758834,0.021646535,0.01261559,0.007990186,0.0015152256],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017671324,0.00060709164,0.01816363,0.0013019044,0.0007371357,0.00031021002,0.0009414972,0.25984362,0.005876109,0.10251893,0.019828446,0.58810437],"study_design_scores_gemma":[0.00010696717,0.00009245071,0.0012671915,0.0001977312,0.000091991446,0.00022607596,0.00015772686,0.88763446,0.0039791143,0.104467146,0.0017298401,0.000049378286],"about_ca_topic_score_codex":0.010055188,"about_ca_topic_score_gemma":0.0091021685,"teacher_disagreement_score":0.09286975,"about_ca_system_score_codex":0.004350682,"about_ca_system_score_gemma":0.008509657,"threshold_uncertainty_score":0.49114805},"labels":[],"label_agreement":null},{"id":"W191887720","doi":"10.1007/s00357-001-0018-x","title":"Optimal Variable Weighting for Ultrametric and Additive Trees and K-means Partitioning: Methods and Software","year":2001,"lang":"en","type":"article","venue":"Journal of Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":96,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Ultrametric space; Weighting; Outlier; Cluster analysis; Recursive partitioning; Variable (mathematics); Partition (number theory); Computer science; Monte Carlo method; Algorithm; Data mining; Tree (set theory); Mathematics; Statistics; Artificial intelligence; Machine learning; Combinatorics","score_opus":0.05001129746715327,"score_gpt":0.3724863672266711,"score_spread":0.32247506975951784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W191887720","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002044137,0.00014621305,0.9974004,0.00003842065,0.000022507036,0.000016495593,0.000013240444,0.00019734893,0.00012128633],"genre_scores_gemma":[0.02586521,0.00015172624,0.9728602,0.000024127512,0.000043540123,0.00011086659,0.00008319924,0.00019661066,0.0006644399],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99472946,0.002558931,0.0003978141,0.00082459935,0.0012626793,0.00022650698],"domain_scores_gemma":[0.9892697,0.006981115,0.0004912786,0.0012973113,0.001704729,0.00025592075],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008898418,0.0012730409,0.002567802,0.0023673917,0.001412633,0.0028389276,0.003776597,0.002133588,0.0031884857],"category_scores_gemma":[0.023992287,0.001667014,0.0020839733,0.0035730444,0.0015233771,0.0047191796,0.003360583,0.003349282,0.0011968167],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002089806,0.00013504208,0.0008346335,0.00027137392,0.00012941545,0.00002811904,0.0003352667,0.2004033,0.0025582383,0.0916884,0.0032549086,0.7001524],"study_design_scores_gemma":[0.000027304048,0.0000349376,0.00022978494,0.00002303432,0.000038603823,0.00003680491,0.000037472924,0.92560834,0.0015250107,0.07050094,0.0019114024,0.000026427444],"about_ca_topic_score_codex":0.0054918705,"about_ca_topic_score_gemma":0.0077487766,"teacher_disagreement_score":0.008898418,"about_ca_system_score_codex":0.0017232603,"about_ca_system_score_gemma":0.0025962356,"threshold_uncertainty_score":0.047059894},"labels":[],"label_agreement":null},{"id":"W1966755259","doi":"10.1109/nafips.2010.5548181","title":"A new approach to improve the accuracy of online clustering algorithm based on scatter/gather model","year":2010,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Process (computing); Cluster (spacecraft); Canopy clustering algorithm; Algorithm; CURE data clustering algorithm; Correlation clustering; Algorithm design; Data stream clustering; Artificial intelligence; Machine learning","score_opus":0.02906525681003538,"score_gpt":0.31646283583198503,"score_spread":0.2873975790219497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1966755259","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005850518,0.0001453611,0.9919842,0.0001268681,0.00007189311,0.00003874099,0.000023295268,0.0012065682,0.00055260136],"genre_scores_gemma":[0.14079359,0.00026262566,0.8543912,0.00016863513,0.00018582586,0.00016449872,0.0002240242,0.0004580458,0.0033514209],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99677247,0.00064293796,0.00019567402,0.00059915,0.0015711555,0.00021863016],"domain_scores_gemma":[0.9926509,0.0020304278,0.00039313769,0.0016046672,0.0030812947,0.00023959127],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030308058,0.001370899,0.0023574918,0.002981686,0.001747721,0.0021076398,0.0032688645,0.0020530138,0.002462932],"category_scores_gemma":[0.010058962,0.00074086385,0.0014022426,0.0031263595,0.001126244,0.005541166,0.0020096735,0.0023666148,0.0016070267],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007097164,0.00062455575,0.004773955,0.00029604035,0.00026556558,0.00018134192,0.0007592812,0.19778992,0.03002225,0.029796742,0.011454324,0.72332627],"study_design_scores_gemma":[0.000041150066,0.00012947374,0.00047951116,0.000014693751,0.0000443913,0.00020181965,0.00006906306,0.9766634,0.009999996,0.0074605374,0.0048532058,0.000042770655],"about_ca_topic_score_codex":0.004766865,"about_ca_topic_score_gemma":0.0044750897,"teacher_disagreement_score":0.004766865,"about_ca_system_score_codex":0.0010877101,"about_ca_system_score_gemma":0.0016083983,"threshold_uncertainty_score":0.016028643},"labels":[],"label_agreement":null},{"id":"W1972132022","doi":"10.1145/1046456.1046468","title":"Subspace clustering for high dimensional categorical data","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Computer science; Clustering high-dimensional data; Categorical variable; Correlation clustering; CURE data clustering algorithm; Data mining; Subspace topology; Canopy clustering algorithm; Data stream clustering; Focus (optics); Algorithm; Pattern recognition (psychology); Artificial intelligence; Machine learning","score_opus":0.11230554248111278,"score_gpt":0.34489126193553477,"score_spread":0.23258571945442197,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1972132022","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029771768,0.00018914425,0.99621105,0.00008878906,0.000013166352,0.000021611699,0.000062117135,0.00019685042,0.0002400615],"genre_scores_gemma":[0.08437975,0.00045641002,0.9132391,0.000067568246,0.000060282993,0.00015289032,0.00065029063,0.00010090363,0.0008928911],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958442,0.0018400035,0.00022451645,0.0005922827,0.0013391597,0.00015977501],"domain_scores_gemma":[0.9945497,0.0025171295,0.00042908706,0.0009376222,0.0013997749,0.00016661272],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036670242,0.0007325431,0.0013952,0.0031338092,0.0016159528,0.0015248206,0.0013621288,0.001086037,0.0013659481],"category_scores_gemma":[0.011253672,0.00035343794,0.0013862362,0.0048139407,0.0013641313,0.0019349727,0.0018699398,0.0015861645,0.000713491],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015771468,0.00009320082,0.0037724364,0.0005072683,0.00017550368,0.00023284531,0.0008529113,0.3639867,0.009032694,0.17664865,0.0070194677,0.43752062],"study_design_scores_gemma":[0.000009494041,0.000051233143,0.0009633853,0.00002273448,0.00001552045,0.00015240135,0.00015880106,0.88820785,0.0031883654,0.101581015,0.0056013856,0.00004775618],"about_ca_topic_score_codex":0.0037131135,"about_ca_topic_score_gemma":0.0043203845,"teacher_disagreement_score":0.0037131135,"about_ca_system_score_codex":0.0011755482,"about_ca_system_score_gemma":0.0017140458,"threshold_uncertainty_score":0.019393325},"labels":[],"label_agreement":null},{"id":"W1973892392","doi":"10.1016/s0167-8655(02)00130-7","title":"Collaborative fuzzy clustering","year":2002,"lang":"en","type":"article","venue":"Pattern Recognition Letters","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":282,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Partition (number theory); Computer science; Data mining; Fuzzy clustering; Fuzzy logic; Theoretical computer science; Basis (linear algebra); Artificial intelligence; Mathematics","score_opus":0.034607468860997824,"score_gpt":0.26777838524257236,"score_spread":0.23317091638157453,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1973892392","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0061030714,0.00044101643,0.98456097,0.00017241077,0.00013838371,0.00009449061,0.00011681405,0.0004595972,0.0079133],"genre_scores_gemma":[0.32006827,0.000595285,0.6516533,0.00023227464,0.00024357224,0.0002819571,0.0009802406,0.0002291414,0.025716042],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965055,0.00065651984,0.0001902982,0.0010911927,0.0013113351,0.00024509832],"domain_scores_gemma":[0.9968076,0.00060816715,0.00015450726,0.0011302758,0.0011488955,0.00015069281],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026012014,0.0012086147,0.0025549985,0.0030774479,0.0028538161,0.0030928073,0.0033144194,0.0030058573,0.008628493],"category_scores_gemma":[0.0056710723,0.00085267023,0.0024920425,0.0028215728,0.0013229387,0.00218887,0.0032197575,0.0013852952,0.0037313178],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057552545,0.000196886,0.0015402199,0.00030096032,0.00047058478,0.00028204446,0.000351867,0.2879498,0.015317221,0.09350444,0.018081153,0.5814293],"study_design_scores_gemma":[0.000033126526,0.00007774737,0.0005726445,0.000024357902,0.00008061058,0.00023703526,0.00010243214,0.93713474,0.008283947,0.04225315,0.01114771,0.000052501597],"about_ca_topic_score_codex":0.005578012,"about_ca_topic_score_gemma":0.0071241837,"teacher_disagreement_score":0.008628493,"about_ca_system_score_codex":0.0015132005,"about_ca_system_score_gemma":0.0018032076,"threshold_uncertainty_score":0.028865159},"labels":[],"label_agreement":null},{"id":"W1977437243","doi":"10.1007/s10489-008-0129-8","title":"Parallel clustering of high dimensional data by integrating multi-objective genetic algorithm with divide and conquer","year":2008,"lang":"en","type":"article","venue":"Applied Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Divide and conquer algorithms; Centroid; CURE data clustering algorithm; Single-linkage clustering; Correlation clustering; Tree (set theory); Scalability; Canopy clustering algorithm; Fuzzy clustering; Process (computing); Node (physics); Algorithm; Artificial intelligence; Mathematics; Database","score_opus":0.039269687142191013,"score_gpt":0.2876830857060188,"score_spread":0.2484133985638278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1977437243","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023709053,0.00014738498,0.9737385,0.0001104129,0.00004791991,0.000076410935,0.00001983852,0.0009076494,0.0012428372],"genre_scores_gemma":[0.1587982,0.00008717865,0.8393562,0.00005398646,0.000032019,0.000188719,0.000099620964,0.00013696474,0.001247112],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994149,0.00014168517,0.00002887398,0.000116903946,0.00023374356,0.00006387001],"domain_scores_gemma":[0.9993332,0.00026747774,0.00006027022,0.00009077527,0.00021294659,0.000035236735],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012222048,0.0009197684,0.0015497118,0.0014519459,0.0013088513,0.0009823855,0.0014382463,0.0010681715,0.001347307],"category_scores_gemma":[0.0021110855,0.0006190113,0.0010074337,0.0018995075,0.00074303657,0.00120486,0.0010016961,0.001065962,0.00034582196],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020149123,0.000226213,0.0011550892,0.00010018459,0.00016864482,0.00008142751,0.00022028347,0.73603994,0.009119728,0.006305497,0.0014966088,0.24488488],"study_design_scores_gemma":[0.00002054639,0.000021105368,0.00012430687,0.0000023517619,0.000012825112,0.000014736689,0.000015638801,0.9958264,0.0014129931,0.0022216602,0.00032211642,0.0000052658],"about_ca_topic_score_codex":0.013366381,"about_ca_topic_score_gemma":0.012477972,"teacher_disagreement_score":0.013366381,"about_ca_system_score_codex":0.001114614,"about_ca_system_score_gemma":0.0018303032,"threshold_uncertainty_score":0.026577115},"labels":[],"label_agreement":null},{"id":"W1977655121","doi":"10.1145/1281192.1281229","title":"Constraint-driven clustering","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Cluster analysis; Computer science; Constrained clustering; Data mining; Constraint (computer-aided design); Scalability; Node (physics); A priori and a posteriori; Tree (set theory); Correlation clustering; CURE data clustering algorithm; Machine learning; Mathematics; Database","score_opus":0.027766444093233224,"score_gpt":0.32127261763251275,"score_spread":0.29350617353927955,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1977655121","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023399042,0.00031701894,0.99292326,0.00023275608,0.000051518116,0.00022059865,0.0005171405,0.00049041206,0.0029073376],"genre_scores_gemma":[0.11432271,0.0009876251,0.8717874,0.0005129195,0.00013888393,0.000854106,0.0044856532,0.00055035733,0.0063604014],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959966,0.0010932628,0.00025660434,0.0010191477,0.0014072979,0.00022712191],"domain_scores_gemma":[0.9946839,0.0019542554,0.00035354102,0.0010742005,0.0017535207,0.00018056473],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023612552,0.0018680717,0.0020887305,0.002082394,0.001821717,0.0026412208,0.004688531,0.00224124,0.0057739113],"category_scores_gemma":[0.009121219,0.00095773133,0.002093435,0.0041194414,0.0012940987,0.0029786653,0.0033199878,0.0020392586,0.0029896237],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001628257,0.000121808305,0.0013712613,0.0009201903,0.00024420605,0.00022922085,0.00032887023,0.6835683,0.005053641,0.09426099,0.034362793,0.17937589],"study_design_scores_gemma":[0.000032480326,0.000034178407,0.00022023688,0.000040517632,0.0000215622,0.00016965499,0.000080190875,0.9206127,0.002284711,0.058483567,0.017977845,0.000042351545],"about_ca_topic_score_codex":0.0071732,"about_ca_topic_score_gemma":0.009649424,"teacher_disagreement_score":0.0071732,"about_ca_system_score_codex":0.002295663,"about_ca_system_score_gemma":0.0033198032,"threshold_uncertainty_score":0.0193156},"labels":[],"label_agreement":null},{"id":"W1983989808","doi":"10.1007/s00357-005-0018-3","title":"Analysis of Global k-Means, an Incremental Heuristic for Minimum Sum-of-Squares Clustering","year":2005,"lang":"en","type":"article","venue":"Journal of Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal; Group for Research in Decision Analysis; HEC Montréal","funders":"","keywords":"Cluster analysis; Heuristic; Explained sum of squares; Mathematics; Artificial intelligence; Pattern recognition (psychology); Combinatorics; Computer science; Statistics","score_opus":0.05400124671072944,"score_gpt":0.37078076021751716,"score_spread":0.31677951350678774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1983989808","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06870507,0.0012902197,0.92482984,0.0003955549,0.00010132238,0.00010389991,0.000119295546,0.0006948782,0.0037599006],"genre_scores_gemma":[0.6132319,0.0006084343,0.38115734,0.00018831331,0.00014519105,0.00017506472,0.0007003073,0.0007445176,0.0030489669],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997959,0.00075025356,0.00005940099,0.00028367073,0.000733383,0.00021420821],"domain_scores_gemma":[0.9853253,0.011016349,0.00045545326,0.00091337407,0.0020725287,0.00021697803],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038584084,0.0013715737,0.0017215016,0.0015277644,0.0011153123,0.0018825307,0.002697124,0.0013191635,0.0037733607],"category_scores_gemma":[0.020613922,0.0007924715,0.0013107025,0.0015355576,0.0012533715,0.0030412697,0.0015742703,0.0016535596,0.00045553994],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00078470045,0.00022561902,0.0029473118,0.00035844892,0.00020502921,0.000093567825,0.0002269529,0.79640937,0.003618287,0.043784074,0.005363621,0.145983],"study_design_scores_gemma":[0.000013980657,0.00006235634,0.0004206551,0.000008332356,0.000028934275,0.000026140686,0.00003231671,0.9891867,0.0006814051,0.00910072,0.0004297025,0.000008768442],"about_ca_topic_score_codex":0.006103414,"about_ca_topic_score_gemma":0.00700483,"teacher_disagreement_score":0.006103414,"about_ca_system_score_codex":0.0017896089,"about_ca_system_score_gemma":0.0025633653,"threshold_uncertainty_score":0.020405412},"labels":[],"label_agreement":null},{"id":"W1986303633","doi":"10.1049/el:20073093","title":"Performance of kernel-based fuzzy clustering","year":2007,"lang":"en","type":"article","venue":"Electronics Letters","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Kernel (algebra); Artificial intelligence; Cluster analysis; Fuzzy logic; Data mining; Fuzzy clustering; Computer science; Machine learning; Pattern recognition (psychology); Mathematics","score_opus":0.010267746433416589,"score_gpt":0.26131967321908706,"score_spread":0.25105192678567045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1986303633","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5436,0.005084153,0.43611658,0.00042834887,0.00021018872,0.000108120534,0.00036816503,0.0050679864,0.009016397],"genre_scores_gemma":[0.94509846,0.00044703373,0.052876774,0.000020761529,0.000013399755,0.000019010959,0.00033317658,0.00009928168,0.0010920678],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99782276,0.0006653055,0.00014766029,0.00028705722,0.00089321536,0.00018406226],"domain_scores_gemma":[0.994089,0.0026943728,0.00032919872,0.0007388005,0.0020272778,0.00012134104],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037470756,0.0008315761,0.0012480108,0.0016889595,0.00084643665,0.0017450624,0.0011582385,0.0015196081,0.0010825384],"category_scores_gemma":[0.012987283,0.00021262931,0.00051487656,0.0017806387,0.000608918,0.0013537755,0.00092966703,0.00055006007,0.00059256423],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00234191,0.00017021965,0.0037292754,0.00041466157,0.0002605021,0.000093174254,0.00023377249,0.67512304,0.009298543,0.00522139,0.0020537106,0.30105984],"study_design_scores_gemma":[0.000017434995,0.00014937088,0.0018086355,0.000012355295,0.000032738135,0.00007696562,0.00006071696,0.98767895,0.008291383,0.0011350315,0.00070578814,0.00003072042],"about_ca_topic_score_codex":0.011980476,"about_ca_topic_score_gemma":0.003871437,"teacher_disagreement_score":0.011980476,"about_ca_system_score_codex":0.0016693345,"about_ca_system_score_gemma":0.00095531635,"threshold_uncertainty_score":0.023821473},"labels":[],"label_agreement":null},{"id":"W1987111416","doi":"10.1002/widm.30","title":"Density‐based clustering","year":2011,"lang":"en","type":"article","venue":"Wiley Interdisciplinary Reviews Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":810,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Outlier; Data set; Computer science; Data mining; Cluster (spacecraft); Set (abstract data type); Single-linkage clustering; DBSCAN; Pattern recognition (psychology); Correlation clustering; CURE data clustering algorithm; Artificial intelligence","score_opus":0.15651942522856147,"score_gpt":0.37289851207918545,"score_spread":0.216379086850624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1987111416","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060664024,0.002601064,0.97213453,0.000663111,0.00022460337,0.00034320817,0.0015029783,0.0016075409,0.014856623],"genre_scores_gemma":[0.30667707,0.0058786785,0.65727484,0.00065458927,0.00049744034,0.00077690487,0.009142983,0.0010073734,0.018090159],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99501246,0.0011331846,0.0002726587,0.0011418659,0.0021684736,0.00027127058],"domain_scores_gemma":[0.9942855,0.0018765172,0.00047513025,0.0012324377,0.0019745734,0.00015582213],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034005304,0.0014905431,0.0020547162,0.008395691,0.0020712565,0.004955584,0.0031010837,0.0023518149,0.008533414],"category_scores_gemma":[0.017203083,0.0009357687,0.0018201969,0.009168972,0.0017023634,0.0035161148,0.0035317,0.0018224838,0.0066979406],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023933119,0.00012533936,0.007886333,0.0011763432,0.0005308698,0.00031956765,0.00094165123,0.26438347,0.0048684445,0.28152376,0.061402865,0.37660202],"study_design_scores_gemma":[0.000039690352,0.00006405476,0.0037962047,0.00029913202,0.0001236625,0.00066472864,0.00038614453,0.6702391,0.0060344557,0.22271296,0.095447265,0.00019260378],"about_ca_topic_score_codex":0.008082308,"about_ca_topic_score_gemma":0.0049816337,"teacher_disagreement_score":0.008533414,"about_ca_system_score_codex":0.0027955754,"about_ca_system_score_gemma":0.0026942843,"threshold_uncertainty_score":0.028547108},"labels":[],"label_agreement":null},{"id":"W1990092496","doi":"10.1016/j.ins.2014.03.106","title":"A new credibilistic clustering algorithm","year":2014,"lang":"en","type":"article","venue":"Information Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Computer science; FLAME clustering; Data mining; Correlation clustering; CURE data clustering algorithm; Measure (data warehouse); Pattern recognition (psychology); Fuzzy logic; Credibility; Artificial intelligence","score_opus":0.025568710353121037,"score_gpt":0.3085933236606262,"score_spread":0.28302461330750517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1990092496","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048823603,0.0003129625,0.98915535,0.0003682848,0.00026782893,0.00012657956,0.00021635361,0.00096665847,0.0037035493],"genre_scores_gemma":[0.055536482,0.00024506927,0.93425894,0.0004092499,0.00025405452,0.00021444703,0.00081855594,0.00024384393,0.008019363],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99751663,0.00053651765,0.00016974767,0.0007137571,0.0009229061,0.00014055033],"domain_scores_gemma":[0.9971698,0.0006114518,0.00014728415,0.0008384091,0.0010273252,0.00020572843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002364921,0.0008225071,0.0016720506,0.0031971494,0.0027992227,0.0032349022,0.0043312614,0.0029162783,0.008636435],"category_scores_gemma":[0.0077327164,0.00061362213,0.0009972502,0.003963693,0.0013721995,0.003723201,0.0032280106,0.002585558,0.004586506],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005256451,0.0002743099,0.0013170178,0.00024987423,0.00013173274,0.00009165914,0.00018236577,0.06678295,0.008830116,0.14045267,0.02585397,0.7553077],"study_design_scores_gemma":[0.00015574344,0.00011174363,0.00052181154,0.000051314673,0.000050229235,0.00027842543,0.00006398199,0.86583334,0.005109157,0.10479344,0.02296761,0.0000632525],"about_ca_topic_score_codex":0.003334033,"about_ca_topic_score_gemma":0.0048242174,"teacher_disagreement_score":0.008636435,"about_ca_system_score_codex":0.0014728844,"about_ca_system_score_gemma":0.002952846,"threshold_uncertainty_score":0.028891802},"labels":[],"label_agreement":null},{"id":"W1993343070","doi":"10.1109/icdm.2013.28","title":"Classification-Based Clustering Evaluation","year":2013,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Generality; Computer science; Classifier (UML); Conceptual clustering; Data mining; Machine learning; Artificial intelligence; Task (project management); Process (computing); Pattern recognition (psychology); Fuzzy clustering; CURE data clustering algorithm","score_opus":0.06908488101264255,"score_gpt":0.34634609223937074,"score_spread":0.2772612112267282,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993343070","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24818747,0.005193539,0.66549426,0.0021524546,0.0013713709,0.0030492085,0.016204415,0.011401012,0.046946257],"genre_scores_gemma":[0.7014675,0.00062968215,0.26545364,0.0005115195,0.0002187774,0.00077060674,0.022916814,0.0010367841,0.0069946772],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.96471304,0.009557173,0.0035944926,0.004295221,0.016304312,0.0015358119],"domain_scores_gemma":[0.94537145,0.014292157,0.0035179916,0.008026457,0.027301349,0.0014906466],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.025111057,0.0022542751,0.0025098692,0.01308871,0.0017326609,0.005338585,0.0036789302,0.0032196075,0.0064773853],"category_scores_gemma":[0.06686461,0.00035909246,0.002080937,0.006688475,0.001358555,0.0043196464,0.0034100604,0.0017426686,0.0027343142],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00221688,0.0008746242,0.07088349,0.001936113,0.0016005138,0.0002681738,0.00070216524,0.13538578,0.015029745,0.025578,0.07069879,0.67482585],"study_design_scores_gemma":[0.00013011896,0.0009138939,0.033273052,0.00025281293,0.00029768367,0.00045809124,0.0008532978,0.89388794,0.029759094,0.020660765,0.019343175,0.00017016396],"about_ca_topic_score_codex":0.0062706424,"about_ca_topic_score_gemma":0.00667448,"teacher_disagreement_score":0.025111057,"about_ca_system_score_codex":0.004717597,"about_ca_system_score_gemma":0.0022328077,"threshold_uncertainty_score":0.13280159},"labels":[],"label_agreement":null},{"id":"W1997668352","doi":"10.1109/tkde.2012.190","title":"Efficient Cluster Labeling for Support Vector Clustering","year":2013,"lang":"en","type":"article","venue":"IEEE Transactions on Knowledge and Data Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Computer science; Cluster analysis; Disjoint sets; Interconnection; Cluster (spacecraft); Process (computing); Algorithm; Point (geometry); Function (biology); Data mining; Pattern recognition (psychology); Artificial intelligence; Mathematics","score_opus":0.03179128225384418,"score_gpt":0.3012782757759483,"score_spread":0.2694869935221041,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1997668352","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029353092,0.00006056556,0.99608743,0.00006429996,0.000023023176,0.000041884217,0.00003764731,0.00047278914,0.00027709623],"genre_scores_gemma":[0.07008161,0.000102904356,0.92780983,0.00007054501,0.000060881368,0.00022508447,0.00057870545,0.00015884665,0.0009116412],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99715024,0.0008122285,0.00016651078,0.0004910038,0.0011793119,0.00020064761],"domain_scores_gemma":[0.9964534,0.0012893475,0.00029985636,0.0005711472,0.0012636313,0.00012258666],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00191557,0.0012650839,0.0015826938,0.0022702182,0.0012219178,0.0019497873,0.0023749762,0.001817733,0.0025055],"category_scores_gemma":[0.008627493,0.0006456508,0.0008624028,0.0027280555,0.0010480289,0.0021725965,0.0017237008,0.0021725977,0.0020775513],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003282766,0.00015491535,0.0012965833,0.00025129065,0.000090963804,0.00011685231,0.00027195338,0.30648518,0.013459126,0.043268397,0.009463143,0.62481326],"study_design_scores_gemma":[0.000015342628,0.0000354663,0.00011771765,0.000009074528,0.000005473014,0.00003265209,0.000027600117,0.98147947,0.0032551822,0.013159409,0.0018499985,0.000012704066],"about_ca_topic_score_codex":0.00331953,"about_ca_topic_score_gemma":0.0026776206,"teacher_disagreement_score":0.00331953,"about_ca_system_score_codex":0.0011974336,"about_ca_system_score_gemma":0.0020901344,"threshold_uncertainty_score":0.010130644},"labels":[],"label_agreement":null},{"id":"W1998344252","doi":"10.1145/2597008.2602159","title":"The MoJo family: a story about clustering evaluation (invited talk)","year":2014,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Computer science; Software; Cluster (spacecraft); Domain (mathematical analysis); Data mining; Data science; Machine learning; Programming language; Mathematics","score_opus":0.033363713175748204,"score_gpt":0.32044618986596457,"score_spread":0.28708247669021636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1998344252","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0077746557,0.031793773,0.20181476,0.67834336,0.02365828,0.00037450544,0.00054268155,0.0045973123,0.051100668],"genre_scores_gemma":[0.18350853,0.023517506,0.37915695,0.26276115,0.03278671,0.002075995,0.0014796007,0.01840427,0.09630939],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9536122,0.03035774,0.0008437644,0.0032618926,0.010637966,0.0012863993],"domain_scores_gemma":[0.9206415,0.04846873,0.0017906077,0.004644855,0.019216469,0.005237903],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.07601739,0.0015975587,0.0017706453,0.004598388,0.012053268,0.016056359,0.0045351284,0.010211585,0.0057337796],"category_scores_gemma":[0.14369056,0.0013162011,0.00220645,0.004898119,0.014145098,0.02625715,0.0077995686,0.017458681,0.0051335753],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023290837,0.00012999006,0.0013469082,0.00035759903,0.00011594017,0.0002538539,0.008611954,0.0022686394,0.00088229764,0.08386295,0.72454643,0.17739058],"study_design_scores_gemma":[0.000068588866,0.00015986066,0.001759833,0.0007104542,0.00004392535,0.00046399483,0.005801653,0.008223066,0.0019006251,0.124480344,0.85607094,0.00031665742],"about_ca_topic_score_codex":0.012356009,"about_ca_topic_score_gemma":0.011021754,"teacher_disagreement_score":0.9239826,"about_ca_system_score_codex":0.008633165,"about_ca_system_score_gemma":0.0036107034,"threshold_uncertainty_score":0.4020232},"labels":[],"label_agreement":null},{"id":"W1999646319","doi":"10.1016/j.patrec.2006.01.015","title":"An objective approach to cluster validation","year":2006,"lang":"en","type":"article","venue":"Pattern Recognition Letters","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":122,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Covariance; Cluster (spacecraft); Data mining; Computer science; Measure (data warehouse); Separable space; Fuzzy clustering; Index (typography); Mathematics; Rand index; Artificial intelligence; Pattern recognition (psychology); Statistics","score_opus":0.02420805172161281,"score_gpt":0.27613756821425073,"score_spread":0.2519295164926379,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1999646319","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020477974,0.00005056924,0.99709463,0.0000903353,0.000018172019,0.000057258752,0.000042157997,0.00012920909,0.0004698944],"genre_scores_gemma":[0.11577475,0.00014446783,0.87897706,0.00026799573,0.00009279756,0.0006121332,0.0006621641,0.00038949348,0.0030791117],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9848262,0.0083360635,0.0008555701,0.001533478,0.0038110488,0.00063756714],"domain_scores_gemma":[0.9777974,0.0108830845,0.001160113,0.0025145686,0.0071885684,0.00045627786],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020324884,0.0018522507,0.002238298,0.0042404938,0.0022653802,0.0050879335,0.0040592235,0.0029134622,0.002601517],"category_scores_gemma":[0.03390421,0.0011910847,0.0019994208,0.0023496745,0.0024417753,0.0030513578,0.00461418,0.0031200335,0.00075483567],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045855728,0.00040950047,0.0073782485,0.0006439066,0.00089622807,0.00013893786,0.0007204088,0.37651253,0.007123979,0.18335126,0.008523396,0.41384304],"study_design_scores_gemma":[0.00003725158,0.00013016757,0.0010542659,0.00007089828,0.00006329097,0.00007011942,0.00014312779,0.9453444,0.0040842295,0.04628874,0.0026736518,0.000039848994],"about_ca_topic_score_codex":0.0032235505,"about_ca_topic_score_gemma":0.0043262593,"teacher_disagreement_score":0.020324884,"about_ca_system_score_codex":0.0018619807,"about_ca_system_score_gemma":0.0039009931,"threshold_uncertainty_score":0.107489526},"labels":[],"label_agreement":null},{"id":"W1999677481","doi":"10.1007/s12065-013-0103-1","title":"On evolutionary subspace clustering with symbiosis","year":2014,"lang":"en","type":"article","venue":"Evolutionary Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Benchmarking; Scalability; Subspace topology; Population; Artificial intelligence; Machine learning","score_opus":0.015414830467287739,"score_gpt":0.26708960754385996,"score_spread":0.2516747770765722,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1999677481","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017806105,0.00057359296,0.9764598,0.0002888338,0.00009999933,0.000034422534,0.000013600509,0.00009177137,0.0046318243],"genre_scores_gemma":[0.46574947,0.00094842474,0.5245901,0.00032498097,0.00020051247,0.00015388324,0.00012207696,0.000113875074,0.0077967173],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99851424,0.00072784454,0.000051311526,0.00020836158,0.0004046941,0.000093533374],"domain_scores_gemma":[0.99841094,0.0007150971,0.00010233631,0.00030242436,0.00035860064,0.00011057472],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020369454,0.0006496016,0.0009961411,0.0009665853,0.0011373991,0.0010260962,0.0016768258,0.001154628,0.0019850328],"category_scores_gemma":[0.0062686517,0.00036312055,0.000785749,0.0014644181,0.0014018067,0.00182264,0.004509958,0.0010719565,0.0004020291],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013628113,0.00012996746,0.002149342,0.00014807546,0.0001885282,0.00017570415,0.00040622367,0.46887547,0.0060967766,0.22733468,0.003457478,0.29090157],"study_design_scores_gemma":[0.00001282681,0.000063692394,0.0003114535,0.000014888715,0.000015771244,0.000099291,0.000053294792,0.9345969,0.00075395877,0.06084106,0.0032230706,0.0000137826455],"about_ca_topic_score_codex":0.0014348679,"about_ca_topic_score_gemma":0.0013642041,"teacher_disagreement_score":0.0020369454,"about_ca_system_score_codex":0.00056957826,"about_ca_system_score_gemma":0.00060029747,"threshold_uncertainty_score":0.010772526},"labels":[],"label_agreement":null},{"id":"W2005934890","doi":"10.1007/s10618-013-0336-8","title":"Clustering categorical data in projected spaces","year":2013,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Cluster analysis; Categorical variable; Computer science; Data mining; Outlier; CURE data clustering algorithm; Curse of dimensionality; Clustering high-dimensional data; Correlation clustering; Canopy clustering algorithm; Data stream clustering; Anomaly detection; Probabilistic logic; Set (abstract data type); Artificial intelligence; Pattern recognition (psychology); Machine learning","score_opus":0.09257415986957443,"score_gpt":0.3536923389834265,"score_spread":0.26111817911385204,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2005934890","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026683636,0.0002873314,0.9716309,0.0002125753,0.000037361515,0.000056232595,0.00029771763,0.00020170689,0.00059259066],"genre_scores_gemma":[0.4011313,0.00083621166,0.5930611,0.00010979081,0.00016111461,0.00030804108,0.0023523725,0.00010892633,0.001931086],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939646,0.002668276,0.00037810704,0.00089254667,0.0018633271,0.00023310866],"domain_scores_gemma":[0.9905229,0.00468479,0.00065285,0.0015052943,0.00226564,0.00036852362],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035654467,0.0006601155,0.0014429196,0.0032962754,0.0009856101,0.0041706287,0.0016828232,0.00086424215,0.00182067],"category_scores_gemma":[0.019665848,0.0004967592,0.001414593,0.0059513766,0.0018691374,0.0038612979,0.0032101902,0.0020124977,0.0006795465],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067602674,0.00034449837,0.015950983,0.00076244806,0.00041160182,0.00025594697,0.0018769094,0.21932463,0.006321073,0.28259137,0.005340605,0.4661439],"study_design_scores_gemma":[0.000023607923,0.000114582086,0.0023722316,0.00006407911,0.00003795208,0.00016626359,0.00059378677,0.7229326,0.001872669,0.26872396,0.0030531061,0.00004528843],"about_ca_topic_score_codex":0.0027820426,"about_ca_topic_score_gemma":0.0021047967,"teacher_disagreement_score":0.0041706287,"about_ca_system_score_codex":0.0010668829,"about_ca_system_score_gemma":0.0012404603,"threshold_uncertainty_score":0.018856108},"labels":[],"label_agreement":null},{"id":"W2010564375","doi":"10.1145/1389095.1389118","title":"Particle swarm clustering ensemble","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; CURE data clustering algorithm; Correlation clustering; Computer science; Consensus clustering; Data mining; Canopy clustering algorithm; Particle swarm optimization; Artificial intelligence; Robustness (evolution); Pattern recognition (psychology); Machine learning","score_opus":0.051617333080655387,"score_gpt":0.2997592064093752,"score_spread":0.24814187332871981,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2010564375","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015358212,0.0007655798,0.97788846,0.0002530035,0.00022517022,0.00009384854,0.00017220445,0.000609632,0.004633886],"genre_scores_gemma":[0.43144378,0.001802541,0.5514519,0.00025415284,0.00034838656,0.00045003957,0.0015067733,0.00018627684,0.01255605],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99905044,0.00021176273,0.000044759272,0.00026759898,0.0003547319,0.00007062855],"domain_scores_gemma":[0.99875104,0.00036146273,0.000114800816,0.00026545543,0.00044323166,0.000064004475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001404728,0.0011477525,0.001921644,0.0014664005,0.0008669548,0.0014029907,0.0011617878,0.0012830946,0.0018418202],"category_scores_gemma":[0.0034287958,0.000483963,0.0009870849,0.0015189716,0.00044957382,0.0015786438,0.0014124507,0.0012632963,0.00086242246],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014519702,0.00007677205,0.0026378895,0.00016880277,0.00032090655,0.00009492441,0.00015771392,0.6790947,0.0052272375,0.019797932,0.008650615,0.28362727],"study_design_scores_gemma":[0.0000068517706,0.000021983642,0.00037863586,0.000009128462,0.00001874565,0.000025547237,0.000016530754,0.99075925,0.00112557,0.0043172413,0.0033102226,0.00001027804],"about_ca_topic_score_codex":0.0048714424,"about_ca_topic_score_gemma":0.004059137,"teacher_disagreement_score":0.0048714424,"about_ca_system_score_codex":0.000772338,"about_ca_system_score_gemma":0.0009137936,"threshold_uncertainty_score":0.009686172},"labels":[],"label_agreement":null},{"id":"W2010971425","doi":"10.1007/s10618-011-0212-3","title":"Measuring the component overlapping in the Gaussian mixture model","year":2011,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Gaussian; Computer science; Algorithm; Set (abstract data type); Component (thermodynamics); Multivariate normal distribution; Mixture model; Measure (data warehouse); Mathematics; Data mining; Multivariate statistics; Pattern recognition (psychology); Artificial intelligence; Machine learning","score_opus":0.16375358936821682,"score_gpt":0.3177989488666944,"score_spread":0.15404535949847759,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2010971425","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.104691975,0.0005797641,0.8925421,0.00016761836,0.00004299302,0.00004140039,0.00009155746,0.00034483965,0.0014977237],"genre_scores_gemma":[0.7774403,0.0006471138,0.2202113,0.00010289531,0.00009189534,0.00009057267,0.0004989449,0.00019375661,0.00072315236],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99365073,0.0023711636,0.00030646432,0.0011681983,0.0020321736,0.00047132323],"domain_scores_gemma":[0.98016584,0.013533266,0.0011743645,0.002590504,0.002015208,0.00052077783],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075767036,0.0011496784,0.0017119062,0.0037605762,0.0012339038,0.003047348,0.0021406831,0.0027547693,0.0011378662],"category_scores_gemma":[0.042159148,0.00081434485,0.0012458283,0.0037610896,0.0020286303,0.005212673,0.0029722885,0.0017939276,0.0005117311],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017720963,0.00038802528,0.055629756,0.00055310264,0.0010347493,0.0004554652,0.0016608009,0.5435554,0.026670482,0.101823375,0.0029652668,0.26349142],"study_design_scores_gemma":[0.000015135876,0.00007522311,0.008059748,0.000028314049,0.0001262048,0.00032398463,0.00013852125,0.9529648,0.0042932983,0.032975197,0.00093927386,0.00006025826],"about_ca_topic_score_codex":0.0048893066,"about_ca_topic_score_gemma":0.0027131308,"teacher_disagreement_score":0.0075767036,"about_ca_system_score_codex":0.0011677141,"about_ca_system_score_gemma":0.0012546225,"threshold_uncertainty_score":0.040069938},"labels":[],"label_agreement":null},{"id":"W2011191813","doi":"10.1007/s10115-009-0226-y","title":"Subspace and projected clustering: experimental evaluation and analysis","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Subspace topology; Computer science; Data mining; Range (aeronautics); Clustering high-dimensional data; Consensus clustering; Artificial intelligence; Correlation clustering; CURE data clustering algorithm; Engineering","score_opus":0.0260607623732046,"score_gpt":0.33560897402197565,"score_spread":0.30954821164877105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2011191813","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40804046,0.006439429,0.5602527,0.0006380794,0.0008176843,0.0014385884,0.00501834,0.008960683,0.008393994],"genre_scores_gemma":[0.52160794,0.0017161208,0.4629073,0.00016908461,0.00013561579,0.0006875365,0.008576652,0.0010966514,0.0031031456],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9872727,0.0064764973,0.00074814237,0.0015509478,0.0036129134,0.00033878352],"domain_scores_gemma":[0.971338,0.014628467,0.0008990652,0.004903384,0.007769272,0.00046187983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013229835,0.001851313,0.0018636707,0.0032065532,0.001857989,0.0025930197,0.0022148916,0.0017709277,0.0035736097],"category_scores_gemma":[0.03895741,0.0005320774,0.0011267269,0.004688549,0.0012483097,0.003248808,0.0026502665,0.00090894464,0.0015774567],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00834869,0.0021639126,0.010794014,0.0030181555,0.0020957224,0.00012254338,0.00085040875,0.15634094,0.019388966,0.0060581276,0.0169387,0.77387977],"study_design_scores_gemma":[0.0007027799,0.0023358054,0.012846154,0.00012653916,0.00073565694,0.00052775413,0.000969531,0.9347551,0.03266923,0.008646244,0.0055059553,0.00017932878],"about_ca_topic_score_codex":0.008395977,"about_ca_topic_score_gemma":0.010121288,"teacher_disagreement_score":0.013229835,"about_ca_system_score_codex":0.00094478595,"about_ca_system_score_gemma":0.001399791,"threshold_uncertainty_score":0.06996685},"labels":[],"label_agreement":null},{"id":"W2012187238","doi":"10.1109/isspa.2012.6310493","title":"Improving X-means clustering with MNDL","year":2012,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Cluster analysis; Bayesian information criterion; Determining the number of clusters in a data set; Computer science; Correlation clustering; Single-linkage clustering; Set (abstract data type); Cluster (spacecraft); Data mining; CURE data clustering algorithm; Mathematics; Algorithm; Artificial intelligence","score_opus":0.019568434327145962,"score_gpt":0.27481909939934496,"score_spread":0.255250665072199,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012187238","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011174834,0.00032202495,0.98714733,0.00016310017,0.00003332875,0.000042028034,0.000046303332,0.00063898147,0.00043212483],"genre_scores_gemma":[0.20028166,0.0003635513,0.7953023,0.0004454704,0.0000961855,0.00026333585,0.0008465232,0.00024389986,0.002157138],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99613035,0.0014865507,0.00027645595,0.00074448803,0.0011522047,0.00020997907],"domain_scores_gemma":[0.99570113,0.001877743,0.00029810055,0.0007691316,0.0012253271,0.000128562],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043111267,0.0013457194,0.0019614915,0.0020475588,0.0013248597,0.0016945604,0.0024904856,0.0021058766,0.0010790208],"category_scores_gemma":[0.012411336,0.0007720874,0.0013679229,0.001725065,0.0011183876,0.002521604,0.0035872785,0.0018378936,0.00096994557],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043679573,0.00018767566,0.003974384,0.0002948346,0.0002741794,0.000114036404,0.00062022073,0.51686376,0.010792339,0.025412424,0.006566242,0.43446314],"study_design_scores_gemma":[0.000020374615,0.000031874282,0.00022364133,0.000008897971,0.000014200812,0.000026775515,0.000032940727,0.98966366,0.002162894,0.006852911,0.00094748824,0.000014320711],"about_ca_topic_score_codex":0.00914634,"about_ca_topic_score_gemma":0.007150388,"teacher_disagreement_score":0.00914634,"about_ca_system_score_codex":0.0017180576,"about_ca_system_score_gemma":0.002133785,"threshold_uncertainty_score":0.02279973},"labels":[],"label_agreement":null},{"id":"W2012862892","doi":"10.1007/s10462-004-2901-4","title":"Statistical, Evolutionary, and Neurocomputing Clustering Techniques: Cluster-Based vs Object-Based Approaches","year":2005,"lang":"en","type":"article","venue":"Artificial Intelligence Review","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"Natural Sciences and Engineering Research Council of Canada; University of Regina","keywords":"Cluster analysis; Computer science; Artificial neural network; Set (abstract data type); Self-organizing map; Artificial intelligence; Hierarchical clustering; Data mining; Genetic algorithm; Object (grammar); Machine learning; Correlation clustering; Cluster (spacecraft); Pattern recognition (psychology)","score_opus":0.10217262922586404,"score_gpt":0.35377746189472176,"score_spread":0.2516048326688577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012862892","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013651658,0.49575004,0.45991316,0.0054376028,0.0018870395,0.00013193057,0.00017588957,0.00037635854,0.022676427],"genre_scores_gemma":[0.1801836,0.41108918,0.39344388,0.0018299524,0.004105698,0.00023971939,0.00035209878,0.00017139179,0.008584448],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998187,0.00040365505,0.000082842605,0.00021409967,0.0010572007,0.00005511113],"domain_scores_gemma":[0.99697614,0.0013607404,0.00018658515,0.00014020196,0.0012744865,0.00006185254],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028482324,0.0006854215,0.0018233905,0.002977301,0.0005491352,0.002290635,0.0019658061,0.0014604512,0.0015284547],"category_scores_gemma":[0.0057062646,0.00028610733,0.00077985233,0.00797277,0.0014777284,0.002189368,0.0005525129,0.0011608065,0.00087684725],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012882419,0.00009896179,0.002974082,0.0026314692,0.00040712743,0.00007818748,0.0002229998,0.01818401,0.0029291925,0.09921184,0.010749775,0.86238354],"study_design_scores_gemma":[0.00013488822,0.0006028212,0.029798696,0.0026303646,0.0010592372,0.0027991408,0.0017596089,0.19918881,0.012018823,0.3911268,0.35850742,0.00037333198],"about_ca_topic_score_codex":0.002963891,"about_ca_topic_score_gemma":0.004264153,"teacher_disagreement_score":0.002977301,"about_ca_system_score_codex":0.0011994039,"about_ca_system_score_gemma":0.0013473155,"threshold_uncertainty_score":0.015063107},"labels":[],"label_agreement":null},{"id":"W2013358657","doi":"10.2333/bhmk.41.115","title":"A General Formulation of Cluster Analysis with Dimension Reduction and Subspace Separation","year":2014,"lang":"en","type":"article","venue":"Behaviormetrika","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Japan Society for the Promotion of Science","keywords":"Subspace topology; Dimensionality reduction; Dimension (graph theory); Cluster analysis; Reduction (mathematics); Mathematics; Cluster (spacecraft); Factorial; Separation (statistics); Principal component analysis; Extension (predicate logic); Computer science; Pattern recognition (psychology); Algorithm; Artificial intelligence; Data mining; Statistics; Combinatorics","score_opus":0.019064604396722078,"score_gpt":0.32239569922186634,"score_spread":0.30333109482514425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2013358657","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00085327175,0.00024455856,0.9965305,0.00031049183,0.00008475852,0.00003100141,0.00008859994,0.000054259523,0.0018025852],"genre_scores_gemma":[0.08006937,0.0014322028,0.904025,0.00071422476,0.00086936774,0.00045233048,0.0006767967,0.00021620472,0.011544452],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973749,0.0010211796,0.00012736133,0.00062668946,0.0006983603,0.00015145418],"domain_scores_gemma":[0.9986125,0.00042746874,0.00012304597,0.0003250482,0.00044008766,0.00007171707],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025828823,0.0012939847,0.0014448735,0.0016910029,0.0013127896,0.0026903423,0.0032410717,0.0019292587,0.0041520004],"category_scores_gemma":[0.0048693414,0.0006317445,0.0019310833,0.0034152644,0.0024781958,0.0037564235,0.0030980587,0.003002077,0.0020284788],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003147048,0.000036421894,0.00027962957,0.0002211605,0.000082461345,0.00010292198,0.0002352421,0.048401617,0.0029790115,0.89612097,0.007471937,0.04403721],"study_design_scores_gemma":[0.00001304298,0.00003303418,0.00035450963,0.00003519684,0.00002896478,0.00018063083,0.00009422281,0.34126225,0.0008803551,0.64276636,0.014313279,0.000038187853],"about_ca_topic_score_codex":0.0034294296,"about_ca_topic_score_gemma":0.003751065,"teacher_disagreement_score":0.0041520004,"about_ca_system_score_codex":0.0012205627,"about_ca_system_score_gemma":0.001768765,"threshold_uncertainty_score":0.013889849},"labels":[],"label_agreement":null},{"id":"W2013578504","doi":"10.1109/icassp.2014.6855216","title":"Efficient unimodality test in clustering by signature testing","year":2014,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Unimodality; Signature (topology); Cluster analysis; Hierarchical clustering; Statistic; Statistical hypothesis testing; Computer science; Test statistic; Algorithm; Mathematics; Pattern recognition (psychology); Data mining; Statistics; Artificial intelligence","score_opus":0.01849794104842726,"score_gpt":0.28076863114923545,"score_spread":0.26227069010080817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2013578504","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04410692,0.0002218843,0.9531594,0.00011200262,0.000045718687,0.00008108443,0.00013966029,0.0011536379,0.0009797523],"genre_scores_gemma":[0.599966,0.00019907107,0.39706862,0.00018142021,0.00012201636,0.0002736914,0.0008398944,0.00026955898,0.0010797053],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9931804,0.0025402377,0.00040759664,0.00124071,0.0022692934,0.0003617638],"domain_scores_gemma":[0.9811756,0.012304591,0.0013775284,0.001826852,0.0026743156,0.0006411535],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004477215,0.0008528507,0.0015999014,0.0038324671,0.0009539925,0.0015268226,0.0021903047,0.0013450666,0.002634271],"category_scores_gemma":[0.027636709,0.00038645978,0.00083864876,0.0027419718,0.002261834,0.00271242,0.002376926,0.0016726101,0.001291682],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016384622,0.00038201545,0.018030176,0.00035970137,0.00026060187,0.00046636644,0.0003398236,0.15708472,0.043216377,0.04105475,0.0035897575,0.7335772],"study_design_scores_gemma":[0.000065383174,0.00034978273,0.00383499,0.00002037554,0.0000423269,0.00050450413,0.00010794049,0.9312458,0.028459743,0.03336744,0.0018964229,0.000105305495],"about_ca_topic_score_codex":0.001254346,"about_ca_topic_score_gemma":0.0007575759,"teacher_disagreement_score":0.004477215,"about_ca_system_score_codex":0.0009639,"about_ca_system_score_gemma":0.0020831572,"threshold_uncertainty_score":0.023678064},"labels":[],"label_agreement":null},{"id":"W2020287868","doi":"10.1016/j.eswa.2007.11.045","title":"A genetic fuzzy <mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" altimg=\"si218.gif\" overflow=\"scroll\"><mml:mrow><mml:mi>k</mml:mi></mml:mrow></mml:math>-Modes algorithm for clustering categorical data","year":2007,"lang":"lv","type":"article","venue":"Expert Systems with Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Algorithm; Fuzzy logic; Crossover; Categorical variable; Computer science; Genetic algorithm; Operator (biology); Cluster analysis; Mathematics; Artificial intelligence; Machine learning","score_opus":0.034403805891581714,"score_gpt":0.29495871503319476,"score_spread":0.26055490914161306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2020287868","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008415041,0.00007960118,0.9840109,0.00023249422,0.00007005004,0.00009829771,0.0003482506,0.0008099753,0.005935405],"genre_scores_gemma":[0.0777313,0.00010412305,0.9109325,0.00018553941,0.000040742754,0.00019030052,0.0007285099,0.00015826891,0.009928742],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990734,0.00014823006,0.00004404081,0.00025891335,0.00042165854,0.000053769403],"domain_scores_gemma":[0.9991124,0.00026624068,0.000041910494,0.00013845158,0.00039444646,0.000046456742],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012238625,0.00042908717,0.0006065027,0.0015195856,0.0011399243,0.00144051,0.002129828,0.0013203941,0.0068039345],"category_scores_gemma":[0.0041244444,0.00035092,0.0010874004,0.0016498184,0.00070089777,0.00070406473,0.0008425842,0.0011579462,0.0023952548],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000205394,0.00018911317,0.0023041272,0.00018631626,0.0001334223,0.00017146883,0.00035146295,0.21650404,0.015859602,0.08961814,0.017643727,0.65683323],"study_design_scores_gemma":[0.000042922667,0.000074712465,0.0006493525,0.000048076243,0.0000525729,0.0001478668,0.0000673916,0.95056814,0.007548164,0.029149188,0.011610466,0.00004110689],"about_ca_topic_score_codex":0.024853459,"about_ca_topic_score_gemma":0.02555266,"teacher_disagreement_score":0.024853459,"about_ca_system_score_codex":0.0019123082,"about_ca_system_score_gemma":0.0024847584,"threshold_uncertainty_score":0.049417615},"labels":[],"label_agreement":null},{"id":"W2020492796","doi":"10.2333/bhmk.28.153","title":"Space Distortion and Monotone Admissibility in Agglomerative Clustering","year":2001,"lang":"en","type":"article","venue":"Behaviormetrika","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Paul University","funders":"","keywords":"Cluster analysis; Distortion (music); Monotone polygon; Hierarchical clustering; Mathematics; Monotonic function; Space (punctuation); Algorithm; Computer science; Statistics","score_opus":0.03308515863390581,"score_gpt":0.33067564014191975,"score_spread":0.2975904815080139,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2020492796","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10198211,0.0017578935,0.8864825,0.0023518414,0.00010904397,0.00010690672,0.00021535656,0.00020788406,0.006786513],"genre_scores_gemma":[0.73030376,0.0017306459,0.2563318,0.000434277,0.0003567591,0.00031338382,0.00069859496,0.0003892292,0.009441508],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.988456,0.0062554125,0.000639853,0.0017355065,0.0022066983,0.00070663384],"domain_scores_gemma":[0.9051592,0.06951631,0.005482104,0.009947974,0.0066116774,0.0032827505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016709246,0.0016170156,0.0032767367,0.0031425597,0.0032802194,0.0051871114,0.0074592414,0.004308512,0.0036593527],"category_scores_gemma":[0.092156455,0.0019033452,0.0024640427,0.0044629383,0.008880021,0.01177357,0.0065221004,0.00653925,0.0006570347],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005093199,0.00012858675,0.003197092,0.00034031717,0.00013975076,0.00017441562,0.0010043739,0.12041954,0.0017695245,0.8450687,0.0019433025,0.025305154],"study_design_scores_gemma":[0.000045825178,0.000118239994,0.00085756736,0.000047454825,0.00003914133,0.0002634634,0.00017856363,0.32626864,0.0010142163,0.66956776,0.0015490474,0.000050123916],"about_ca_topic_score_codex":0.0038437338,"about_ca_topic_score_gemma":0.003143755,"teacher_disagreement_score":0.016709246,"about_ca_system_score_codex":0.0052734087,"about_ca_system_score_gemma":0.0023810475,"threshold_uncertainty_score":0.08836806},"labels":[],"label_agreement":null},{"id":"W2022006917","doi":"10.1145/1645953.1646069","title":"Clustering web queries","year":2009,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Weighting; Classifier (UML); Set (abstract data type); Population; Search engine; Information retrieval; Metric (unit); Artificial intelligence","score_opus":0.018877849317098013,"score_gpt":0.3006797422883326,"score_spread":0.2818018929712346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2022006917","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45175213,0.0031227244,0.5162309,0.00086865976,0.00026947036,0.0026156488,0.008343474,0.0060230186,0.010774006],"genre_scores_gemma":[0.59794754,0.00064183533,0.3789138,0.0002359231,0.00016740481,0.0006054989,0.017177522,0.0005495529,0.003760896],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9896288,0.002486056,0.001045799,0.0022788106,0.0040921774,0.0004684131],"domain_scores_gemma":[0.9841325,0.0056608394,0.0010153676,0.0031251675,0.0057897926,0.00027636415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005239871,0.0014596254,0.0014732421,0.010397583,0.0015028327,0.003605348,0.0017784534,0.0014841987,0.0011759906],"category_scores_gemma":[0.022911793,0.00033474228,0.001321454,0.006619292,0.000658394,0.0024866113,0.0010687287,0.0008734359,0.0010546754],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010467567,0.0012051599,0.10303855,0.0014504303,0.0009951032,0.0003160553,0.0020400062,0.09833182,0.035228666,0.011506995,0.026095679,0.7187448],"study_design_scores_gemma":[0.00010400596,0.00059995824,0.073924795,0.00017888748,0.00030523358,0.0013200055,0.0023104632,0.8230383,0.04919229,0.021388426,0.027407045,0.00023056711],"about_ca_topic_score_codex":0.010476983,"about_ca_topic_score_gemma":0.012305375,"teacher_disagreement_score":0.010476983,"about_ca_system_score_codex":0.0022819303,"about_ca_system_score_gemma":0.0015376067,"threshold_uncertainty_score":0.027711451},"labels":[],"label_agreement":null},{"id":"W2022201359","doi":"10.1016/j.eswa.2013.07.002","title":"Cluster center initialization algorithm for K-modes clustering","year":2013,"lang":"en","type":"article","venue":"Expert Systems with Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":120,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Initialization; Computer science; Cluster analysis; Cluster (spacecraft); Center (category theory); Algorithm; Data mining; Artificial intelligence; Pattern recognition (psychology); Computer network","score_opus":0.02231767915479643,"score_gpt":0.304274552149248,"score_spread":0.28195687299445155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2022201359","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025500455,0.0001486374,0.9949216,0.00005938262,0.00008269643,0.000067016124,0.00010761667,0.000990444,0.0010725566],"genre_scores_gemma":[0.07133662,0.00016959292,0.9226873,0.000077120894,0.000052779997,0.00024573097,0.0007641893,0.0004059235,0.004260613],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988613,0.00025563993,0.00006858325,0.00029303753,0.00039141582,0.00013008782],"domain_scores_gemma":[0.99860674,0.00025961504,0.00006609828,0.00023903386,0.00075892534,0.0000696179],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012627749,0.0010355863,0.001137247,0.0016456352,0.0018583658,0.0014491072,0.0025928507,0.0014765321,0.006314137],"category_scores_gemma":[0.00413323,0.000676921,0.0010272211,0.002037435,0.0006528165,0.001280901,0.0016088133,0.0021921166,0.0047424845],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008200851,0.00014894202,0.0015148624,0.00027587908,0.00018559818,0.000113702474,0.00041735786,0.17120343,0.018519975,0.027605856,0.030791426,0.7484029],"study_design_scores_gemma":[0.00006263667,0.00005237019,0.0008768007,0.000036911606,0.00004077988,0.00013171026,0.000092408605,0.96120375,0.013099275,0.013819383,0.010521603,0.0000623698],"about_ca_topic_score_codex":0.011170032,"about_ca_topic_score_gemma":0.013015608,"teacher_disagreement_score":0.011170032,"about_ca_system_score_codex":0.0011322491,"about_ca_system_score_gemma":0.0027797627,"threshold_uncertainty_score":0.022210002},"labels":[],"label_agreement":null},{"id":"W2023420371","doi":"10.5539/mas.v3n2p75","title":"Genetic Algorithm for Document Clustering with Simultaneous and Ranked Mutation","year":2009,"lang":"en","type":"article","venue":"Modern Applied Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Mutation; Mutation rate; Computer science; Genetic algorithm; Population; Similarity (geometry); Local optimum; Operator (biology); Chromosome; Algorithm; Data mining; Correlation clustering; Document clustering; Artificial intelligence; Machine learning; Genetics; Biology","score_opus":0.010406362118062418,"score_gpt":0.2725208109599084,"score_spread":0.262114448841846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2023420371","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010789814,0.00063636503,0.9856479,0.00013106305,0.00005555897,0.00010719061,0.000058066154,0.0007250915,0.0018488838],"genre_scores_gemma":[0.18982132,0.00079669117,0.8031501,0.00010413395,0.000064207765,0.0005327803,0.00038095322,0.00013138655,0.0050184075],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99919814,0.00022131673,0.000044144,0.00016883793,0.00030565631,0.00006190012],"domain_scores_gemma":[0.9995377,0.0002074937,0.00003714329,0.00004555142,0.00015295002,0.000019123756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093131687,0.000904929,0.0011771483,0.0015746691,0.0009299031,0.00091762026,0.0015864968,0.0011152168,0.0012346096],"category_scores_gemma":[0.0020940974,0.000348095,0.00084739976,0.0021019313,0.0006527441,0.0008269784,0.00066835445,0.0011078888,0.0005601149],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009556787,0.000097037584,0.0010706186,0.00014531102,0.00015319006,0.00016083283,0.00021649625,0.6618096,0.0067300666,0.020908449,0.004026907,0.30458587],"study_design_scores_gemma":[0.00004607107,0.000044746288,0.00030007606,0.000015488684,0.00003068822,0.00011321529,0.000024178946,0.985269,0.0021708023,0.008385441,0.0035744496,0.000025802647],"about_ca_topic_score_codex":0.010424407,"about_ca_topic_score_gemma":0.006245771,"teacher_disagreement_score":0.010424407,"about_ca_system_score_codex":0.0014350762,"about_ca_system_score_gemma":0.0018593771,"threshold_uncertainty_score":0.020727456},"labels":[],"label_agreement":null},{"id":"W2026643052","doi":"10.1007/s11047-009-9173-5","title":"Flocking based approach for data clustering","year":2009,"lang":"en","type":"article","venue":"Natural Computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Particle swarm optimization; Fuzzy clustering; Correlation clustering; CURE data clustering algorithm; Data mining; Canopy clustering algorithm; Artificial intelligence; Consensus clustering; Machine learning","score_opus":0.0709870533393512,"score_gpt":0.3606660983964147,"score_spread":0.28967904505706354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2026643052","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00535459,0.00025483462,0.9931467,0.00008780608,0.000059758462,0.00008611967,0.000043765947,0.00017581737,0.00079045567],"genre_scores_gemma":[0.14284004,0.00049369206,0.8498669,0.00019248945,0.00012402183,0.0003451483,0.00043781367,0.00011881843,0.005581049],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985745,0.00038511338,0.00009481851,0.000320956,0.0005300054,0.000094586765],"domain_scores_gemma":[0.99815613,0.0006617058,0.00012893984,0.00034974687,0.00061384536,0.00008956667],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016999834,0.00070724153,0.0015098765,0.00391337,0.0020142635,0.0011439295,0.0023856848,0.0015614615,0.0025963676],"category_scores_gemma":[0.0051820865,0.0004705577,0.0014782266,0.0028821607,0.0010950749,0.0019310713,0.0016972359,0.0015010893,0.0007621138],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003485145,0.00025724724,0.0028946844,0.0005285127,0.00041065767,0.0002556814,0.00085031026,0.2502554,0.021317448,0.12179348,0.010850624,0.5902375],"study_design_scores_gemma":[0.00002142241,0.00011235207,0.0010906358,0.000038338934,0.00005364423,0.00025486597,0.00012404259,0.9450967,0.003612828,0.043320447,0.0062374496,0.00003732292],"about_ca_topic_score_codex":0.0069892886,"about_ca_topic_score_gemma":0.0071506756,"teacher_disagreement_score":0.0069892886,"about_ca_system_score_codex":0.0013336696,"about_ca_system_score_gemma":0.0010245362,"threshold_uncertainty_score":0.01389724},"labels":[],"label_agreement":null},{"id":"W2028768445","doi":"10.1109/ifsa-nafips.2013.6608463","title":"A granular recursive fuzzy meta-clustering algorithm for social networks","year":2013,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Data mining; Computer science; Fuzzy logic; FLAME clustering; Artificial intelligence; Canopy clustering algorithm","score_opus":0.049427383107694296,"score_gpt":0.3102288841149434,"score_spread":0.26080150100724914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2028768445","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040623834,0.00013629963,0.99450004,0.000060003047,0.00001953692,0.000043135726,0.00003913252,0.00026081115,0.0008786455],"genre_scores_gemma":[0.13432215,0.00018182085,0.8630131,0.00006594321,0.00002946256,0.00017859325,0.0002343466,0.00012233212,0.0018522843],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99896955,0.00024857884,0.000076393684,0.00023840435,0.0003840032,0.00008308914],"domain_scores_gemma":[0.9991284,0.00028092606,0.00010524354,0.00016625914,0.00027280036,0.000046348534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001391867,0.0007605945,0.0012111708,0.0025993846,0.0012902349,0.0018746323,0.0023125878,0.0014147296,0.002052911],"category_scores_gemma":[0.003505002,0.0005137206,0.0014821958,0.002383527,0.0008114695,0.002273969,0.0015277431,0.0013389807,0.00082304096],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016652285,0.00008860881,0.0014495683,0.00018114144,0.00018285138,0.00013157862,0.0004798999,0.51523304,0.0072294576,0.09820419,0.004128319,0.37252486],"study_design_scores_gemma":[0.00001084084,0.000021405787,0.00020607225,0.000016166645,0.000015611555,0.000052139756,0.00004396184,0.97285044,0.0011561171,0.02307567,0.0025332023,0.000018367402],"about_ca_topic_score_codex":0.0064919377,"about_ca_topic_score_gemma":0.0062214783,"teacher_disagreement_score":0.0064919377,"about_ca_system_score_codex":0.0018853525,"about_ca_system_score_gemma":0.0011471469,"threshold_uncertainty_score":0.013679266},"labels":[],"label_agreement":null},{"id":"W2029272463","doi":"10.1016/j.ejor.2011.01.016","title":"Globally optimal clusterwise regression by mixed logical-quadratic programming","year":2011,"lang":"en","type":"article","venue":"European Journal of Operational Research","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"HEC Montréal; Group for Research in Decision Analysis","funders":"","keywords":"Computer science; Regression analysis; Integer programming; Quadratic equation; Mathematical optimization; Operations research; Mathematics; Statistics","score_opus":0.1333836776367603,"score_gpt":0.376373881362056,"score_spread":0.24299020372529573,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2029272463","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002895299,0.000092352195,0.99596715,0.00019124652,0.000026871252,0.000025088071,0.000046972804,0.00027385185,0.00048119418],"genre_scores_gemma":[0.1649798,0.00019144447,0.82586825,0.00032635135,0.00015181291,0.00045495713,0.0005728436,0.0008855866,0.0065688626],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958402,0.0026263765,0.000120614604,0.0006329706,0.0005265382,0.00025337652],"domain_scores_gemma":[0.99160016,0.006047683,0.00039796537,0.00051165884,0.0011461297,0.0002963257],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070080385,0.002264598,0.0035225379,0.0012051746,0.0011218842,0.0022449072,0.004205144,0.0031299295,0.004692959],"category_scores_gemma":[0.016793834,0.0018070816,0.0017793442,0.0018377973,0.0024283654,0.0028268758,0.004156441,0.003443881,0.0015190851],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028087397,0.000116534706,0.00041988379,0.0002143984,0.0001179427,0.000047848516,0.0000883392,0.8927337,0.0018198593,0.043870725,0.005682905,0.054606903],"study_design_scores_gemma":[0.000013196234,0.000020234176,0.00002620848,0.0000041921126,0.0000059419935,0.000005504819,0.000005531579,0.99018705,0.0002844829,0.009210305,0.00023194407,0.0000055252312],"about_ca_topic_score_codex":0.0063670417,"about_ca_topic_score_gemma":0.0066756536,"teacher_disagreement_score":0.0070080385,"about_ca_system_score_codex":0.0020061152,"about_ca_system_score_gemma":0.0045034285,"threshold_uncertainty_score":0.037062466},"labels":[],"label_agreement":null},{"id":"W2029698681","doi":"10.1145/2395116.2395117","title":"The effectiveness of lloyd-type methods for the k-means problem","year":2012,"lang":"en","type":"article","venue":"Journal of the ACM","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":185,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Office of Naval Research; Israel Science Foundation; United States-Israel Binational Science Foundation; National Science Foundation","keywords":"Computer science; Cluster analysis; Heuristic; Probabilistic logic; Popularity; Process (computing); Algorithm; Type (biology); Quality (philosophy); Data mining; Artificial intelligence; Mathematical optimization; Mathematics","score_opus":0.05066538846456285,"score_gpt":0.4219683914484055,"score_spread":0.3713030029838426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2029698681","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0057546017,0.0004963841,0.99025214,0.00032455206,0.000058707763,0.0000826612,0.000029015371,0.00028952584,0.002712403],"genre_scores_gemma":[0.12982607,0.0006841094,0.86510026,0.00034249324,0.00015937655,0.00036768944,0.00018722551,0.0003585848,0.0029742625],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9943504,0.002783767,0.00029491333,0.0008053196,0.0014359428,0.00032973496],"domain_scores_gemma":[0.98251927,0.012685323,0.0008956263,0.0017143651,0.001682381,0.00050295214],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010329079,0.0020699115,0.0024521437,0.002573962,0.0021657506,0.0039461697,0.004857445,0.004114964,0.003192442],"category_scores_gemma":[0.04864528,0.0015001183,0.001993753,0.002356225,0.0034054369,0.00596827,0.0041668676,0.003744483,0.001633201],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031596163,0.0002196324,0.0018424125,0.0003823439,0.00015786583,0.000094485426,0.0004428424,0.6936516,0.0020592061,0.17239138,0.006618295,0.121823944],"study_design_scores_gemma":[0.00004710043,0.000071582646,0.00013996007,0.000047110225,0.000014313403,0.000041120806,0.000038699887,0.93569076,0.0011128739,0.06030454,0.002467051,0.000024844172],"about_ca_topic_score_codex":0.0065538962,"about_ca_topic_score_gemma":0.0051600346,"teacher_disagreement_score":0.010329079,"about_ca_system_score_codex":0.0029445565,"about_ca_system_score_gemma":0.0034216826,"threshold_uncertainty_score":0.054626048},"labels":[],"label_agreement":null},{"id":"W2032084149","doi":"10.1016/j.ins.2013.11.004","title":"Relative entropy fuzzy c-means clustering","year":2013,"lang":"en","type":"article","venue":"Information Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":113,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Pattern recognition (psychology); Artificial intelligence; Fuzzy logic; Data mining; Entropy (arrow of time); Computer science; FLAME clustering; Mathematics; Machine learning; CURE data clustering algorithm","score_opus":0.02682990070830507,"score_gpt":0.30065639858746507,"score_spread":0.27382649787916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2032084149","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023176039,0.00068235194,0.9707427,0.00018462277,0.00010219366,0.00009029652,0.00022195399,0.0003715697,0.0044282237],"genre_scores_gemma":[0.41910812,0.00053215655,0.57412136,0.000103023296,0.00015084074,0.00018451967,0.00083960925,0.00016996672,0.0047903885],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966486,0.00071064476,0.00024592402,0.0005395438,0.0017078375,0.00014733864],"domain_scores_gemma":[0.9972562,0.000877328,0.00017924453,0.00041126122,0.0012214611,0.000054454144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028140363,0.00069947884,0.0012909214,0.0038302399,0.0018825817,0.002466691,0.0015145125,0.0013299431,0.0030720078],"category_scores_gemma":[0.009360615,0.0003807458,0.0011409547,0.0030377035,0.0010285925,0.0018636307,0.001149554,0.00091537036,0.00082604744],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005813787,0.00013133867,0.0025515747,0.00055890175,0.00026732672,0.00015260767,0.0003792745,0.3873131,0.014473926,0.1259692,0.0073291166,0.46029225],"study_design_scores_gemma":[0.000017607703,0.00005858805,0.0017592515,0.000034414883,0.0000613569,0.00014778053,0.00006914468,0.9534481,0.008317977,0.031802624,0.004224219,0.000058953374],"about_ca_topic_score_codex":0.0047228723,"about_ca_topic_score_gemma":0.0038237807,"teacher_disagreement_score":0.0047228723,"about_ca_system_score_codex":0.0016642666,"about_ca_system_score_gemma":0.0016943518,"threshold_uncertainty_score":0.0148822665},"labels":[],"label_agreement":null},{"id":"W2032920196","doi":"10.1016/j.eswa.2011.09.154","title":"Evolutionary feature selection via structure retention","year":2011,"lang":"en","type":"article","venue":"Expert Systems with Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Granular computing; Curse of dimensionality; Cluster analysis; Computer science; Particle swarm optimization; Dimensionality reduction; Granulation; Fuzzy set; Artificial intelligence; Mathematics; Pattern recognition (psychology); Feature selection; Feature (linguistics); Fuzzy logic; Reduction (mathematics); Algorithm; Data mining; Rough set","score_opus":0.021056378203984503,"score_gpt":0.2632772240430977,"score_spread":0.2422208458391132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2032920196","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14763042,0.00027397773,0.84738654,0.00031917763,0.0000906877,0.0001365839,0.00006572745,0.0008331132,0.0032637706],"genre_scores_gemma":[0.7880882,0.00010087049,0.20646939,0.00023932822,0.00007870395,0.00016957874,0.00021584707,0.00019441277,0.0044436497],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994808,0.00015162728,0.000032559168,0.0001273075,0.00013193206,0.000075910255],"domain_scores_gemma":[0.99797255,0.0010031136,0.000116611154,0.00036536317,0.00045722086,0.00008522867],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013424949,0.0004783091,0.0010214006,0.0012233403,0.00065195194,0.0006918764,0.0014146132,0.0011301021,0.003253869],"category_scores_gemma":[0.004441292,0.00041066686,0.0009269892,0.0009769496,0.0006081635,0.001149381,0.0012641901,0.00087762205,0.00056776847],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039013365,0.0004005087,0.0049819145,0.00011796283,0.00019646013,0.0003143013,0.00022856965,0.27085584,0.033839043,0.020213202,0.003517244,0.6649449],"study_design_scores_gemma":[0.000056925222,0.00016816676,0.00079630833,0.000008754259,0.000053116848,0.000121621626,0.000026342661,0.9856908,0.004820104,0.0074360846,0.0008073806,0.000014385568],"about_ca_topic_score_codex":0.0008738563,"about_ca_topic_score_gemma":0.0013520714,"teacher_disagreement_score":0.003253869,"about_ca_system_score_codex":0.00038559444,"about_ca_system_score_gemma":0.00056802446,"threshold_uncertainty_score":0.010885298},"labels":[],"label_agreement":null},{"id":"W2034996776","doi":"10.2333/bhmk.40.1","title":"Hierarchically Structured Fuzzy c-Means Clustering","year":2013,"lang":"en","type":"article","venue":"Behaviormetrika","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Centroid; Fuzzy logic; Data mining; Cluster analysis; Fuzzy clustering; Cluster (spacecraft); Extension (predicate logic); Hierarchical clustering; Extant taxon; Single-linkage clustering; Pattern recognition (psychology); Mathematics; Hierarchy; Computer science; Artificial intelligence; FLAME clustering; CURE data clustering algorithm","score_opus":0.021141757707111235,"score_gpt":0.28846440525270595,"score_spread":0.2673226475455947,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2034996776","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016029336,0.00010537801,0.98132485,0.00009877704,0.000029063114,0.00010676241,0.00020628746,0.0003798939,0.001719723],"genre_scores_gemma":[0.24413542,0.00012380694,0.750479,0.00009707861,0.00003065594,0.00021018213,0.0011553251,0.00016280479,0.0036057818],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983614,0.00036023307,0.00009711227,0.00040773727,0.00062360754,0.0001499115],"domain_scores_gemma":[0.9979075,0.0003999822,0.00014350445,0.00037685348,0.0011044547,0.00006775064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012359027,0.0006655003,0.000698258,0.001973609,0.0015938886,0.0013162763,0.001966042,0.0011774449,0.002930979],"category_scores_gemma":[0.005119356,0.0003925253,0.0011302808,0.0020078486,0.00076013803,0.0009366335,0.0010765735,0.00095434656,0.001156007],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035340025,0.0002495698,0.00401665,0.000284255,0.00016446013,0.00013250051,0.0006126788,0.40880507,0.01847797,0.05342308,0.010850767,0.5026296],"study_design_scores_gemma":[0.000008362694,0.000026776881,0.0010441522,0.000015627904,0.000014687555,0.000041168038,0.00007528204,0.9804372,0.002475862,0.014199555,0.0016385638,0.000022675624],"about_ca_topic_score_codex":0.01693235,"about_ca_topic_score_gemma":0.02282965,"teacher_disagreement_score":0.01693235,"about_ca_system_score_codex":0.0015415584,"about_ca_system_score_gemma":0.002724607,"threshold_uncertainty_score":0.033667564},"labels":[],"label_agreement":null},{"id":"W2036163493","doi":"10.1145/1141277.1141422","title":"CLUC","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Cluster analysis; Categorical variable; Cohesion (chemistry); Data mining; Scalability; Similarity (geometry); Artificial intelligence; Pattern recognition (psychology); Machine learning","score_opus":0.007823885945915845,"score_gpt":0.25728392612711454,"score_spread":0.2494600401811987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2036163493","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002847955,0.0006393109,0.9599191,0.00046389506,0.00027563146,0.00030016023,0.0022230307,0.022026725,0.011304122],"genre_scores_gemma":[0.061682384,0.00049186934,0.90982795,0.00092975964,0.00018158599,0.00072558003,0.013222864,0.002077892,0.010860195],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960849,0.0008611615,0.00032272682,0.0011151286,0.0013588946,0.00025719352],"domain_scores_gemma":[0.99308115,0.0011099181,0.0003554537,0.0025992459,0.0026396306,0.00021453391],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00243337,0.0015249031,0.001161088,0.005619882,0.0020154505,0.00421994,0.0043036807,0.0020772684,0.016492065],"category_scores_gemma":[0.014918784,0.00069357606,0.0014397228,0.0053879344,0.000798881,0.0037317292,0.0034424935,0.0015448249,0.013296822],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022315182,0.000105043364,0.0021999406,0.0004701929,0.000100043224,0.00015656163,0.00022985981,0.033469494,0.0042258785,0.04463339,0.10496436,0.80922204],"study_design_scores_gemma":[0.00011184985,0.0001757774,0.0016206212,0.00020003051,0.000059790564,0.0007968518,0.00023162887,0.5206306,0.01627108,0.09526608,0.36448684,0.00014882644],"about_ca_topic_score_codex":0.006022548,"about_ca_topic_score_gemma":0.0061733196,"teacher_disagreement_score":0.98350793,"about_ca_system_score_codex":0.0020872361,"about_ca_system_score_gemma":0.0026850163,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2040963032","doi":"10.1142/s0219649211002973","title":"Effectiveness of Heuristic Based Approach on the Performance of Indexing and Clustering of High Dimensional Data","year":2011,"lang":"en","type":"article","venue":"Journal of Information & Knowledge Management","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Search engine indexing; Data mining; Curse of dimensionality; Hierarchical clustering; DBSCAN; Dimensionality reduction; Clustering high-dimensional data; Heuristic; CURE data clustering algorithm; Correlation clustering; Machine learning; Artificial intelligence","score_opus":0.04405171062011411,"score_gpt":0.2708279659291963,"score_spread":0.2267762553090822,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2040963032","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5142997,0.007109352,0.43462148,0.0010546495,0.0005202477,0.0003885241,0.0005330834,0.008743626,0.03272929],"genre_scores_gemma":[0.75311166,0.00069051853,0.24317905,0.00022234445,0.000070761635,0.00013152146,0.0005831626,0.00034751266,0.0016634729],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934737,0.0017175287,0.00040234413,0.0008458412,0.0031142456,0.00044628864],"domain_scores_gemma":[0.99023485,0.005802547,0.00075742224,0.0012563203,0.0017344612,0.00021453112],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043908083,0.0009709478,0.0013165696,0.0019125593,0.00097208173,0.0025149204,0.002032923,0.0015682733,0.0016225753],"category_scores_gemma":[0.015142499,0.00044522248,0.00060885533,0.0022085726,0.0009558503,0.001593124,0.00077516976,0.0008805781,0.00067051186],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021193959,0.0010035281,0.009805784,0.0007813816,0.0003538473,0.00025290338,0.00035318313,0.56090087,0.02912876,0.009925558,0.0054444377,0.37993044],"study_design_scores_gemma":[0.00010149095,0.0005296602,0.0036939993,0.000045335917,0.00006858856,0.00022866737,0.00014674668,0.9725337,0.016011592,0.0034951274,0.003091945,0.00005321005],"about_ca_topic_score_codex":0.005739831,"about_ca_topic_score_gemma":0.0049886894,"teacher_disagreement_score":0.005739831,"about_ca_system_score_codex":0.001878742,"about_ca_system_score_gemma":0.0024154636,"threshold_uncertainty_score":0.023221076},"labels":[],"label_agreement":null},{"id":"W2049769908","doi":"10.1007/bf02960771","title":"Clustering DTDs: An interactive two-level approach","year":2002,"lang":"en","type":"article","venue":"Journal of Computer Science and Technology","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Cluster analysis; Document type definition; Document Structure Description; XML; Information retrieval; Data mining; Artificial intelligence","score_opus":0.04570929441094207,"score_gpt":0.31330563583075194,"score_spread":0.26759634141980987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2049769908","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016715705,0.000041177096,0.99419034,0.0000952473,0.000021678059,0.00011145854,0.00010391654,0.0028195865,0.00094499416],"genre_scores_gemma":[0.029182145,0.00006700669,0.9673196,0.000077825985,0.000022345586,0.0001518771,0.0004287563,0.00074502657,0.0020054805],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99634296,0.00080269587,0.00027302175,0.0005442979,0.0017029622,0.0003340242],"domain_scores_gemma":[0.9954294,0.0020058742,0.00017908825,0.0010906238,0.0009670733,0.00032792432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002858201,0.0017212763,0.0024917994,0.00416361,0.002556429,0.006072129,0.0058384077,0.003626963,0.019718684],"category_scores_gemma":[0.008563782,0.0018329691,0.0032432403,0.0030596864,0.0012083942,0.004565442,0.006371656,0.0028463844,0.0040657506],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00078138633,0.00050595606,0.0019205969,0.0011903724,0.00035644675,0.0005514278,0.0015315933,0.14127722,0.04019524,0.07563295,0.023485076,0.71257174],"study_design_scores_gemma":[0.00009352887,0.00008321916,0.00039819564,0.000058872934,0.00008515065,0.00027349088,0.00026528357,0.92275864,0.015748441,0.041113276,0.019032383,0.00008949244],"about_ca_topic_score_codex":0.00611634,"about_ca_topic_score_gemma":0.012694247,"teacher_disagreement_score":0.019718684,"about_ca_system_score_codex":0.0019153676,"about_ca_system_score_gemma":0.0021878267,"threshold_uncertainty_score":0.06596553},"labels":[],"label_agreement":null},{"id":"W2051549497","doi":"10.1109/foci.2007.371524","title":"Fuzzy Clustering and Mapping of Ordinal Values to Numerical","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Feature (linguistics); Measure (data warehouse); Mathematics; Pattern recognition (psychology); Fuzzy set; Fuzzy logic; Ordinal optimization; Data mining; Artificial intelligence; Ordinal regression; Ordinal data; Object (grammar); Computer science; FLAME clustering; Algorithm; CURE data clustering algorithm; Statistics","score_opus":0.030559089090553892,"score_gpt":0.32468563521561433,"score_spread":0.29412654612506045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2051549497","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01742139,0.00023561675,0.97956467,0.0001173158,0.000038190887,0.000045044562,0.00005671772,0.000119998025,0.0024010045],"genre_scores_gemma":[0.29356882,0.00030608728,0.70287186,0.000044940134,0.0000436826,0.00015376115,0.00014988448,0.000051542203,0.0028093855],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.998268,0.00060359336,0.00009995538,0.0003718533,0.00057031243,0.000086318265],"domain_scores_gemma":[0.9977181,0.0010589511,0.00029304123,0.00033982488,0.00051939854,0.00007063762],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018904497,0.0004871705,0.0006183326,0.0019455669,0.0008246004,0.0016290009,0.00093784346,0.00068192167,0.0025174818],"category_scores_gemma":[0.010601174,0.00030263563,0.00066576485,0.0019990725,0.00155194,0.0018404191,0.0010333837,0.0009845819,0.00059179944],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019575762,0.00006583577,0.0029701574,0.0003042584,0.00010066078,0.0001859231,0.0010248362,0.35164726,0.0085804295,0.3340924,0.0018714577,0.298961],"study_design_scores_gemma":[0.000019228224,0.0000648879,0.00168892,0.000047506197,0.000019940222,0.00016840498,0.00018746736,0.74877656,0.00457001,0.23765954,0.006743546,0.0000539941],"about_ca_topic_score_codex":0.0023639267,"about_ca_topic_score_gemma":0.0014917473,"teacher_disagreement_score":0.0025174818,"about_ca_system_score_codex":0.0011860953,"about_ca_system_score_gemma":0.0007957322,"threshold_uncertainty_score":0.009997785},"labels":[],"label_agreement":null},{"id":"W2052493879","doi":"10.1109/cec.2012.6252895","title":"Symbiotic evolutionary subspace clustering","year":2012,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Institute for Materials Science; Natural Sciences and Engineering Research Council of Canada; Dalhousie University","keywords":"Cluster analysis; Data mining; Computer science; Centroid; Subspace topology; CURE data clustering algorithm; Clustering high-dimensional data; Correlation clustering; Single-linkage clustering; Canopy clustering algorithm; Cardinality (data modeling); Artificial intelligence; Pattern recognition (psychology)","score_opus":0.025139009780897566,"score_gpt":0.2933772242592968,"score_spread":0.26823821447839924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2052493879","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044352945,0.00027298645,0.94883806,0.00026545345,0.000048901737,0.00005874569,0.00004411736,0.0002857536,0.005833024],"genre_scores_gemma":[0.46826887,0.00037771373,0.5234591,0.00015574276,0.000039906372,0.00022298186,0.00024139213,0.00010405916,0.007130219],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99893457,0.00037127198,0.00004046157,0.00019049503,0.00036692183,0.000096314165],"domain_scores_gemma":[0.999175,0.00025101003,0.00006500508,0.00017344908,0.00027226153,0.000063239255],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001329781,0.0005979409,0.0008376481,0.0010069285,0.0008787878,0.0013256908,0.0012110757,0.0010638711,0.0017225147],"category_scores_gemma":[0.0030691861,0.0003316825,0.0007049624,0.0012207604,0.0009305878,0.0011238636,0.0025765367,0.0007055118,0.00051143364],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007238018,0.0000902429,0.0019634757,0.000067330184,0.00008007615,0.0001715504,0.0003954985,0.7130692,0.007165916,0.06361558,0.0017257226,0.211583],"study_design_scores_gemma":[0.0000083901305,0.000036836114,0.0002667481,0.000007148167,0.000005874608,0.00007444592,0.000049097653,0.9773301,0.0011942843,0.018018186,0.0029992566,0.000009671921],"about_ca_topic_score_codex":0.0021601624,"about_ca_topic_score_gemma":0.0017702745,"teacher_disagreement_score":0.0021601624,"about_ca_system_score_codex":0.0007059093,"about_ca_system_score_gemma":0.00084702764,"threshold_uncertainty_score":0.007032633},"labels":[],"label_agreement":null},{"id":"W2063036182","doi":"10.1504/ijdmb.2006.009920","title":"Bi-level clustering of mixed categorical and numerical biomedical data","year":2006,"lang":"en","type":"article","venue":"International Journal of Data Mining and Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Categorical variable; Cluster analysis; Data type; Computer science; Data mining; Artificial intelligence; Machine learning","score_opus":0.10190478292986778,"score_gpt":0.35695398446423293,"score_spread":0.25504920153436517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2063036182","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014177139,0.00027820998,0.9830005,0.00025581164,0.000039591137,0.0001327103,0.0005236165,0.00093672145,0.0006557521],"genre_scores_gemma":[0.113558084,0.0001591445,0.8808381,0.00025414806,0.000044324523,0.00044139303,0.0035087303,0.0002048828,0.0009911795],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9861137,0.0048648342,0.0014354355,0.0023585206,0.0045317938,0.0006956628],"domain_scores_gemma":[0.9808457,0.007942857,0.0015093823,0.0039037147,0.0053180316,0.00048030872],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008788549,0.0008889947,0.0016824435,0.0066573676,0.0016571253,0.004179042,0.0028796513,0.0019336712,0.0015851635],"category_scores_gemma":[0.038714204,0.00071442284,0.0024220126,0.007358713,0.0014938427,0.002877817,0.0042454153,0.0023634129,0.0016850083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013851052,0.0004713289,0.04300907,0.0013763146,0.0010638266,0.0004983288,0.0025693749,0.1135197,0.041616682,0.07877936,0.013475291,0.7022356],"study_design_scores_gemma":[0.00008546578,0.0001684616,0.01270235,0.00016300753,0.00015592099,0.00060130475,0.0005239307,0.7944673,0.018504215,0.15719897,0.0152473645,0.00018172461],"about_ca_topic_score_codex":0.003568818,"about_ca_topic_score_gemma":0.00529055,"teacher_disagreement_score":0.008788549,"about_ca_system_score_codex":0.0020696248,"about_ca_system_score_gemma":0.0026203352,"threshold_uncertainty_score":0.046478868},"labels":[],"label_agreement":null},{"id":"W2065468535","doi":"10.1504/ijsise.2013.051504","title":"A correlation based stochastic partitional algorithm for accurate cluster analysis","year":2013,"lang":"en","type":"article","venue":"International Journal of Signal and Imaging Systems Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Correlation; Cluster (spacecraft); Computer science; Algorithm; Artificial intelligence; Pattern recognition (psychology); Mathematics","score_opus":0.009204340715160656,"score_gpt":0.2568217004772642,"score_spread":0.24761735976210356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2065468535","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012310594,0.00006056745,0.99812406,0.00003717206,0.000020762322,0.000025468291,0.000018222869,0.00021901599,0.00026370268],"genre_scores_gemma":[0.059534058,0.00013524282,0.93842185,0.00009390841,0.0000614704,0.00022470298,0.00021284657,0.00016733962,0.0011485508],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99734026,0.0008035177,0.00014353837,0.00049548846,0.0010859263,0.00013130295],"domain_scores_gemma":[0.9976708,0.00090782,0.00023812734,0.00032919602,0.0007682469,0.00008579225],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025363234,0.00097139564,0.001225376,0.0018727711,0.0012330936,0.0013457162,0.0016160004,0.0013024564,0.0020615875],"category_scores_gemma":[0.007532747,0.00061960384,0.0011332365,0.0029451824,0.0011481001,0.0013268661,0.0017458342,0.0020585074,0.0013968374],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021140222,0.000076119824,0.0011584952,0.0001660214,0.00016044948,0.00014468387,0.00030395418,0.5212064,0.017029565,0.07296629,0.009031065,0.37754548],"study_design_scores_gemma":[0.000008790172,0.000022062208,0.00019788368,0.0000072424136,0.000008076915,0.00007107832,0.000010666084,0.98901725,0.001711774,0.0061601982,0.0027673268,0.00001766804],"about_ca_topic_score_codex":0.0047994335,"about_ca_topic_score_gemma":0.0042714416,"teacher_disagreement_score":0.0047994335,"about_ca_system_score_codex":0.0011939086,"about_ca_system_score_gemma":0.0022248153,"threshold_uncertainty_score":0.0134135485},"labels":[],"label_agreement":null},{"id":"W2067021889","doi":"10.1016/j.patcog.2008.08.037","title":"A distance-relatedness dynamic model for clustering high dimensional data of arbitrary shapes and densities","year":2008,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":45,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Outlier; Computer science; Algorithm; Cluster (spacecraft); Measure (data warehouse); Data mining; Correlation clustering; Determining the number of clusters in a data set; Visualization; Process (computing); CURE data clustering algorithm; Pattern recognition (psychology); Mathematics; Artificial intelligence","score_opus":0.0906646566591676,"score_gpt":0.30605559204328986,"score_spread":0.21539093538412224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2067021889","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008384351,0.00023593225,0.98981464,0.00030919892,0.000039212842,0.00004057691,0.00012533445,0.0002177593,0.0008330174],"genre_scores_gemma":[0.5316864,0.0015408648,0.4454771,0.00059370516,0.00034258587,0.0007696158,0.0016656368,0.00047943584,0.017444525],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975158,0.000813329,0.0001308439,0.0008892277,0.00047254865,0.00017826342],"domain_scores_gemma":[0.9966363,0.001576814,0.00040496845,0.0005870191,0.00060097565,0.00019391638],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038681782,0.0009953063,0.0021840117,0.0024053634,0.0012994423,0.0025519961,0.005831437,0.003312424,0.0031934772],"category_scores_gemma":[0.012459641,0.001189808,0.00210598,0.0037996704,0.0020379333,0.005450691,0.0029667134,0.0031161236,0.0019000934],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013910241,0.00011692169,0.0013940409,0.00011984131,0.0001280902,0.00011397774,0.00032656651,0.7653623,0.0018472972,0.15732473,0.0031816927,0.069945514],"study_design_scores_gemma":[0.000006429617,0.000019175986,0.00017231877,0.000007064691,0.000013216164,0.000039631617,0.000012742994,0.9690521,0.00012378817,0.029857319,0.00068080216,0.000015465532],"about_ca_topic_score_codex":0.012967914,"about_ca_topic_score_gemma":0.010084911,"teacher_disagreement_score":0.012967914,"about_ca_system_score_codex":0.0027172174,"about_ca_system_score_gemma":0.0015210087,"threshold_uncertainty_score":0.02578485},"labels":[],"label_agreement":null},{"id":"W2067475588","doi":"10.5539/ijsp.v2n1p73","title":"Modelling Students' Length of Stay at University Using Coxian Phase-type Distributions","year":2013,"lang":"en","type":"article","venue":"International Journal of Statistics and Probability","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Type (biology); Term (time); Phase (matter); Order (exchange); Mathematics education; Computer science; Sociology; Psychology; Business; Chemistry; Physics","score_opus":0.04458694132785159,"score_gpt":0.345093190856324,"score_spread":0.30050624952847244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2067475588","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8472493,0.00063080585,0.14752756,0.0009589486,0.00010434092,0.00020788104,0.0012680135,0.00011647543,0.0019366113],"genre_scores_gemma":[0.9885182,0.00031674054,0.007649764,0.000045825604,0.00005741124,0.00013567893,0.00080126565,0.000026551257,0.002448641],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978325,0.0009514996,0.00008956494,0.00044035094,0.00024913132,0.00043700726],"domain_scores_gemma":[0.97220653,0.02053932,0.0039056789,0.0010457906,0.0012915244,0.0010111588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00877984,0.00078598567,0.00077717344,0.0023772651,0.0006673005,0.0020342139,0.0022151438,0.0018158488,0.003916414],"category_scores_gemma":[0.028539425,0.0004016395,0.0016630739,0.0022519105,0.0011502634,0.0017297692,0.0013755228,0.0019652224,0.00072458433],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001548339,0.0005999132,0.38318115,0.00020222833,0.0005760598,0.00048203298,0.0020257502,0.48539254,0.0011881352,0.070847094,0.00373233,0.05022437],"study_design_scores_gemma":[0.0000653539,0.00029079593,0.041936,0.000036415673,0.00010047654,0.00011315028,0.00059127505,0.9377404,0.00029160295,0.017530143,0.0012552342,0.000049264017],"about_ca_topic_score_codex":0.017801989,"about_ca_topic_score_gemma":0.013072016,"teacher_disagreement_score":0.017801989,"about_ca_system_score_codex":0.0021054076,"about_ca_system_score_gemma":0.0015248626,"threshold_uncertainty_score":0.046432793},"labels":[],"label_agreement":null},{"id":"W2071486909","doi":"10.1007/s00357-009-9028-x","title":"Assessing Congruence Among Ultrametric Distance Matrices","year":2009,"lang":"en","type":"article","venue":"Journal of Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Ultrametric space; Mathematics; Resampling; Congruence (geometry); Type I and type II errors; Permutation (music); Mantel test; Distance matrices in phylogeny; Combinatorics; Statistics; Statistic; Normalization (sociology); Discrete mathematics; Biology; Genetics","score_opus":0.042940639641067625,"score_gpt":0.35797349114821986,"score_spread":0.3150328515071522,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2071486909","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72373146,0.0008354628,0.2674646,0.00052050024,0.00012095046,0.00014337603,0.00059398595,0.00027443454,0.0063153175],"genre_scores_gemma":[0.943681,0.00018629064,0.054724805,0.000063283514,0.000086683445,0.00010313232,0.00080908893,0.00006288752,0.00028281804],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9695267,0.01333373,0.0033042033,0.0039221253,0.009029598,0.0008835287],"domain_scores_gemma":[0.751346,0.18777448,0.016586322,0.01677227,0.023209965,0.00431097],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019001858,0.0007794474,0.0014065111,0.009130489,0.0020729208,0.0056097545,0.0024474557,0.0019978124,0.0023242335],"category_scores_gemma":[0.21541218,0.00090674247,0.0010457957,0.0067938017,0.003933574,0.008256392,0.006484689,0.001988923,0.000585747],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026797196,0.00049493514,0.3392873,0.00093288335,0.00092074525,0.00097181427,0.010040252,0.042650986,0.01748925,0.29076958,0.0037840703,0.28997847],"study_design_scores_gemma":[0.00022055654,0.0015095586,0.07179579,0.00036263195,0.0003030727,0.001961833,0.008377644,0.2777443,0.008407819,0.62380755,0.005311518,0.00019774331],"about_ca_topic_score_codex":0.0012376303,"about_ca_topic_score_gemma":0.001108028,"teacher_disagreement_score":0.019001858,"about_ca_system_score_codex":0.0012191422,"about_ca_system_score_gemma":0.0012222164,"threshold_uncertainty_score":0.1004926},"labels":[],"label_agreement":null},{"id":"W2072723323","doi":"10.1016/j.eswa.2014.03.042","title":"An interval weighed fuzzy c-means clustering by genetically guided alternating optimization","year":2014,"lang":"en","type":"article","venue":"Expert Systems with Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Fuzzy clustering; Weighting; Mathematics; Correlation clustering; Data mining; CURE data clustering algorithm; Fuzzy logic; FLAME clustering; k-medians clustering; Heuristic; Interval (graph theory); Partition (number theory); Computer science; Pattern recognition (psychology); Mathematical optimization; Algorithm; Artificial intelligence; Combinatorics","score_opus":0.016637616564232523,"score_gpt":0.30557719769351444,"score_spread":0.2889395811292819,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2072723323","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014705873,0.00009363945,0.9822489,0.000080953374,0.000062472274,0.000056183322,0.00002283103,0.00025140482,0.0024777085],"genre_scores_gemma":[0.20804134,0.00009135456,0.78806216,0.000090670874,0.00003212114,0.00021429021,0.000101261656,0.000120333796,0.003246389],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993337,0.00014466853,0.00003275793,0.00018043739,0.00025337774,0.000055091943],"domain_scores_gemma":[0.9994267,0.0001552164,0.000045925743,0.000055754354,0.00028098957,0.000035510802],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001013126,0.00092778844,0.0010648546,0.0009983977,0.0010730671,0.0010061868,0.0016436826,0.0016877948,0.0020077205],"category_scores_gemma":[0.0022068657,0.0005193561,0.0010312457,0.0010717324,0.00069905765,0.0008557621,0.0013155369,0.0009394461,0.00040813492],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013701343,0.000081212995,0.0005653915,0.00013235846,0.000087293505,0.0000793255,0.00015482688,0.8128554,0.012518358,0.01736503,0.0026382718,0.15338542],"study_design_scores_gemma":[0.000008080578,0.000019610641,0.00006722358,0.0000046889963,0.000009910033,0.000018024299,0.000008387476,0.996516,0.0011509574,0.0017339538,0.00045410226,0.0000090946],"about_ca_topic_score_codex":0.0075760903,"about_ca_topic_score_gemma":0.0059184595,"teacher_disagreement_score":0.0075760903,"about_ca_system_score_codex":0.0011176937,"about_ca_system_score_gemma":0.0018719274,"threshold_uncertainty_score":0.0150639415},"labels":[],"label_agreement":null},{"id":"W2072732320","doi":"10.1007/s10618-005-0019-1","title":"Data Clustering with Partial Supervision","year":2006,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":75,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence; Classifier (UML); Pattern recognition (psychology); Fuzzy clustering; Machine learning; Support vector machine; Data mining","score_opus":0.07186957610495115,"score_gpt":0.33223066228592213,"score_spread":0.260361086180971,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2072732320","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00929492,0.00024162678,0.9879447,0.00025821116,0.00003962943,0.00007452391,0.00026017454,0.0010523914,0.0008337929],"genre_scores_gemma":[0.3790835,0.00038088727,0.61392075,0.00018762748,0.00021651624,0.00037325994,0.002162853,0.00025771192,0.0034168917],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9943514,0.0021377027,0.00041709328,0.0013617767,0.0015583296,0.0001736776],"domain_scores_gemma":[0.9855563,0.0041386504,0.00079656957,0.007122415,0.002040097,0.00034593552],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045554065,0.0008085172,0.0021845875,0.001374606,0.0013022163,0.0020903272,0.0026164216,0.0012032968,0.0028480361],"category_scores_gemma":[0.022083798,0.0011221461,0.001975679,0.0021141586,0.0017512761,0.004056206,0.003653304,0.001552107,0.0013926757],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009985493,0.0002153772,0.008193932,0.00078276347,0.00061026827,0.00030593295,0.0005634647,0.2593059,0.007974184,0.06632493,0.012525273,0.64219946],"study_design_scores_gemma":[0.000030296835,0.000077334764,0.0007943041,0.000021960823,0.000055325727,0.00015783259,0.000039333376,0.93922615,0.0036000486,0.053064764,0.0029143463,0.000018352805],"about_ca_topic_score_codex":0.0024099296,"about_ca_topic_score_gemma":0.0039341343,"teacher_disagreement_score":0.0045554065,"about_ca_system_score_codex":0.00080778525,"about_ca_system_score_gemma":0.0026417396,"threshold_uncertainty_score":0.024091542},"labels":[],"label_agreement":null},{"id":"W2074553505","doi":"10.1080/02664760903186049","title":"Clustering probability distributions","year":2010,"lang":"en","type":"article","venue":"Journal of Applied Statistics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":57,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Moncton","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Joint probability distribution; Probability distribution; Computer science; Mathematics; Probability density function; Algorithm; Statistical physics; Data mining; Statistics; Physics","score_opus":0.01810154605205502,"score_gpt":0.29800518970734263,"score_spread":0.27990364365528764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2074553505","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011275623,0.0006525295,0.9811172,0.0003869126,0.000053776723,0.00009080359,0.00030979054,0.0002976435,0.0058156042],"genre_scores_gemma":[0.51999825,0.0037333206,0.45723552,0.00048563775,0.00050500676,0.00075978297,0.0020843048,0.00059716357,0.014601031],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9957515,0.0012031832,0.00022689912,0.0011985508,0.0012680135,0.0003519065],"domain_scores_gemma":[0.9895739,0.0059603993,0.00073041464,0.0013389793,0.0020343654,0.00036199862],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050532618,0.0012623522,0.0016423349,0.0063486933,0.0018775593,0.0040158927,0.0032498566,0.0026578773,0.005400025],"category_scores_gemma":[0.025323853,0.00075236347,0.0014994522,0.0040468317,0.0026531566,0.005260281,0.0025270106,0.002211621,0.0025453374],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008072009,0.000042985484,0.0022632454,0.00022653324,0.00008972787,0.00019994462,0.00047309767,0.1683348,0.002177608,0.7651075,0.0043791244,0.056624707],"study_design_scores_gemma":[0.000016467591,0.00004699074,0.0013215536,0.000118815,0.000044031123,0.00045083094,0.00024024224,0.48586956,0.0023223392,0.49809086,0.011398397,0.00007998671],"about_ca_topic_score_codex":0.0013614531,"about_ca_topic_score_gemma":0.001032174,"teacher_disagreement_score":0.0063486933,"about_ca_system_score_codex":0.0024933787,"about_ca_system_score_gemma":0.000962411,"threshold_uncertainty_score":0.026724577},"labels":[],"label_agreement":null},{"id":"W2076089275","doi":"10.1007/s00357-006-0018-y","title":"Generation of Random Clusters with Specified Degree of Separation","year":2006,"lang":"en","type":"article","venue":"Journal of Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":122,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Mathematics; Cluster (spacecraft); Constraint (computer-aided design); Dimension (graph theory); Degree (music); Outlier; Covariance; Set (abstract data type); Separation (statistics); Combinatorics; Pattern recognition (psychology); Algorithm; Artificial intelligence; Computer science; Statistics","score_opus":0.12329393467636098,"score_gpt":0.3420131434182264,"score_spread":0.21871920874186546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2076089275","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25491083,0.00016717575,0.7369395,0.00048685123,0.0002448337,0.00096722355,0.0003602703,0.0014617285,0.004461527],"genre_scores_gemma":[0.7322626,0.00007461704,0.26218885,0.00018497738,0.000032611235,0.0008232982,0.0007136718,0.00032538234,0.0033939455],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983619,0.0005692735,0.00008852066,0.00037554838,0.00038741098,0.00021733517],"domain_scores_gemma":[0.99005425,0.0042649936,0.00036693626,0.0023809546,0.0025560132,0.0003768146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031838906,0.0006837457,0.0009940829,0.001431462,0.0010380282,0.001154957,0.0021674703,0.0015860596,0.004110915],"category_scores_gemma":[0.017296238,0.00075149356,0.0009582366,0.0011115554,0.0009323521,0.0012609919,0.0019762772,0.0011740632,0.0010585266],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033267075,0.00078740384,0.007430605,0.00049275527,0.00025259957,0.00067507575,0.001339301,0.66712415,0.05402616,0.06855782,0.012843171,0.18314426],"study_design_scores_gemma":[0.00030447345,0.00015997046,0.00064852234,0.000020813393,0.00004629567,0.0001428325,0.00012195657,0.96113783,0.019804858,0.015705772,0.0018614016,0.000045333174],"about_ca_topic_score_codex":0.00099074,"about_ca_topic_score_gemma":0.0013030823,"teacher_disagreement_score":0.004110915,"about_ca_system_score_codex":0.0011939825,"about_ca_system_score_gemma":0.0010070158,"threshold_uncertainty_score":0.016838193},"labels":[],"label_agreement":null},{"id":"W2077025240","doi":"10.1142/s0218488508005133","title":"FUZZY CLUSTERING OF FEATURE VECTORS WITH SOME ORDINAL VALUED ATTRIBUTES USING GRADIENT DESCENT FOR LEARNING","year":2008,"lang":"en","type":"article","venue":"International Journal of Uncertainty Fuzziness and Knowledge-Based Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Mathematics; Ordinal data; Ordinal regression; Ordinal Scale; Cluster analysis; Gradient descent; Pattern recognition (psychology); Fuzzy clustering; Artificial intelligence; Feature (linguistics); Ordinal optimization; Fuzzy set; Data mining; Fuzzy logic; Statistics; Computer science; Artificial neural network","score_opus":0.04545949455616138,"score_gpt":0.30771333468718276,"score_spread":0.2622538401310214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077025240","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0112890415,0.00011939015,0.9879479,0.00010019297,0.000013506625,0.00003887599,0.000015958618,0.000130679,0.00034440332],"genre_scores_gemma":[0.25233576,0.00014640833,0.7455036,0.00007225157,0.000024754543,0.0001995725,0.00013669056,0.000057137608,0.0015238409],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99904305,0.0004171509,0.00005787754,0.00015882593,0.00025981446,0.00006315761],"domain_scores_gemma":[0.9987373,0.0005861719,0.00012171802,0.00010508562,0.00040831268,0.000041471874],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024481635,0.00066059857,0.0012303283,0.0012438524,0.0006564842,0.0010080276,0.0011037751,0.0010718937,0.00074880326],"category_scores_gemma":[0.005606994,0.00046414777,0.0008362393,0.001192376,0.000904729,0.0011007881,0.00081585173,0.0010694003,0.0003202369],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009726473,0.00007839313,0.0011093082,0.00011245094,0.00009244177,0.000052314594,0.00017417871,0.85368884,0.0032231717,0.016204298,0.0013264177,0.123840876],"study_design_scores_gemma":[0.0000034415295,0.000013325171,0.000112428395,0.0000053489216,0.0000024791243,0.0000063939683,0.000006443122,0.995865,0.0003512622,0.0034617954,0.00016700356,0.000005078593],"about_ca_topic_score_codex":0.0071864086,"about_ca_topic_score_gemma":0.0056682955,"teacher_disagreement_score":0.0071864086,"about_ca_system_score_codex":0.001614906,"about_ca_system_score_gemma":0.0013238406,"threshold_uncertainty_score":0.014289141},"labels":[],"label_agreement":null},{"id":"W2079040080","doi":"10.1007/s10618-013-0311-4","title":"A framework for semi-supervised and unsupervised optimal extraction of clusters from hierarchies","year":2013,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":83,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Tree (set theory); Cluster (spacecraft); Variety (cybernetics); Data mining; Artificial intelligence; Extraction (chemistry); Machine learning; Mathematics","score_opus":0.06040773208466434,"score_gpt":0.3443362713711306,"score_spread":0.2839285392864663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2079040080","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003578806,0.000065082546,0.999009,0.00004375724,0.000009231741,0.000023238008,0.000052601205,0.00028746756,0.00015162032],"genre_scores_gemma":[0.01760601,0.00011916919,0.9808699,0.000058420785,0.00004026802,0.00013292319,0.00038353426,0.00014142206,0.00064836163],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9961482,0.0011623132,0.00030949723,0.0008732014,0.0012579786,0.00024887783],"domain_scores_gemma":[0.9951792,0.0017755142,0.0003824513,0.001027828,0.0013939227,0.00024102046],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047129043,0.0011738804,0.0021359995,0.0034727857,0.0019374635,0.0037106322,0.0047548567,0.0019750022,0.002734884],"category_scores_gemma":[0.011648168,0.001068743,0.0025070822,0.0040500406,0.002121405,0.0038019721,0.0046154764,0.0029571534,0.0018301978],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021089043,0.00020919877,0.0012902771,0.00051547284,0.00028583367,0.00025367865,0.0008870711,0.15777513,0.0106559545,0.27801174,0.01594564,0.5339591],"study_design_scores_gemma":[0.000033424214,0.00004685997,0.0003688262,0.00006355796,0.000050944753,0.00015926121,0.000119053155,0.7681822,0.0027566475,0.21802625,0.010142959,0.000050032293],"about_ca_topic_score_codex":0.008027991,"about_ca_topic_score_gemma":0.015189968,"teacher_disagreement_score":0.008027991,"about_ca_system_score_codex":0.0016600689,"about_ca_system_score_gemma":0.0038711927,"threshold_uncertainty_score":0.024924517},"labels":[],"label_agreement":null},{"id":"W2079184215","doi":"10.5539/cis.v3n3p23","title":"An Improved Clustering Algorithm Based on Density Distribution Function","year":2010,"lang":"en","type":"article","venue":"Computer and Information Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Computer science; Point (geometry); Algorithm; Scale (ratio); Distribution (mathematics); Core (optical fiber); Function (biology); Pattern recognition (psychology); Mathematics; Artificial intelligence; Physics; Geometry; Mathematical analysis","score_opus":0.008231189996876725,"score_gpt":0.2644429853575031,"score_spread":0.2562117953606264,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2079184215","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003751332,0.0004144528,0.99387854,0.0001097542,0.00008587473,0.00005365024,0.0000557562,0.00062733836,0.0010233354],"genre_scores_gemma":[0.08341564,0.0009196794,0.907783,0.00016688084,0.00018919204,0.00030172805,0.0006307273,0.00023727283,0.0063559967],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997529,0.00034672508,0.00014159392,0.0006171665,0.0011994245,0.00016617657],"domain_scores_gemma":[0.9983615,0.00028819693,0.00009647829,0.00015346013,0.001052206,0.000048096907],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015675253,0.0009700195,0.0019908908,0.0034044045,0.0013561435,0.001594918,0.0026476951,0.0017671543,0.0023663018],"category_scores_gemma":[0.0038689866,0.000655757,0.0015470329,0.0043600323,0.0007172187,0.0028471276,0.0013934439,0.0013829092,0.001958762],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017243011,0.000104678445,0.0022757768,0.00033835552,0.00019693,0.00018758085,0.00040683942,0.25447425,0.01660447,0.03440346,0.011929201,0.678906],"study_design_scores_gemma":[0.00002670591,0.00003221194,0.00086177274,0.000017222974,0.00003171429,0.00024674408,0.000041802465,0.97736496,0.00377073,0.008211785,0.009335896,0.00005856244],"about_ca_topic_score_codex":0.013772424,"about_ca_topic_score_gemma":0.005851522,"teacher_disagreement_score":0.013772424,"about_ca_system_score_codex":0.0016133031,"about_ca_system_score_gemma":0.0020310993,"threshold_uncertainty_score":0.02738452},"labels":[],"label_agreement":null},{"id":"W2080476941","doi":"10.1016/j.patcog.2007.11.011","title":"A convergence theorem for the fuzzy subspace clustering (FSC) algorithm","year":2007,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":100,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Subsequence; Convergence (economics); Cluster analysis; Subspace topology; Sequence (biology); Algorithm; Mathematics; Fuzzy logic; Set (abstract data type); Computer science; Mathematical optimization; Artificial intelligence","score_opus":0.044762448138194626,"score_gpt":0.3136870178531478,"score_spread":0.26892456971495315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2080476941","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023173012,0.00029273643,0.99321425,0.00026273378,0.00011279766,0.00004582816,0.000071661845,0.00012482931,0.0035578492],"genre_scores_gemma":[0.13869578,0.0013066345,0.8432359,0.00054226926,0.0004774946,0.000546206,0.00056084845,0.0004845273,0.014150323],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99750936,0.0006884316,0.00010386015,0.00033744736,0.0012085388,0.0001524255],"domain_scores_gemma":[0.9928807,0.0032049215,0.0001782057,0.0004965369,0.0030268715,0.00021271774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054078824,0.0011054045,0.0014221302,0.0023312,0.0014982888,0.0022184532,0.002264584,0.0020335112,0.0060141734],"category_scores_gemma":[0.020150553,0.0004568423,0.0011313682,0.0021848017,0.0025099386,0.0026908147,0.0029070226,0.0032454114,0.0025246488],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017295631,0.00006516273,0.00082656415,0.00030861807,0.000100106125,0.000092832706,0.00035051105,0.18215835,0.0067471247,0.575193,0.013984748,0.21999997],"study_design_scores_gemma":[0.000022792334,0.000060428334,0.00031979225,0.000050510083,0.000019122492,0.00018027841,0.00005704252,0.83825266,0.0025477551,0.14919196,0.009259198,0.000038434548],"about_ca_topic_score_codex":0.005302574,"about_ca_topic_score_gemma":0.0031306602,"teacher_disagreement_score":0.0060141734,"about_ca_system_score_codex":0.001375741,"about_ca_system_score_gemma":0.0021803821,"threshold_uncertainty_score":0.028599977},"labels":[],"label_agreement":null},{"id":"W2082539276","doi":"10.1016/j.ins.2012.02.042","title":"Feature selection using structural similarity","year":2012,"lang":"en","type":"article","venue":"Information Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cardinality (data modeling); Feature selection; Pattern recognition (psychology); Similarity (geometry); Data mining; Cluster analysis; Partition (number theory); Curse of dimensionality; Feature (linguistics); Computer science; Mathematics; Artificial intelligence; Selection (genetic algorithm); Combinatorics","score_opus":0.057503571108182994,"score_gpt":0.37055909314857566,"score_spread":0.31305552204039266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2082539276","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09471186,0.00056140486,0.89964783,0.00025319803,0.00017853381,0.00023458383,0.00038602087,0.0015890527,0.0024375583],"genre_scores_gemma":[0.7512818,0.00031609618,0.24129382,0.00015594302,0.00026014334,0.00027402648,0.0023346534,0.00021780953,0.0038657254],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99908686,0.00016282825,0.000090325055,0.00020499341,0.00035616144,0.00009893538],"domain_scores_gemma":[0.9985709,0.00042839325,0.00008510331,0.0002160254,0.0006366171,0.00006293421],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009351105,0.00075858936,0.0016960972,0.0034028583,0.000890763,0.0012485957,0.0009021496,0.0008396555,0.0026912712],"category_scores_gemma":[0.003602042,0.00026194495,0.0015525756,0.003065841,0.0003431665,0.001064565,0.0008574332,0.00064273603,0.0014090046],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00089653244,0.0004475108,0.005842826,0.00019557048,0.00034063892,0.000203449,0.00007378526,0.026698107,0.03955876,0.0045084152,0.009076702,0.9121576],"study_design_scores_gemma":[0.00016285009,0.000634095,0.007883973,0.000028175065,0.00036438458,0.00051541993,0.00012305517,0.94964087,0.024722476,0.010813681,0.005056765,0.00005423733],"about_ca_topic_score_codex":0.0013383114,"about_ca_topic_score_gemma":0.0017544917,"teacher_disagreement_score":0.0034028583,"about_ca_system_score_codex":0.00032175574,"about_ca_system_score_gemma":0.00093019824,"threshold_uncertainty_score":0.009003162},"labels":[],"label_agreement":null},{"id":"W2082738974","doi":"10.1109/cicare.2014.7007849","title":"A novel mixed values k-prototypes algorithm with application to health care databases mining","year":2014,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval","funders":"Institute of Genetics","keywords":"Categorical variable; Cluster analysis; Data mining; Computer science; Variable (mathematics); Field (mathematics); Representation (politics); Algorithm; Database; Artificial intelligence; Machine learning; Mathematics","score_opus":0.02655961038866402,"score_gpt":0.3337965133803607,"score_spread":0.3072369029916967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2082738974","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008171251,0.0005843401,0.9883808,0.00027081513,0.00013440334,0.00017882588,0.00020951853,0.001201316,0.0008687905],"genre_scores_gemma":[0.056409486,0.00020726446,0.94084936,0.00010298243,0.00007469822,0.00026026185,0.0004967227,0.00014856973,0.0014507368],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99612904,0.001217802,0.00034212894,0.00078107114,0.0013434268,0.0001865865],"domain_scores_gemma":[0.9961863,0.0012909157,0.0002475494,0.00042965054,0.0016679153,0.00017759886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030983917,0.0011547277,0.0019534894,0.0034671745,0.001805957,0.0027114295,0.004601355,0.0021953606,0.0025316605],"category_scores_gemma":[0.012653417,0.0009291534,0.0013354003,0.006429686,0.00080235576,0.0029631217,0.0021766392,0.0014392396,0.0014934253],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051297125,0.00024426496,0.0035293875,0.00045728448,0.000371495,0.0002963506,0.00053722586,0.087263174,0.0052832323,0.018512215,0.013554921,0.86943746],"study_design_scores_gemma":[0.00008357122,0.00013937686,0.0009035794,0.00004941963,0.000054302836,0.00043282087,0.00022588797,0.9624248,0.0043793367,0.020072296,0.011168805,0.000065790955],"about_ca_topic_score_codex":0.009969314,"about_ca_topic_score_gemma":0.010634581,"teacher_disagreement_score":0.009969314,"about_ca_system_score_codex":0.001129536,"about_ca_system_score_gemma":0.002181318,"threshold_uncertainty_score":0.019822598},"labels":[],"label_agreement":null},{"id":"W2086478515","doi":"10.1126/science.1151268","title":"Response to Comment on \"Clustering by Passing Messages Between Data Points\"","year":2008,"lang":"en","type":"article","venue":"Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Generalization; Heuristic; Cluster analysis; Vertex (graph theory); Affinity propagation; Probabilistic logic; Substitution (logic); Computer science; Algorithm; Combinatorics; Graph; Chemistry; Mathematics; Theoretical computer science; Artificial intelligence; Correlation clustering; Mathematical analysis","score_opus":0.10641529095600079,"score_gpt":0.37877880338952974,"score_spread":0.27236351243352896,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2086478515","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00016990908,0.00074619567,0.00060254085,0.9255188,0.07112769,0.00003560802,0.00030630312,0.00019035068,0.0013026885],"genre_scores_gemma":[0.0010310602,0.00046707416,0.00037567274,0.9608483,0.032451775,0.000081845916,0.00007552277,0.00007000906,0.0045987912],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9939535,0.0013638898,0.0008157606,0.0009819712,0.0023317898,0.0005530901],"domain_scores_gemma":[0.96165776,0.019554345,0.0020659214,0.0012198229,0.013419147,0.0020829858],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077980557,0.001833999,0.0017840178,0.0015854383,0.0038466274,0.0036710473,0.004372071,0.04037102,0.02125883],"category_scores_gemma":[0.062163666,0.0011288906,0.0021231046,0.0017315428,0.004018871,0.003845166,0.0028741292,0.044747565,0.024884086],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040113595,0.000011890848,0.00012904286,0.000057199497,0.00001459314,0.000074838645,0.000051910996,0.00006188552,0.00009115791,0.0006976354,0.9965701,0.0021996102],"study_design_scores_gemma":[0.0001636553,0.00006706645,0.002424501,0.00041618414,0.000050223938,0.0002977011,0.00038089062,0.00062631717,0.0006001433,0.0036735178,0.99119896,0.00010073573],"about_ca_topic_score_codex":0.021460911,"about_ca_topic_score_gemma":0.017750021,"teacher_disagreement_score":0.04037102,"about_ca_system_score_codex":0.006216642,"about_ca_system_score_gemma":0.004589908,"threshold_uncertainty_score":0.07111788},"labels":[],"label_agreement":null},{"id":"W2086582257","doi":"10.1007/bf02962204","title":"Sequential combination methods for data clustering analysis","year":2002,"lang":"en","type":"article","venue":"Journal of Computer Science and Technology","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Computer science; CURE data clustering algorithm; Correlation clustering; Canopy clustering algorithm; Data stream clustering; Constrained clustering; Fuzzy clustering; Data mining; Clustering high-dimensional data; Algorithm; Artificial intelligence","score_opus":0.0811634130080792,"score_gpt":0.41144146351408606,"score_spread":0.3302780505060069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2086582257","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00166235,0.00043067161,0.996863,0.00004597038,0.0000679143,0.000071866554,0.00008836374,0.00053101435,0.00023881749],"genre_scores_gemma":[0.034240816,0.00052960624,0.9620907,0.00006990105,0.00016398191,0.00045155297,0.0006437394,0.00028443264,0.0015253546],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99181527,0.0036976216,0.00061747007,0.0013181447,0.002312546,0.0002389619],"domain_scores_gemma":[0.9866599,0.007976015,0.0007139035,0.0022647663,0.002103712,0.00028159475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075704027,0.0020984455,0.0037904473,0.0040859464,0.0016991494,0.0018000749,0.0029701272,0.0013897364,0.004255399],"category_scores_gemma":[0.01536579,0.0019135763,0.003444755,0.008516218,0.0011204211,0.0027245407,0.0022007956,0.0031119108,0.0021735374],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007070606,0.00022725771,0.0020474733,0.0008253524,0.0009595724,0.00012422638,0.000256841,0.11414776,0.008629548,0.020177789,0.0076693995,0.84422773],"study_design_scores_gemma":[0.00009521714,0.00021877218,0.001315508,0.000046490575,0.00028478602,0.00025185593,0.00006157209,0.9206611,0.0045632496,0.0636488,0.008785234,0.00006754018],"about_ca_topic_score_codex":0.0039648567,"about_ca_topic_score_gemma":0.005541536,"teacher_disagreement_score":0.0075704027,"about_ca_system_score_codex":0.0010459227,"about_ca_system_score_gemma":0.0019869003,"threshold_uncertainty_score":0.04003656},"labels":[],"label_agreement":null},{"id":"W2086829791","doi":"10.1007/s10618-011-0221-2","title":"DHCC: Divisive hierarchical clustering of categorical data","year":2011,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":77,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Data mining; Initialization; Similarity (geometry); Linear subspace; Single-linkage clustering; Hierarchical clustering; Artificial intelligence; Pattern recognition (psychology); CURE data clustering algorithm; Correlation clustering; Machine learning; Mathematics","score_opus":0.19241456294856188,"score_gpt":0.35658341041610714,"score_spread":0.16416884746754526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2086829791","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033370464,0.00033460552,0.9810245,0.00016454906,0.00016145674,0.00042559084,0.0024336348,0.011159172,0.0009594607],"genre_scores_gemma":[0.029774936,0.00020604397,0.961478,0.0001240128,0.000074891825,0.0005166616,0.0052012703,0.00087264186,0.0017515081],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9966754,0.0011234827,0.00023696703,0.0005648898,0.0011925034,0.00020682611],"domain_scores_gemma":[0.9951391,0.0014741125,0.0002268712,0.0015837203,0.0013670189,0.00020921614],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031861435,0.0013503027,0.0016416556,0.0051628626,0.0016267915,0.0026188253,0.0034556084,0.0013924185,0.0061198743],"category_scores_gemma":[0.014504861,0.0008620228,0.0020251097,0.007430254,0.0006719409,0.0016278939,0.0027894499,0.0022480043,0.0042519635],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042408897,0.00024775352,0.0036280137,0.00076729053,0.00041261627,0.00016022353,0.000566676,0.047848776,0.0059847413,0.02416062,0.06596284,0.84983635],"study_design_scores_gemma":[0.00016969623,0.00016590423,0.0036284556,0.00012955096,0.00019260119,0.0003882017,0.00029826586,0.839159,0.014006497,0.07377815,0.06792869,0.00015500785],"about_ca_topic_score_codex":0.012267906,"about_ca_topic_score_gemma":0.017289916,"teacher_disagreement_score":0.012267906,"about_ca_system_score_codex":0.0012010592,"about_ca_system_score_gemma":0.003998428,"threshold_uncertainty_score":0.024393022},"labels":[],"label_agreement":null},{"id":"W2088308919","doi":"10.1109/icdmw.2011.162","title":"TeamExp: Top-k Team Formation in Social Networks","year":2011,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Set (abstract data type); Computer science; Social network (sociolinguistics); Communication skills; Knowledge management; Social media; World Wide Web; Medical education","score_opus":0.04316863286285745,"score_gpt":0.2991666233186445,"score_spread":0.2559979904557871,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2088308919","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022962129,0.0003032078,0.8743365,0.00051988574,0.00015381665,0.00060326676,0.006862257,0.089575596,0.004683418],"genre_scores_gemma":[0.09646497,0.00014544431,0.8890072,0.00013489426,0.00004187159,0.00042474954,0.0073536225,0.0014381971,0.0049890513],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986084,0.00037799237,0.000114317954,0.00048094397,0.00032560222,0.00009281087],"domain_scores_gemma":[0.9974273,0.00094755477,0.0001677819,0.00095235667,0.00032620708,0.00017885142],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021075779,0.0013470427,0.001053348,0.0016123895,0.001956311,0.0020738987,0.0026135996,0.0014773154,0.011870955],"category_scores_gemma":[0.0069364305,0.0009804907,0.0010907608,0.0016405516,0.0006056793,0.0040537296,0.00380967,0.001462644,0.0066913264],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018047141,0.00052364037,0.008368152,0.0010786799,0.00031175127,0.0003900925,0.0011322892,0.07435524,0.021322004,0.027192727,0.15258545,0.7109353],"study_design_scores_gemma":[0.000245293,0.00015000839,0.0013676088,0.00006910729,0.000058718757,0.00027086856,0.00034130568,0.8842179,0.02074121,0.04207155,0.05040041,0.000066004286],"about_ca_topic_score_codex":0.0034136726,"about_ca_topic_score_gemma":0.0077070193,"teacher_disagreement_score":0.011870955,"about_ca_system_score_codex":0.0010433159,"about_ca_system_score_gemma":0.0011564839,"threshold_uncertainty_score":0.03971231},"labels":[],"label_agreement":null},{"id":"W2089055054","doi":"10.1145/1401890.1401956","title":"Finding non-redundant, statistically significant regions in high dimensional data","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":109,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Disjoint sets; Computer science; Cluster (spacecraft); Outlier; Data mining; Single-linkage clustering; Clustering high-dimensional data; Subspace topology; Complete-linkage clustering; Data point; Correlation clustering; Determining the number of clusters in a data set; CURE data clustering algorithm; Pattern recognition (psychology); Artificial intelligence; Mathematics; Combinatorics","score_opus":0.10470996700155884,"score_gpt":0.3439923630461722,"score_spread":0.23928239604461338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2089055054","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13539061,0.003208513,0.85776657,0.00064101414,0.00009281497,0.00016679932,0.0005835518,0.0014384782,0.00071156886],"genre_scores_gemma":[0.4636392,0.0015603607,0.5313199,0.00018522929,0.00023060646,0.00024298744,0.0017985025,0.00022059685,0.00080258655],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.995773,0.000988354,0.0003646212,0.0009911594,0.001570388,0.0003124692],"domain_scores_gemma":[0.9897725,0.0052969605,0.0020315659,0.001378428,0.0012884128,0.00023207547],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004778348,0.0014314314,0.0028883792,0.0068135015,0.00144491,0.002659874,0.0022120327,0.0022366703,0.00074170483],"category_scores_gemma":[0.014123462,0.0009785583,0.001629969,0.006830544,0.0024439308,0.0023585227,0.0021620535,0.0013128539,0.000916635],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012696218,0.0007136622,0.051303793,0.002320428,0.0010983722,0.0032360817,0.0028504147,0.14683278,0.09589059,0.02324325,0.0102096405,0.66103137],"study_design_scores_gemma":[0.00012766536,0.000659034,0.04528477,0.00030321884,0.00073287834,0.003986267,0.003171764,0.7972945,0.055572692,0.08063266,0.011955459,0.0002791392],"about_ca_topic_score_codex":0.0018363013,"about_ca_topic_score_gemma":0.0022341637,"teacher_disagreement_score":0.0068135015,"about_ca_system_score_codex":0.00061746186,"about_ca_system_score_gemma":0.0024787749,"threshold_uncertainty_score":0.025270581},"labels":[],"label_agreement":null},{"id":"W2093156537","doi":"10.1007/s10044-007-0088-4","title":"Hierarchical clustering of subpopulations with a dissimilarity based on the likelihood ratio statistic: application to clustering massive data sets","year":2007,"lang":"en","type":"article","venue":"Pattern Analysis and Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Cluster analysis; Mathematics; Statistic; Hierarchical clustering; Multinomial distribution; Poisson distribution; Multivariate normal distribution; Statistics; Data mining; Computer science; Multivariate statistics","score_opus":0.03091776160389113,"score_gpt":0.3354578783523037,"score_spread":0.3045401167484125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2093156537","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027019199,0.0001413865,0.9719282,0.00010407336,0.000016072372,0.000072704126,0.000068282534,0.00037230173,0.00027786475],"genre_scores_gemma":[0.27715188,0.00011362157,0.7213452,0.000045223434,0.000053191132,0.00018780703,0.00032009737,0.00017928313,0.0006037787],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99744284,0.0012218377,0.00015848098,0.0004536676,0.00062783645,0.0000952978],"domain_scores_gemma":[0.9923184,0.0046004453,0.0008729516,0.0009341915,0.0009749034,0.00029911706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005676434,0.0006505508,0.0016673958,0.0037589332,0.0013631787,0.0017893915,0.0019353016,0.0013227252,0.0011218596],"category_scores_gemma":[0.02383688,0.0004696299,0.0015378109,0.0033513836,0.0010985417,0.0016475376,0.0023915933,0.0013253739,0.00042618348],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008912798,0.00036805603,0.017356526,0.00059529766,0.0006520575,0.00043686066,0.0015675556,0.4425353,0.020740416,0.061276004,0.0041583152,0.44942227],"study_design_scores_gemma":[0.000052041458,0.00010000428,0.002413548,0.000013949739,0.000039254115,0.00014409822,0.00010465688,0.96526694,0.001994581,0.028903672,0.00092347,0.000043791504],"about_ca_topic_score_codex":0.002334434,"about_ca_topic_score_gemma":0.0027447818,"teacher_disagreement_score":0.005676434,"about_ca_system_score_codex":0.0010439095,"about_ca_system_score_gemma":0.0011448838,"threshold_uncertainty_score":0.030020237},"labels":[],"label_agreement":null},{"id":"W2095980495","doi":"10.1145/2740908.2742474","title":"Short-Text Clustering using Statistical Semantics","year":2015,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Term (time); Similarity (geometry); Semantics (computer science); Representation (politics); Selection (genetic algorithm); Measure (data warehouse); Computation; Space (punctuation); Data mining; Artificial intelligence; Matrix (chemical analysis); Document clustering; Pattern recognition (psychology); Algorithm","score_opus":0.13068816452266865,"score_gpt":0.3815659194037594,"score_spread":0.25087775488109076,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2095980495","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011027045,0.00029631527,0.98724085,0.00006363955,0.0000292376,0.00005791674,0.00010547034,0.0006648969,0.0005147177],"genre_scores_gemma":[0.18993823,0.0007220986,0.8042816,0.00009787614,0.00015488522,0.0002696729,0.0015712997,0.00023900217,0.0027252873],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990458,0.0002266548,0.000093516246,0.00020501587,0.00037752572,0.000051448686],"domain_scores_gemma":[0.99846953,0.000512734,0.00024601386,0.00022965482,0.00048816222,0.000053842938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008108693,0.00084974547,0.0009781558,0.004775308,0.000690201,0.0012580113,0.0010973137,0.00089715124,0.0016245338],"category_scores_gemma":[0.0040362542,0.0003353284,0.0011733337,0.003547653,0.00067806477,0.0024397867,0.000976515,0.0006541029,0.0013776873],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035231165,0.00017794619,0.0029283983,0.00054333993,0.00029374933,0.00028963896,0.0004079729,0.27464926,0.058000356,0.060136832,0.004782367,0.5974378],"study_design_scores_gemma":[0.00001994226,0.000077398894,0.001087886,0.000026914362,0.000034843986,0.00018310931,0.00008600284,0.95146745,0.0088072065,0.033869047,0.004307625,0.00003264505],"about_ca_topic_score_codex":0.0030294447,"about_ca_topic_score_gemma":0.0043192524,"teacher_disagreement_score":0.004775308,"about_ca_system_score_codex":0.0007758929,"about_ca_system_score_gemma":0.0011564774,"threshold_uncertainty_score":0.006023586},"labels":[],"label_agreement":null},{"id":"W2096879261","doi":"","title":"Measures of Clustering Quality: A Working Set of Axioms for Clustering","year":2008,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":141,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Axiom; Impossibility; Constrained clustering; Computer science; Correlation clustering; Theoretical computer science; Fuzzy clustering; Data mining; Mathematics; Set (abstract data type); CURE data clustering algorithm; Artificial intelligence","score_opus":0.17760896120554284,"score_gpt":0.35983169362462686,"score_spread":0.18222273241908402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096879261","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037417412,0.0006511424,0.9909189,0.0016186079,0.00009125115,0.00010387314,0.00022944102,0.00019435023,0.002450815],"genre_scores_gemma":[0.0880386,0.0008861738,0.9074914,0.00085186097,0.0003902321,0.0007049525,0.00065775885,0.00013218295,0.00084684294],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9734018,0.0094330385,0.0037664887,0.0044299955,0.008156549,0.0008121577],"domain_scores_gemma":[0.9396952,0.030070657,0.0066092666,0.012233023,0.010094877,0.0012970302],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02813102,0.0019148435,0.002549851,0.008247268,0.0035668467,0.009385272,0.007253961,0.005496571,0.002826716],"category_scores_gemma":[0.0516832,0.0013603509,0.0036571678,0.008383105,0.017726589,0.018197214,0.005802618,0.00820783,0.0011118847],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002268387,0.000029090577,0.00081539684,0.0002726815,0.000054071355,0.000036680558,0.0003048441,0.007677815,0.00057946926,0.9724894,0.001541748,0.01617609],"study_design_scores_gemma":[0.000021223532,0.000064007974,0.00051911717,0.00013667163,0.000040516737,0.00015698154,0.0001379918,0.03699968,0.001149011,0.9529071,0.007808596,0.000059049406],"about_ca_topic_score_codex":0.0022437135,"about_ca_topic_score_gemma":0.0013234667,"teacher_disagreement_score":0.02813102,"about_ca_system_score_codex":0.004895412,"about_ca_system_score_gemma":0.0034507578,"threshold_uncertainty_score":0.14877284},"labels":[],"label_agreement":null},{"id":"W2097507414","doi":"10.1109/icdm.2006.151","title":"Speedup Clustering with Hierarchical Ranking","year":2006,"lang":"en","type":"article","venue":"Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Canada's Michael Smith Genome Sciences Centre","keywords":"Speedup; Cluster analysis; Computer science; Ranking (information retrieval); Pairwise comparison; Data mining; Hierarchical clustering; Algorithm; Artificial intelligence; Parallel computing","score_opus":0.009744494059777837,"score_gpt":0.2418840382192615,"score_spread":0.23213954415948365,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097507414","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029018374,0.00040886184,0.9580048,0.00039020102,0.00014271926,0.00015320742,0.00022296526,0.006338497,0.0053204084],"genre_scores_gemma":[0.21467173,0.00020755462,0.778137,0.0001490389,0.00008394255,0.00012826321,0.00078677526,0.0005466737,0.0052890605],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982052,0.00040209998,0.00008085638,0.00030014815,0.00086798746,0.00014377937],"domain_scores_gemma":[0.9967456,0.0010046704,0.0001689336,0.0014016847,0.0005843518,0.00009468349],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014197445,0.0011861242,0.001152997,0.0012158534,0.00081777305,0.001106244,0.0014118663,0.0010124868,0.0063827783],"category_scores_gemma":[0.006596499,0.0005145454,0.00082833203,0.002218371,0.000581513,0.0027139992,0.0017475846,0.0009553032,0.0040303026],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038536993,0.0001652041,0.0019671295,0.0003369715,0.0001223234,0.0001147672,0.00031445656,0.21451913,0.041032415,0.031060625,0.018747421,0.6912342],"study_design_scores_gemma":[0.00011618645,0.0001560638,0.0013181077,0.000016936407,0.00004831331,0.00020649242,0.0001590507,0.92248094,0.023110569,0.037284695,0.015059207,0.000043445278],"about_ca_topic_score_codex":0.0081898,"about_ca_topic_score_gemma":0.012648747,"teacher_disagreement_score":0.0081898,"about_ca_system_score_codex":0.0010496996,"about_ca_system_score_gemma":0.0015075948,"threshold_uncertainty_score":0.02135259},"labels":[],"label_agreement":null},{"id":"W2099837874","doi":"","title":"Human Cluster Evaluation and Formal Quality Measures: A Comparative Study","year":2012,"lang":"en","type":"article","venue":"eScholarship (California Digital Library)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Science Foundation","keywords":"Cluster analysis; Computer science; Data mining; Disjoint sets; Partition (number theory); Consistency (knowledge bases); Set (abstract data type); Contrast (vision); Fuzzy clustering; Machine learning; Artificial intelligence; Mathematics","score_opus":0.09701167708169886,"score_gpt":0.3631882371902808,"score_spread":0.2661765601085819,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099837874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9332653,0.0016478491,0.048088692,0.00053938193,0.00007820854,0.00021532952,0.00023655321,0.00023572177,0.015692936],"genre_scores_gemma":[0.99134177,0.00016537169,0.007944169,0.00004285827,0.00001834945,0.00005840493,0.000112029236,0.000026128722,0.00029089392],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9659104,0.019387,0.0020384875,0.0021391176,0.009968936,0.0005560012],"domain_scores_gemma":[0.74114096,0.19193149,0.01787586,0.01440578,0.031786665,0.0028591543],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03624295,0.00041643318,0.0005833763,0.0064250245,0.0008490547,0.0028616413,0.0011019884,0.00066407514,0.0021075488],"category_scores_gemma":[0.18428741,0.00017192264,0.00040733247,0.0044447533,0.0028586441,0.0030449242,0.0023139636,0.00052726327,0.00025318004],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016555509,0.0009544067,0.50384414,0.0014921018,0.0006920061,0.00026081753,0.038429346,0.008183569,0.004064854,0.02900077,0.006403729,0.4050187],"study_design_scores_gemma":[0.00022644184,0.003994707,0.77613026,0.0008739586,0.00029683154,0.0013985512,0.044941034,0.10529962,0.008194723,0.04113302,0.017127275,0.0003836193],"about_ca_topic_score_codex":0.0021814476,"about_ca_topic_score_gemma":0.0023163545,"teacher_disagreement_score":0.03624295,"about_ca_system_score_codex":0.0019775718,"about_ca_system_score_gemma":0.0009773066,"threshold_uncertainty_score":0.19167334},"labels":[],"label_agreement":null},{"id":"W2100405508","doi":"10.1109/nafips.2008.4531218","title":"Discovering structure in labeled data","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Computer science; Hierarchical clustering; Class (philosophy); Similarity (geometry); Data mining; Measure (data warehouse); Cluster (spacecraft); Artificial intelligence; Similarity measure; Data structure; Pattern recognition (psychology); Mathematics","score_opus":0.08347915006166517,"score_gpt":0.329163074094694,"score_spread":0.2456839240330288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100405508","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011382882,0.00026180487,0.9863521,0.00032993956,0.000045811394,0.00009498417,0.00034595773,0.00040987477,0.0007766549],"genre_scores_gemma":[0.1501999,0.0004092749,0.84530824,0.00026963936,0.00016419795,0.00027198571,0.002323846,0.000093879906,0.0009590607],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9935917,0.0023751564,0.00035365165,0.0014777959,0.001938098,0.00026343894],"domain_scores_gemma":[0.9815042,0.009555252,0.0021200355,0.003961768,0.0025438557,0.00031494178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004639487,0.00082329276,0.0014402041,0.0046424256,0.0017661595,0.0028308695,0.0022161452,0.0019293855,0.00096376677],"category_scores_gemma":[0.021936808,0.0008165502,0.0011927441,0.0041400366,0.0018560595,0.004341541,0.002735205,0.002113548,0.0006815513],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040369143,0.00045150326,0.019818751,0.0008440035,0.00041656196,0.0007919188,0.0021166669,0.17973894,0.01092538,0.15888329,0.013239448,0.6123699],"study_design_scores_gemma":[0.000037167472,0.00007862285,0.0013702676,0.000105216524,0.00006742918,0.00020270499,0.00031884524,0.77154535,0.0054207956,0.2129793,0.007827743,0.000046622674],"about_ca_topic_score_codex":0.0025276924,"about_ca_topic_score_gemma":0.0048738006,"teacher_disagreement_score":0.0046424256,"about_ca_system_score_codex":0.0014474454,"about_ca_system_score_gemma":0.0024119434,"threshold_uncertainty_score":0.024536252},"labels":[],"label_agreement":null},{"id":"W2103991913","doi":"10.1093/bioinformatics/btg030","title":"<i>K</i>-ary clustering with optimal leaf ordering for gene expression data","year":2003,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University","funders":"Leverhulme Trust; Florida State University; Natural Sciences and Engineering Research Council of Canada; Burroughs Wellcome Fund","keywords":"Cluster analysis; Hierarchical clustering; Computer science; Tree (set theory); Binary tree; Pairwise comparison; Hierarchical clustering of networks; Single-linkage clustering; Robustness (evolution); CURE data clustering algorithm; Data mining; Algorithm; Nearest-neighbor chain algorithm; Correlation clustering; Canopy clustering algorithm; Tree structure; Mathematics; Artificial intelligence; Gene; Biology; Combinatorics; Genetics","score_opus":0.055061850156336696,"score_gpt":0.306350724116407,"score_spread":0.2512888739600703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103991913","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062975967,0.0001145863,0.9838688,0.00017130915,0.000030390158,0.00011409299,0.00042088248,0.008386512,0.0005958102],"genre_scores_gemma":[0.024293406,0.000060835966,0.97310793,0.000059419115,0.0000144547985,0.00015367572,0.0011783591,0.00056860235,0.0005633194],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99793774,0.00044617016,0.00018854489,0.00050060335,0.0007726861,0.00015428054],"domain_scores_gemma":[0.99694866,0.0009310616,0.00032539913,0.0009291266,0.00068641175,0.00017933661],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021153945,0.0010722531,0.0012031817,0.0024068283,0.0018573427,0.0016160012,0.0027395776,0.0018598372,0.004141501],"category_scores_gemma":[0.0102513395,0.0008082603,0.0012197417,0.0041939137,0.0012350102,0.0019335304,0.0021857354,0.0019722062,0.0054497398],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006282751,0.000357411,0.004625145,0.0006794899,0.00012865187,0.0002718981,0.00086226827,0.18105958,0.04641232,0.021830961,0.036616053,0.7065279],"study_design_scores_gemma":[0.00008747509,0.00008242814,0.0013178393,0.000039893373,0.00002166478,0.00021629201,0.000089672656,0.9379001,0.019475104,0.030690582,0.010027176,0.000051761876],"about_ca_topic_score_codex":0.0061216233,"about_ca_topic_score_gemma":0.008840216,"teacher_disagreement_score":0.0061216233,"about_ca_system_score_codex":0.0015504106,"about_ca_system_score_gemma":0.0024545384,"threshold_uncertainty_score":0.013854742},"labels":[],"label_agreement":null},{"id":"W2105120842","doi":"10.1007/s10898-011-9691-4","title":"An efficient algorithm for maximal margin clustering","year":2011,"lang":"en","type":"article","venue":"Journal of Global Optimization","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Air Force Office of Scientific Research; National Institutes of Health; National Science Foundation","keywords":"Margin (machine learning); Cluster analysis; Mathematics; Curse of dimensionality; Dimensionality reduction; Convergence (economics); Algorithm; Extension (predicate logic); Computational complexity theory; Computer science; Artificial intelligence; Machine learning; Statistics","score_opus":0.027345270433135983,"score_gpt":0.3021882505185445,"score_spread":0.2748429800854085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2105120842","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023629826,0.00012960998,0.9952852,0.00010825426,0.00006160094,0.000066026994,0.00006196384,0.0008172509,0.0011072394],"genre_scores_gemma":[0.036422934,0.00009291076,0.9594899,0.00010625729,0.00006084312,0.00022609583,0.00035200754,0.00027278918,0.0029762567],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99823356,0.00047800734,0.00011970498,0.00039974693,0.00063130737,0.00013770616],"domain_scores_gemma":[0.99818134,0.000697353,0.00010829493,0.00039651946,0.0005040192,0.000112511],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017466473,0.0015517878,0.0020240692,0.0016999106,0.001994764,0.0018781448,0.0037572235,0.002556913,0.008198697],"category_scores_gemma":[0.0061856867,0.0011019506,0.0013555256,0.0022704438,0.0011520265,0.003003686,0.0043488415,0.0023836235,0.0044522732],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063969125,0.00019278162,0.0005029016,0.00026288227,0.0001397244,0.000102165104,0.00025842956,0.18327475,0.010293711,0.05210717,0.018815666,0.7334101],"study_design_scores_gemma":[0.000091899004,0.00007187677,0.00023114888,0.000026040234,0.000028114726,0.00012442487,0.00006576299,0.92909086,0.004551529,0.05877347,0.006909534,0.000035409448],"about_ca_topic_score_codex":0.0028441765,"about_ca_topic_score_gemma":0.0049598985,"teacher_disagreement_score":0.008198697,"about_ca_system_score_codex":0.0013241671,"about_ca_system_score_gemma":0.0024459043,"threshold_uncertainty_score":0.027427435},"labels":[],"label_agreement":null},{"id":"W2105790475","doi":"10.1109/icpr.2006.88","title":"A K-means-based Algorithm for Projective Clustering","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Computer science; Projective test; Algorithm; Artificial intelligence; Mathematics; Pure mathematics","score_opus":0.020212233679205004,"score_gpt":0.30347532922713905,"score_spread":0.28326309554793405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2105790475","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006179314,0.000101544676,0.9981092,0.000036079404,0.000038350656,0.000042488256,0.000028013395,0.0005525674,0.00047379712],"genre_scores_gemma":[0.012565319,0.00015189644,0.9854407,0.00005010628,0.00004110705,0.00014482647,0.00020650095,0.00017949996,0.0012201453],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973242,0.0005951147,0.00014444029,0.00068541063,0.001122756,0.00012795553],"domain_scores_gemma":[0.99874145,0.00037575688,0.000098385935,0.00022509665,0.00050868734,0.00005066334],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001822391,0.0014399018,0.0016167993,0.0023549418,0.0020496468,0.0018380072,0.0027785643,0.0016934397,0.0050516822],"category_scores_gemma":[0.004784545,0.0008889837,0.0013968386,0.0032657445,0.00126364,0.0025699819,0.0023205467,0.002441486,0.004543422],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018776204,0.00009260459,0.0008391918,0.00032215318,0.00021767405,0.00012504379,0.00028147013,0.13452885,0.015718544,0.04926476,0.013949499,0.7844724],"study_design_scores_gemma":[0.00004300222,0.00010783285,0.0007761685,0.000041967112,0.00006451908,0.00037183976,0.0001007732,0.9179033,0.012076823,0.029975936,0.03842197,0.00011592719],"about_ca_topic_score_codex":0.005058565,"about_ca_topic_score_gemma":0.0060485564,"teacher_disagreement_score":0.005058565,"about_ca_system_score_codex":0.0010947217,"about_ca_system_score_gemma":0.0019590473,"threshold_uncertainty_score":0.016899526},"labels":[],"label_agreement":null},{"id":"W2106899369","doi":"10.1142/s0218001408006922","title":"CLUSTERING QUALITY MEASURES BASED ON COMPARING THE PROXIMITY MATRICES FOR THE MEMBERSHIP VECTORS AND THE OBJECTS","year":2008,"lang":"en","type":"article","venue":"International Journal of Pattern Recognition and Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Cluster analysis; Data mining; Pattern recognition (psychology); Rand index; Partition (number theory); Mathematics; Feature vector; Artificial intelligence; Fuzzy clustering; Correlation clustering; Metric (unit); Feature (linguistics); Single-linkage clustering; Computer science; CURE data clustering algorithm; Combinatorics","score_opus":0.2650123592300616,"score_gpt":0.3802203558425662,"score_spread":0.11520799661250458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106899369","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03740804,0.0014976674,0.9554039,0.00023286525,0.00014217185,0.00026501337,0.00053510204,0.0006771716,0.003838078],"genre_scores_gemma":[0.37319314,0.0011240339,0.62137103,0.00015102273,0.00019534271,0.00042567926,0.0016884614,0.00037450777,0.0014768463],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98857856,0.002611662,0.0012439905,0.0014078451,0.00586509,0.00029286838],"domain_scores_gemma":[0.958984,0.020200595,0.0060584885,0.0034096586,0.010787422,0.00055984606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011929441,0.0011965248,0.0016577132,0.009992859,0.0011611931,0.00370012,0.0012606676,0.0013553642,0.0023898464],"category_scores_gemma":[0.048052594,0.0004615579,0.0012138053,0.006322928,0.0018900504,0.004998588,0.0019159165,0.0014266397,0.001011417],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009909829,0.00033223728,0.053010665,0.0021864504,0.0015269236,0.00024764298,0.0015830192,0.16357915,0.029908095,0.08552748,0.011577464,0.6495299],"study_design_scores_gemma":[0.00015582236,0.0013763128,0.08248232,0.000514141,0.0006668299,0.0021704303,0.0020482864,0.7036814,0.049582,0.12358411,0.03303694,0.0007013541],"about_ca_topic_score_codex":0.0019013516,"about_ca_topic_score_gemma":0.0016665047,"teacher_disagreement_score":0.011929441,"about_ca_system_score_codex":0.0015198085,"about_ca_system_score_gemma":0.0011353083,"threshold_uncertainty_score":0.06308967},"labels":[],"label_agreement":null},{"id":"W2111253950","doi":"10.1109/dexa.2002.1045928","title":"Clustering Web sessions by sequence alignment","year":2004,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":114,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Cluster analysis; Sequence (biology); World Wide Web; Information retrieval; Artificial intelligence","score_opus":0.03441255174344665,"score_gpt":0.32684877228899994,"score_spread":0.2924362205455533,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111253950","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08892272,0.0007211943,0.9019054,0.00021290587,0.00012025764,0.00066951365,0.0017128584,0.0035757674,0.0021594109],"genre_scores_gemma":[0.22014402,0.00070735265,0.76633877,0.00010703593,0.000118650976,0.00069522695,0.007876202,0.00056815374,0.0034444712],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99676263,0.000978131,0.00028854795,0.00084710424,0.0008628016,0.00026079922],"domain_scores_gemma":[0.9953029,0.0014965307,0.00070115534,0.0007544787,0.0014680534,0.00027698575],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016727494,0.0010786724,0.0013580756,0.007600408,0.0012239424,0.0016834778,0.0013668005,0.0014199973,0.0015990827],"category_scores_gemma":[0.008339362,0.0005270716,0.0013101383,0.008935637,0.0005480593,0.0023121631,0.0012530533,0.0010680575,0.002125849],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012573168,0.0009573512,0.029992443,0.00082178164,0.00043944747,0.00040952233,0.0015407699,0.061733436,0.042135447,0.013944789,0.011495623,0.83527195],"study_design_scores_gemma":[0.00010775547,0.0005845633,0.025013918,0.0001333897,0.00026024072,0.0012250431,0.0012558289,0.82826966,0.033597663,0.06915231,0.040189974,0.00020969482],"about_ca_topic_score_codex":0.0037789429,"about_ca_topic_score_gemma":0.0046734107,"teacher_disagreement_score":0.007600408,"about_ca_system_score_codex":0.00071497756,"about_ca_system_score_gemma":0.0019377042,"threshold_uncertainty_score":0.008846402},"labels":[],"label_agreement":null},{"id":"W2111596024","doi":"10.1109/tpami.2007.1138","title":"Cumulative Voting Consensus Method for Partitions with Variable Number of Clusters","year":2007,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":232,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Categorical variable; Probabilistic logic; Voting; Entropy (arrow of time); Consensus clustering; Data mining; Algorithm; Mathematics; Correlation clustering; Artificial intelligence; CURE data clustering algorithm; Machine learning","score_opus":0.034566888328313325,"score_gpt":0.3610294286362877,"score_spread":0.32646254030797434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111596024","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00269997,0.00007996673,0.9963361,0.000043730946,0.00002656988,0.00003662076,0.000020808688,0.00021490337,0.00054142624],"genre_scores_gemma":[0.14021245,0.0001581997,0.85452634,0.000107010404,0.00009081423,0.00031914687,0.00032959264,0.00023190092,0.004024473],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972192,0.0008593307,0.00014553465,0.0006256073,0.00096765975,0.0001826871],"domain_scores_gemma":[0.9969081,0.0011571047,0.00021067039,0.00048540326,0.0011104572,0.00012819657],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040977104,0.00097266806,0.0017387317,0.0019641987,0.0012241033,0.0014201372,0.0028569633,0.0018045943,0.0037500903],"category_scores_gemma":[0.011113053,0.00048363995,0.0011218229,0.0020708514,0.0011508074,0.0025949848,0.0017359746,0.0014166058,0.0012541392],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035111496,0.00006643172,0.0008239607,0.0001737787,0.0001224978,0.00011814251,0.00039373667,0.5128079,0.00761568,0.101954214,0.0051271245,0.37044537],"study_design_scores_gemma":[0.000022854958,0.000030078256,0.00010818918,0.000009946358,0.000013408536,0.000030854066,0.000025069181,0.9717997,0.0019641377,0.023626685,0.0023544198,0.000014721327],"about_ca_topic_score_codex":0.0046302257,"about_ca_topic_score_gemma":0.004506739,"teacher_disagreement_score":0.0046302257,"about_ca_system_score_codex":0.0015740396,"about_ca_system_score_gemma":0.0019051932,"threshold_uncertainty_score":0.021671057},"labels":[],"label_agreement":null},{"id":"W2111827030","doi":"10.1186/1756-0381-7-7","title":"Semi-supervised consensus clustering for gene expression data analysis","year":2014,"lang":"en","type":"article","venue":"BioData Mining","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"National Research Council Canada","keywords":"Cluster analysis; Computer science; Data mining; Consensus clustering; Computational biology; Expression (computer science); Artificial intelligence; Biology; Correlation clustering; CURE data clustering algorithm","score_opus":0.10870116120127883,"score_gpt":0.350048260191916,"score_spread":0.24134709899063717,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111827030","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048176856,0.0003067981,0.9933073,0.000076289616,0.000028293702,0.00008718965,0.0001320129,0.0009445902,0.00029987356],"genre_scores_gemma":[0.12970492,0.00033533238,0.8670255,0.00009254944,0.00006963529,0.0007530796,0.001086307,0.00025969083,0.0006729325],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934989,0.0024991746,0.00041448063,0.001443121,0.0019979652,0.00014630263],"domain_scores_gemma":[0.99022293,0.004703601,0.0010365195,0.0012198902,0.0026371626,0.00017986055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057240664,0.001321783,0.0023112008,0.00288702,0.0010140119,0.0011492053,0.0027144072,0.001711551,0.0016915504],"category_scores_gemma":[0.014415616,0.0007080265,0.0019309913,0.0031342518,0.0014998583,0.0013623812,0.0012694037,0.002125851,0.0015209311],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032022403,0.0002092676,0.0027090423,0.0010150252,0.00053231395,0.00013991639,0.00041757934,0.657284,0.016900728,0.011743088,0.0056023295,0.30312645],"study_design_scores_gemma":[0.000012550233,0.00003164614,0.0006001565,0.000020787982,0.00001940386,0.00005141592,0.000026929085,0.98296356,0.0040430576,0.010800973,0.0014011542,0.000028431314],"about_ca_topic_score_codex":0.004231688,"about_ca_topic_score_gemma":0.0037342615,"teacher_disagreement_score":0.0057240664,"about_ca_system_score_codex":0.001808852,"about_ca_system_score_gemma":0.0026686862,"threshold_uncertainty_score":0.030272126},"labels":[],"label_agreement":null},{"id":"W2112317819","doi":"10.1109/apccas.2008.4746001","title":"Cluster validation for subspace clustering on high dimensional data","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Subspace topology; Computer science; Clustering high-dimensional data; Cluster (spacecraft); Data mining; k-medians clustering; Correlation clustering; Single-linkage clustering; Fuzzy clustering; CURE data clustering algorithm; Pattern recognition (psychology); Determining the number of clusters in a data set; Artificial intelligence","score_opus":0.11038308959026868,"score_gpt":0.3447051931064546,"score_spread":0.2343221035161859,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112317819","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015074954,0.00017164397,0.9834415,0.0001214761,0.00003974773,0.00009984196,0.00008991627,0.00042433094,0.00053673395],"genre_scores_gemma":[0.2941871,0.00019251066,0.70297545,0.00013090605,0.000056550543,0.00050301297,0.0010364077,0.00023527766,0.00068288384],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9833017,0.008290339,0.00083296944,0.0016006044,0.005608906,0.0003655062],"domain_scores_gemma":[0.9710724,0.01360944,0.002545373,0.0048179417,0.007494271,0.00046060514],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017049951,0.0008579677,0.0015121656,0.0037951476,0.002359288,0.0021502962,0.0016981225,0.0015036811,0.0012016543],"category_scores_gemma":[0.054079134,0.00037829854,0.0011268944,0.0030653733,0.0027052255,0.0026478372,0.0028397322,0.0023150789,0.000663298],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010114111,0.00030326637,0.01562112,0.0006295099,0.0005375038,0.00024767066,0.0011295121,0.41935807,0.0216541,0.09198956,0.006815346,0.44070294],"study_design_scores_gemma":[0.000021371445,0.000102452024,0.0022291124,0.000046847745,0.00001691088,0.000098515775,0.00013432714,0.9643608,0.00852301,0.022341847,0.0020689722,0.000055903307],"about_ca_topic_score_codex":0.0022131852,"about_ca_topic_score_gemma":0.0020565414,"teacher_disagreement_score":0.017049951,"about_ca_system_score_codex":0.0014715736,"about_ca_system_score_gemma":0.0023503432,"threshold_uncertainty_score":0.09016985},"labels":[],"label_agreement":null},{"id":"W2113334100","doi":"10.1016/j.jbi.2008.11.004","title":"Efficient layered density-based clustering of categorical data","year":2008,"lang":"en","type":"article","venue":"Journal of Biomedical Informatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Computer science; Categorical variable; Data mining; Scalability; CURE data clustering algorithm; Single-linkage clustering; Clustering high-dimensional data; Correlation clustering; Artificial intelligence; Pattern recognition (psychology); Machine learning; Database","score_opus":0.06798969841812494,"score_gpt":0.32353461207813417,"score_spread":0.2555449136600092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113334100","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010132434,0.000112477464,0.98823905,0.00010092218,0.0000169761,0.000049967395,0.00014226581,0.0009991287,0.00020668683],"genre_scores_gemma":[0.21774054,0.00022924367,0.7782068,0.000106029765,0.000046171725,0.000247516,0.001833674,0.0002379233,0.0013520503],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9965023,0.0012486957,0.00026191963,0.00031775655,0.0013234414,0.00034582245],"domain_scores_gemma":[0.99188143,0.0036963979,0.00038940378,0.0014657258,0.00223701,0.0003300836],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036955136,0.00084594445,0.0024402265,0.0025563755,0.001212391,0.0025512897,0.004327216,0.0015793211,0.0018039116],"category_scores_gemma":[0.01941756,0.0012750638,0.002176781,0.0031194433,0.0008752475,0.00259129,0.004714425,0.0023973717,0.0015703203],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056693144,0.0002833341,0.006224526,0.00028091815,0.00027153073,0.00012512576,0.0006562417,0.45046526,0.009701112,0.023076838,0.007709224,0.50063896],"study_design_scores_gemma":[0.000012639637,0.000020631684,0.0004532759,0.000009695768,0.000014104112,0.00004097215,0.000038540355,0.9893987,0.0012660556,0.008375354,0.00035380924,0.000016131646],"about_ca_topic_score_codex":0.01867349,"about_ca_topic_score_gemma":0.019791337,"teacher_disagreement_score":0.01867349,"about_ca_system_score_codex":0.0021826301,"about_ca_system_score_gemma":0.0034883125,"threshold_uncertainty_score":0.03712958},"labels":[],"label_agreement":null},{"id":"W2113439791","doi":"10.5539/cis.v4n3p88","title":"A Genetic K-means Clustering Algorithm Based on the Optimized Initial Centers","year":2011,"lang":"en","type":"article","venue":"Computer and Information Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Cluster analysis; Cluster (spacecraft); Convergence (economics); Algorithm; Genetic algorithm; k-medoids; k-medians clustering; CURE data clustering algorithm; Correlation clustering; Artificial intelligence; Machine learning","score_opus":0.035164000121881894,"score_gpt":0.2771619605297209,"score_spread":0.24199796040783902,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113439791","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006243119,0.0001677286,0.9912907,0.00007375911,0.000054537173,0.000083202256,0.000040597602,0.00088581647,0.0011605303],"genre_scores_gemma":[0.074962236,0.00021925542,0.9217492,0.000069489135,0.000035236444,0.0003680669,0.0002566964,0.00017984482,0.0021599776],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989618,0.00019392821,0.000045048244,0.00034134148,0.00036638865,0.00009147215],"domain_scores_gemma":[0.99931717,0.0001557422,0.00007279197,0.00007819254,0.0003432251,0.000032886288],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008158673,0.0014735662,0.001639262,0.001779395,0.0016580675,0.0011907121,0.002390823,0.0017892809,0.0014782674],"category_scores_gemma":[0.0024900886,0.0007473806,0.0010747465,0.0024248336,0.0010463584,0.0014646972,0.0010512522,0.0014868327,0.0010356654],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014761643,0.00011324991,0.0020390048,0.00025114798,0.00018590501,0.00013989353,0.00030808512,0.58208185,0.018062577,0.01892113,0.008488504,0.36926112],"study_design_scores_gemma":[0.000043562184,0.000046149966,0.00050053076,0.000020008003,0.000039268172,0.00013534419,0.000046774418,0.98128015,0.0074909762,0.006046073,0.0042955605,0.000055536675],"about_ca_topic_score_codex":0.013695828,"about_ca_topic_score_gemma":0.009586503,"teacher_disagreement_score":0.013695828,"about_ca_system_score_codex":0.0014530898,"about_ca_system_score_gemma":0.0029014575,"threshold_uncertainty_score":0.02723223},"labels":[],"label_agreement":null},{"id":"W2113755034","doi":"10.1145/1569901.1570232","title":"Evolutionary clustering with arbitrary subspaces","year":2009,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Linear subspace; Subspace topology; Computer science; Correlation clustering; Canopy clustering algorithm; Set (abstract data type); Constrained clustering; CURE data clustering algorithm; Evolutionary algorithm; Algorithm; Data mining; Mathematics; Mathematical optimization; Theoretical computer science; Artificial intelligence","score_opus":0.012674477805760215,"score_gpt":0.2639241954487666,"score_spread":0.2512497176430064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113755034","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021474611,0.00015637882,0.9748857,0.0001218687,0.000018612403,0.00003616821,0.000029085322,0.000108290915,0.003169261],"genre_scores_gemma":[0.33773738,0.0003029823,0.65625966,0.00012538431,0.00002667634,0.00022753543,0.00019094736,0.00006676133,0.0050626835],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991793,0.00031928602,0.000039938757,0.00017458531,0.0002104411,0.000076399076],"domain_scores_gemma":[0.99940705,0.00021206247,0.00006078247,0.00013391039,0.00015012706,0.00003606415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012169663,0.00051894586,0.00077799853,0.0008160371,0.0006784739,0.0010080084,0.0009966048,0.0008013823,0.0013667458],"category_scores_gemma":[0.0028703122,0.00034115565,0.000752792,0.0011157544,0.0010455651,0.0012074383,0.0015140041,0.00066800293,0.0003017123],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000036165184,0.00003333784,0.00066702784,0.000053618172,0.000054498076,0.0000712824,0.00013123149,0.83033276,0.0029006617,0.086440586,0.0007348591,0.07854392],"study_design_scores_gemma":[0.0000066126486,0.000028144512,0.00019633191,0.000008229096,0.00000802299,0.000037285918,0.000030017447,0.96363586,0.0007767422,0.032784387,0.0024791085,0.0000092061355],"about_ca_topic_score_codex":0.0018939989,"about_ca_topic_score_gemma":0.0017485897,"teacher_disagreement_score":0.0018939989,"about_ca_system_score_codex":0.00075882976,"about_ca_system_score_gemma":0.0007875042,"threshold_uncertainty_score":0.0064359903},"labels":[],"label_agreement":null},{"id":"W2114217619","doi":"10.1109/nnsp.1997.622414","title":"An improved scheme for the fuzzifier in fuzzy clustering","year":2002,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Fuzzy clustering; Data mining; Computer science; Correlation clustering; Fuzzy set; Embedding; Constrained clustering; Scheme (mathematics); Probabilistic logic; Fuzzy logic; Artificial intelligence; CURE data clustering algorithm; Mathematics","score_opus":0.04832545442196626,"score_gpt":0.31937943448412687,"score_spread":0.27105398006216064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114217619","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0047087315,0.00016201129,0.99415946,0.0001228363,0.000065534936,0.000056869467,0.000023379906,0.00016876805,0.0005324732],"genre_scores_gemma":[0.10491457,0.00020371965,0.89265287,0.00011395924,0.00009208445,0.000112972266,0.0000919993,0.00007108131,0.0017467468],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960146,0.0011492412,0.00030697606,0.00081408507,0.0014426818,0.00027249457],"domain_scores_gemma":[0.99550986,0.0015124236,0.000256591,0.0014732232,0.0010939903,0.00015400519],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059418553,0.0007262581,0.0015346352,0.0017174942,0.0017495004,0.002184779,0.0038968895,0.0027960115,0.0029652668],"category_scores_gemma":[0.013511618,0.0005508951,0.001317619,0.001874006,0.0019112827,0.0045106364,0.0033328736,0.0029083097,0.0011596157],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007345283,0.000120144716,0.0013500841,0.00029031452,0.00010853517,0.00022194767,0.0011483585,0.2330012,0.043494523,0.33149847,0.003186314,0.38484555],"study_design_scores_gemma":[0.00006008782,0.0001274777,0.0003856999,0.000041507657,0.00003726521,0.0001404802,0.000039731498,0.92277086,0.010858793,0.058637105,0.006826804,0.000074221134],"about_ca_topic_score_codex":0.0029633415,"about_ca_topic_score_gemma":0.0022585597,"teacher_disagreement_score":0.0059418553,"about_ca_system_score_codex":0.0017563225,"about_ca_system_score_gemma":0.0016363645,"threshold_uncertainty_score":0.031423926},"labels":[],"label_agreement":null},{"id":"W2114848212","doi":"10.1016/j.neucom.2011.08.032","title":"Clustering in applications with multiple data sources—A mutual subspace clustering approach","year":2012,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Linear subspace; Cluster analysis; Computer science; Subspace topology; Data mining; Clustering high-dimensional data; Set (abstract data type); Data set; Data point; Correlation clustering; Pattern recognition (psychology); Cluster (spacecraft); Artificial intelligence; Mathematics","score_opus":0.057235792663626425,"score_gpt":0.30596122055709757,"score_spread":0.24872542789347113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114848212","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033601788,0.00026114334,0.9954277,0.00012385815,0.00002136992,0.000027226426,0.000019327152,0.00009884546,0.00066024775],"genre_scores_gemma":[0.23719943,0.0009121241,0.7578861,0.00012270185,0.00022738261,0.00023076608,0.00023413113,0.0001418315,0.0030455762],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968227,0.0012703355,0.00016703192,0.00045317048,0.001139554,0.00014709198],"domain_scores_gemma":[0.9978479,0.0007961255,0.00020803955,0.00040256625,0.0006507899,0.0000946203],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027135566,0.00071452867,0.0013268283,0.001632988,0.0013284967,0.0018759388,0.0021461463,0.0013163615,0.0013192199],"category_scores_gemma":[0.0062506534,0.0006478633,0.0015345139,0.0024146386,0.0010515507,0.0031355494,0.0034883886,0.0015751611,0.0005897054],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029058388,0.00025025566,0.002819224,0.00047587018,0.000563913,0.00023753548,0.00076765486,0.35853794,0.012492514,0.12924568,0.0050953194,0.4892235],"study_design_scores_gemma":[0.000007743885,0.000057463025,0.0006456187,0.000019273106,0.000036945527,0.000116015304,0.00010629738,0.96242756,0.0021610458,0.031525627,0.0028644213,0.000031957225],"about_ca_topic_score_codex":0.001737509,"about_ca_topic_score_gemma":0.0025438454,"teacher_disagreement_score":0.0027135566,"about_ca_system_score_codex":0.0005753244,"about_ca_system_score_gemma":0.0013613137,"threshold_uncertainty_score":0.0143508315},"labels":[],"label_agreement":null},{"id":"W2117544284","doi":"10.1145/375663.375672","title":"Data bubbles","year":2001,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":79,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Cluster analysis; Computer science; CURE data clustering algorithm; Canopy clustering algorithm; Data mining; Data stream clustering; Correlation clustering; Hierarchical clustering; Data compression; Data set; Set (abstract data type); Fuzzy clustering; Constrained clustering; Determining the number of clusters in a data set; Artificial intelligence","score_opus":0.12925317243936482,"score_gpt":0.3804818146787029,"score_spread":0.25122864223933805,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117544284","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015993237,0.004075348,0.9381399,0.0061973585,0.0017125609,0.0006598126,0.0034108998,0.0034225995,0.026388329],"genre_scores_gemma":[0.18338038,0.003671493,0.767874,0.0046767252,0.00097550685,0.0013011125,0.008454241,0.0015910002,0.02807548],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9951416,0.0010683516,0.00037993657,0.0011069165,0.0020657296,0.00023747134],"domain_scores_gemma":[0.9858145,0.0064811977,0.0006928246,0.004297899,0.0021996389,0.00051400595],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004271393,0.000717883,0.0010154219,0.0022303541,0.0015853558,0.0035803707,0.0029173167,0.0019068886,0.021606643],"category_scores_gemma":[0.028743545,0.0005578107,0.0010411231,0.0029260837,0.0018663015,0.010513638,0.006357067,0.0024246767,0.007120849],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059662224,0.00013087003,0.0033860994,0.00095909566,0.00009543943,0.0007352084,0.0013554082,0.011408863,0.0067782626,0.5116902,0.06766343,0.39520052],"study_design_scores_gemma":[0.000083944666,0.00020452446,0.001039186,0.00025064667,0.000053819975,0.0016241075,0.0008639693,0.053182658,0.010803821,0.32851547,0.6032785,0.00009935022],"about_ca_topic_score_codex":0.0014907945,"about_ca_topic_score_gemma":0.0010565716,"teacher_disagreement_score":0.021606643,"about_ca_system_score_codex":0.0011905369,"about_ca_system_score_gemma":0.0011855017,"threshold_uncertainty_score":0.07228148},"labels":[],"label_agreement":null},{"id":"W2118001196","doi":"10.1145/2464576.2464653","title":"Flat vs. symbiotic evolutionary subspace clusterings","year":2013,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Subspace topology; Cluster analysis; Cluster (spacecraft); Computer science; Representation (politics); Space (punctuation); Property (philosophy); Artificial intelligence; Pattern recognition (psychology); Mathematics; Data mining","score_opus":0.011335472641119722,"score_gpt":0.24955873640691234,"score_spread":0.23822326376579261,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118001196","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15939236,0.00057637686,0.83044994,0.00027042994,0.000037308564,0.00013959665,0.00018149418,0.00042749665,0.008525113],"genre_scores_gemma":[0.8098971,0.00035773267,0.18503384,0.0001475912,0.000030540235,0.00011341285,0.00038662754,0.00008401862,0.0039492142],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9982035,0.00050782075,0.00011384433,0.000350338,0.00061159435,0.00021283411],"domain_scores_gemma":[0.99705565,0.0007134783,0.00028135648,0.00088798214,0.00080741436,0.00025418098],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026952417,0.0006181608,0.0006844332,0.0013063181,0.000872575,0.0017973968,0.0016388037,0.00093141187,0.002229692],"category_scores_gemma":[0.008584601,0.0002459206,0.00063030294,0.0013959057,0.0015249576,0.0021015683,0.0031847109,0.00083630625,0.00057672797],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044499122,0.00015433694,0.011841179,0.00020952322,0.00021313656,0.0005670917,0.0009858775,0.35938323,0.013069652,0.24758545,0.002625614,0.3629199],"study_design_scores_gemma":[0.000026711396,0.00021273234,0.005542738,0.00004334687,0.000031922467,0.0004556543,0.00036172912,0.87405086,0.004320678,0.11009076,0.004813103,0.0000496935],"about_ca_topic_score_codex":0.0020059424,"about_ca_topic_score_gemma":0.0024446226,"teacher_disagreement_score":0.0026952417,"about_ca_system_score_codex":0.0007286585,"about_ca_system_score_gemma":0.00071783614,"threshold_uncertainty_score":0.014253974},"labels":[],"label_agreement":null},{"id":"W2118013824","doi":"10.1007/s10115-010-0367-z","title":"Statistical semantics for enhancing document clustering","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Semantic similarity; Computer science; Cluster analysis; Document clustering; Similarity (geometry); Vector space model; Explicit semantic analysis; Semantics (computer science); Representation (politics); Benchmark (surveying); Context (archaeology); Artificial intelligence; Information retrieval; Distributional semantics; Data mining; Natural language processing; Semantic computing; Semantic technology; Semantic Web","score_opus":0.012131819968422273,"score_gpt":0.3032534578081119,"score_spread":0.29112163783968964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2118013824","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01615167,0.0005797925,0.980714,0.00023274643,0.0000763006,0.000045160596,0.00024536686,0.00092978746,0.0010251465],"genre_scores_gemma":[0.30144566,0.0009334723,0.6934118,0.00023642472,0.00026576294,0.00018679384,0.0013323271,0.00037024973,0.0018174424],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997378,0.0008933295,0.00025720958,0.00040669402,0.0009632908,0.00010152507],"domain_scores_gemma":[0.9923922,0.0036574795,0.00058113073,0.001400499,0.0017715495,0.00019709062],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002960821,0.00064638007,0.0011618569,0.004362243,0.00089667185,0.002245253,0.0012523493,0.0010406084,0.0016099273],"category_scores_gemma":[0.01577654,0.00048700385,0.0011384137,0.005239945,0.0010953017,0.005560641,0.0018955959,0.0013198333,0.0008123124],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059983664,0.00039364892,0.0039997096,0.000673744,0.00022336084,0.00015717099,0.0005300961,0.07697785,0.025919242,0.22646324,0.0065409737,0.65752107],"study_design_scores_gemma":[0.00007092633,0.00019988291,0.0016641904,0.000079455625,0.00015878527,0.00031444526,0.00019806788,0.60946906,0.016326057,0.35849565,0.012958665,0.00006491266],"about_ca_topic_score_codex":0.0013263229,"about_ca_topic_score_gemma":0.0025237459,"teacher_disagreement_score":0.004362243,"about_ca_system_score_codex":0.00097384804,"about_ca_system_score_gemma":0.0014398901,"threshold_uncertainty_score":0.015658498},"labels":[],"label_agreement":null},{"id":"W2120468118","doi":"10.1007/978-3-540-71701-0_5","title":"Hierarchical Density-Based Clustering of Categorical Data and a Simplification","year":2007,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Categorical variable; Cluster analysis; Data mining; Computer science; CURE data clustering algorithm; Outlier; Pattern recognition (psychology); Basis (linear algebra); Hierarchical clustering; Single-linkage clustering; Canopy clustering algorithm; Artificial intelligence; Correlation clustering; Algorithm; Mathematics; Machine learning","score_opus":0.06047806953614623,"score_gpt":0.33107722062942646,"score_spread":0.27059915109328025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120468118","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031519611,0.0003088859,0.9944622,0.00021459002,0.00004758766,0.000052084397,0.00019347215,0.00019482935,0.0013744126],"genre_scores_gemma":[0.11312783,0.0010307058,0.87678546,0.00023759059,0.00020532536,0.00037456964,0.0017342858,0.00029333823,0.006210853],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99403787,0.0022173298,0.0003918144,0.0008867842,0.002215981,0.00025028715],"domain_scores_gemma":[0.9900878,0.0044384673,0.0003168407,0.0032213747,0.0017696224,0.00016596478],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005645031,0.0005863712,0.0022060291,0.0036518497,0.0016948162,0.0026367176,0.0050120577,0.0013234288,0.0039045953],"category_scores_gemma":[0.034223948,0.0012950932,0.0030672038,0.007533182,0.002215559,0.0042029163,0.0043411064,0.003419992,0.0019491146],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000103728635,0.00008394739,0.0020371105,0.00041637203,0.00016302441,0.00019355232,0.0009836281,0.097563244,0.0029654813,0.605687,0.014733083,0.27506977],"study_design_scores_gemma":[0.000023750645,0.000044317363,0.0025741947,0.00007372838,0.0000730208,0.0004425906,0.00025744535,0.42045188,0.0016200272,0.55973226,0.014637747,0.00006902798],"about_ca_topic_score_codex":0.009517566,"about_ca_topic_score_gemma":0.010255883,"teacher_disagreement_score":0.009517566,"about_ca_system_score_codex":0.0021878632,"about_ca_system_score_gemma":0.0019890147,"threshold_uncertainty_score":0.029854178},"labels":[],"label_agreement":null},{"id":"W2120498775","doi":"10.1109/icsmc.2007.4413889","title":"Hard-fuzzy clustering: A cooperative approach","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Fuzzy clustering; Correlation clustering; CURE data clustering algorithm; Data mining; Canopy clustering algorithm; Data stream clustering; Clustering high-dimensional data; Artificial intelligence; Single-linkage clustering; Fuzzy logic; Consensus clustering; Machine learning","score_opus":0.04187945832137895,"score_gpt":0.3151919733077678,"score_spread":0.27331251498638887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2120498775","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008331546,0.0002605584,0.9881443,0.00026887425,0.000034710483,0.00006720818,0.000023655282,0.00015677763,0.0027123075],"genre_scores_gemma":[0.5687965,0.0005043834,0.41998827,0.00035836798,0.00017039686,0.00031187673,0.0001767813,0.000120113706,0.009573314],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9972378,0.0007144265,0.00011548935,0.0007141805,0.0010182636,0.00019998186],"domain_scores_gemma":[0.99751174,0.0009238311,0.00021150736,0.00041342629,0.00075971434,0.00017966409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030537231,0.0012001868,0.0016202489,0.0018055538,0.0016021323,0.0020842894,0.004995271,0.0026024005,0.0024626276],"category_scores_gemma":[0.0050600343,0.0007809797,0.0013536904,0.0015017412,0.0018437637,0.0026772425,0.003546025,0.0016721961,0.000817987],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013760073,0.000120255,0.0010136665,0.0001847195,0.000115577204,0.00019712011,0.00048591764,0.81462175,0.0047205915,0.04827751,0.0024545568,0.12767078],"study_design_scores_gemma":[0.000008962931,0.000028250024,0.000116848685,0.000007458347,0.000011335848,0.000033669654,0.00004099686,0.9822237,0.0007526772,0.015486129,0.0012751365,0.0000147351575],"about_ca_topic_score_codex":0.007488455,"about_ca_topic_score_gemma":0.0053372327,"teacher_disagreement_score":0.007488455,"about_ca_system_score_codex":0.0018342651,"about_ca_system_score_gemma":0.0018583331,"threshold_uncertainty_score":0.016149819},"labels":[],"label_agreement":null},{"id":"W2122129637","doi":"10.1109/icmlc.2004.1382092","title":"A new efficient validity index for fuzzy clustering","year":2005,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Index (typography); Computer science; Cluster analysis; Fuzzy logic; Fuzzy clustering; Fuzzy set; Data mining; Artificial intelligence; World Wide Web","score_opus":0.05092293168649851,"score_gpt":0.3326614497502807,"score_spread":0.2817385180637822,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122129637","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00768565,0.00055739074,0.9881346,0.00015405331,0.00010155885,0.00014461829,0.00027872954,0.00032739982,0.0026159508],"genre_scores_gemma":[0.13946015,0.0005081387,0.8559637,0.00014992157,0.00030650193,0.00055749546,0.0014254663,0.00030643205,0.0013221191],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97460335,0.0058762445,0.0019967384,0.0023600024,0.014438022,0.00072568265],"domain_scores_gemma":[0.9589667,0.022642102,0.0028576462,0.0042168363,0.010579239,0.0007375563],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01556214,0.0013952286,0.0026604445,0.012591201,0.002581124,0.0060117766,0.003065596,0.0022563676,0.0026176127],"category_scores_gemma":[0.072060265,0.0006506444,0.0020485031,0.008257829,0.003354668,0.008059895,0.0038817024,0.0027918902,0.0012832927],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057468325,0.00022345613,0.013104397,0.0009958276,0.0005832274,0.0003331802,0.000733443,0.15694492,0.018855035,0.20169188,0.011597474,0.5943624],"study_design_scores_gemma":[0.00006829788,0.00027022927,0.0050105373,0.00019370759,0.00014067323,0.0007473404,0.00020682318,0.8104151,0.016116992,0.1526951,0.013886665,0.0002486],"about_ca_topic_score_codex":0.0017311086,"about_ca_topic_score_gemma":0.0010940941,"teacher_disagreement_score":0.01556214,"about_ca_system_score_codex":0.003304124,"about_ca_system_score_gemma":0.002803489,"threshold_uncertainty_score":0.0823015},"labels":[],"label_agreement":null},{"id":"W2126636333","doi":"10.1109/ictai.2005.57","title":"Determining the optimal number of clusters using a new evolutionary algorithm","year":2005,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Evolutionary algorithm; Cluster analysis; Benchmark (surveying); Computer science; Determining the number of clusters in a data set; Algorithm; Entropy (arrow of time); Evolutionary computation; Set (abstract data type); Cluster (spacecraft); Fitness function; Genetic algorithm; Data mining; Mathematical optimization; Mathematics; Artificial intelligence; Machine learning; Correlation clustering; Canopy clustering algorithm","score_opus":0.03369832156042135,"score_gpt":0.33716241651616463,"score_spread":0.3034640949557433,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126636333","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022182588,0.00011815986,0.97650003,0.00015144995,0.000033888773,0.00005859287,0.000017375474,0.00013300625,0.00080491015],"genre_scores_gemma":[0.12308983,0.00012394409,0.87508494,0.00009664418,0.000040918014,0.00020718915,0.00010837694,0.000057816076,0.001190349],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987632,0.0002901751,0.00009680361,0.0003210289,0.00044364645,0.000085168576],"domain_scores_gemma":[0.99794847,0.0008958553,0.00016613741,0.00018652664,0.0007132158,0.00008981898],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021579044,0.0007777684,0.0011585553,0.0019235811,0.00083398336,0.0012111472,0.0017887844,0.0018774701,0.0010183734],"category_scores_gemma":[0.007967368,0.00048570993,0.00089066074,0.0011882834,0.0008167124,0.0018027991,0.0012516256,0.00097343436,0.0001964818],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001240385,0.00012969258,0.0038328755,0.00009254198,0.00016731191,0.00017910756,0.0002859348,0.72780585,0.01583383,0.020168092,0.0012139643,0.23016678],"study_design_scores_gemma":[0.000027125341,0.000023158183,0.0003095332,0.000007883452,0.00001539242,0.00004908164,0.0000170526,0.99365556,0.0014612643,0.003659378,0.00076321495,0.000011442069],"about_ca_topic_score_codex":0.002711315,"about_ca_topic_score_gemma":0.0025350112,"teacher_disagreement_score":0.002711315,"about_ca_system_score_codex":0.00093609194,"about_ca_system_score_gemma":0.0011300235,"threshold_uncertainty_score":0.011412263},"labels":[],"label_agreement":null},{"id":"W2127033147","doi":"10.1109/cidm.2009.4938666","title":"Fuzzy p-mode prototypes: A generalization of frequency-based cluster prototypes for clustering categorical objects","year":2009,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Fuzzy clustering; Cluster analysis; Categorical variable; Pattern recognition (psychology); FLAME clustering; Fuzzy logic; Generalization; Fuzzy set; Artificial intelligence; Mathematics; Data mining; Computer science; Feature (linguistics); CURE data clustering algorithm; Machine learning","score_opus":0.02177035882438115,"score_gpt":0.32354277869719095,"score_spread":0.3017724198728098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127033147","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0132548455,0.00035226738,0.9839834,0.00007443272,0.000078106874,0.0001958064,0.000126861,0.0003510848,0.0015831796],"genre_scores_gemma":[0.19985358,0.00044303152,0.7961252,0.00010005845,0.00012922286,0.00058256974,0.0004019729,0.00014951536,0.0022148343],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975688,0.0005938119,0.00017038618,0.0007168185,0.00081581034,0.00013432192],"domain_scores_gemma":[0.99773717,0.0006701764,0.00022463046,0.00041334503,0.0008555557,0.00009904777],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025462213,0.0008269338,0.0011617402,0.0032939515,0.0011740352,0.001986213,0.0024626954,0.0016142535,0.0019374838],"category_scores_gemma":[0.009426462,0.0006878895,0.0014019689,0.0038584706,0.0017425218,0.0035482277,0.001480513,0.0012327736,0.0008713058],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053516723,0.00022240335,0.0043368973,0.00091563567,0.00035961083,0.00042835856,0.0019099927,0.08903807,0.032719254,0.11496884,0.0067882775,0.74777746],"study_design_scores_gemma":[0.00009129255,0.0007468083,0.004508496,0.00019492375,0.00014927088,0.0014823297,0.0006274454,0.82321316,0.019864699,0.125325,0.023478016,0.0003185281],"about_ca_topic_score_codex":0.0028543577,"about_ca_topic_score_gemma":0.0020800084,"teacher_disagreement_score":0.0032939515,"about_ca_system_score_codex":0.0007955339,"about_ca_system_score_gemma":0.0010370777,"threshold_uncertainty_score":0.013465881},"labels":[],"label_agreement":null},{"id":"W2127258367","doi":"10.1109/icdm.2002.1183901","title":"A parameterless method for efficiently discovering clusters of arbitrary shape in large datasets","year":2003,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Correlation clustering; CURE data clustering algorithm; sort; Single-linkage clustering; Data mining; Canopy clustering algorithm; Data stream clustering; Similarity (geometry); Fuzzy clustering; Artificial intelligence; Range (aeronautics); Scalability; Pattern recognition (psychology); Database","score_opus":0.02762599778356411,"score_gpt":0.35514079866748477,"score_spread":0.32751480088392065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127258367","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001152697,0.00016414539,0.99658304,0.000084046675,0.00004032638,0.00008284725,0.0000952558,0.0015797251,0.00021788181],"genre_scores_gemma":[0.0126346145,0.00013795203,0.98560834,0.000076705306,0.0000447282,0.00021974207,0.00040783538,0.00021450134,0.00065556116],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968143,0.0006637564,0.0002452346,0.0008946281,0.0012608971,0.00012117831],"domain_scores_gemma":[0.9955766,0.0015414635,0.00048298712,0.0014625355,0.0007654026,0.00017098388],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034836808,0.0022249182,0.002419994,0.0044178204,0.0023354248,0.0021860288,0.003613928,0.0028754976,0.0029253839],"category_scores_gemma":[0.011241094,0.0012770292,0.0017586271,0.006689096,0.0015952177,0.003867414,0.0027745466,0.0032047413,0.0034505185],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030694288,0.0001729408,0.0013501315,0.0004211856,0.00029076153,0.00018827408,0.00035838608,0.08737136,0.017526994,0.020958332,0.02518061,0.845874],"study_design_scores_gemma":[0.00019136256,0.00012951893,0.00090377714,0.000050841514,0.000082922,0.0007302831,0.000108152744,0.90589017,0.01180394,0.053088777,0.026892796,0.0001274577],"about_ca_topic_score_codex":0.003503174,"about_ca_topic_score_gemma":0.005487614,"teacher_disagreement_score":0.0044178204,"about_ca_system_score_codex":0.0011042557,"about_ca_system_score_gemma":0.0025609233,"threshold_uncertainty_score":0.018423676},"labels":[],"label_agreement":null},{"id":"W2129184179","doi":"10.1109/ictai.2008.62","title":"An Empirical Analysis on the Stability of Clustering Algorithms","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"","keywords":"Cluster analysis; Computer science; Stability (learning theory); CURE data clustering algorithm; Correlation clustering; Canopy clustering algorithm; Data mining; Scalability; Machine learning; Artificial intelligence; Unsupervised learning; Selection (genetic algorithm); Algorithm","score_opus":0.0953566139620425,"score_gpt":0.3693955361560635,"score_spread":0.27403892219402104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129184179","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7460708,0.0055668224,0.23531102,0.002492785,0.0002032972,0.00029975874,0.0013182274,0.0006885813,0.008048696],"genre_scores_gemma":[0.94522935,0.00072106195,0.050765704,0.00017162359,0.0001451762,0.00014709844,0.0019101641,0.00027810672,0.0006317624],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9635584,0.018182822,0.0026577397,0.00409298,0.010843283,0.0006648732],"domain_scores_gemma":[0.46527785,0.43094525,0.023704587,0.03571937,0.042120863,0.0022320473],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04730218,0.0008290396,0.0011011836,0.006282464,0.0020947293,0.0032759646,0.0018339831,0.0020496391,0.0018915848],"category_scores_gemma":[0.3889703,0.0005469298,0.0008285212,0.0062891035,0.0041461037,0.005647382,0.0019674974,0.0024867863,0.000555418],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030482344,0.000645099,0.47037882,0.0015970746,0.0018777013,0.0007609617,0.0066086734,0.17155679,0.008580952,0.06607986,0.01519987,0.25366598],"study_design_scores_gemma":[0.00014617456,0.0012338237,0.20408708,0.0005882804,0.00034608523,0.003039408,0.0034930909,0.67504025,0.010981468,0.08610374,0.014675407,0.00026516503],"about_ca_topic_score_codex":0.001917178,"about_ca_topic_score_gemma":0.0014829864,"teacher_disagreement_score":0.04730218,"about_ca_system_score_codex":0.0019925344,"about_ca_system_score_gemma":0.0012246176,"threshold_uncertainty_score":0.2501608},"labels":[],"label_agreement":null},{"id":"W2129192715","doi":"10.1109/foci.2007.371525","title":"Likelihood Based Fuzzy Clustering for Data Sets of Mixed Features","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Fuzzy clustering; Cluster analysis; FLAME clustering; Correlation clustering; Canopy clustering algorithm; CURE data clustering algorithm; Computer science; Data mining; Data stream clustering; Pattern recognition (psychology); Feature (linguistics); Artificial intelligence; Algorithm; Mathematics","score_opus":0.05533363372637953,"score_gpt":0.35965094907401307,"score_spread":0.3043173153476335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129192715","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015947954,0.00007556355,0.9979948,0.000051187595,0.0000059483705,0.000016961416,0.000014343429,0.000090228736,0.00015620116],"genre_scores_gemma":[0.07827102,0.0001615801,0.9196178,0.00006733509,0.0000366541,0.00018684742,0.00019136595,0.000103025886,0.0013643746],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969741,0.0009129455,0.00018086933,0.0006366379,0.001199693,0.000095754316],"domain_scores_gemma":[0.99699956,0.0016935053,0.00035800887,0.0002658949,0.00059471215,0.000088337416],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003753115,0.0009575207,0.001588247,0.0023376034,0.001198149,0.0019475493,0.0020606588,0.001276983,0.001277273],"category_scores_gemma":[0.011257695,0.0006907218,0.0015660169,0.0020263465,0.0016443203,0.002663869,0.0022923474,0.0016206484,0.0010623767],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037201334,0.00008666053,0.0021816045,0.0004099297,0.00022201872,0.00028954985,0.0006865377,0.52434766,0.015877822,0.10844702,0.0024786917,0.34460053],"study_design_scores_gemma":[0.000016290593,0.000043532156,0.0003647163,0.000018166123,0.00001579354,0.00012920891,0.000030192863,0.9646895,0.002670003,0.029785626,0.0021913466,0.00004555019],"about_ca_topic_score_codex":0.0025520048,"about_ca_topic_score_gemma":0.0024266345,"teacher_disagreement_score":0.003753115,"about_ca_system_score_codex":0.0019223163,"about_ca_system_score_gemma":0.0012621569,"threshold_uncertainty_score":0.019848585},"labels":[],"label_agreement":null},{"id":"W2129500661","doi":"10.1109/tkde.2008.162","title":"Mining Projected Clusters in High-Dimensional Spaces","year":2008,"lang":"en","type":"article","venue":"IEEE Transactions on Knowledge and Data Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Computer science; Curse of dimensionality; Linear subspace; Outlier; Clustering high-dimensional data; Data mining; CURE data clustering algorithm; Data point; Correlation clustering; Computation; Pattern recognition (psychology); Canopy clustering algorithm; Single-linkage clustering; Algorithm; Artificial intelligence; Mathematics","score_opus":0.03923416376241975,"score_gpt":0.2902071390881266,"score_spread":0.2509729753257069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129500661","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.062254626,0.00037787625,0.93497145,0.00025137703,0.00004252632,0.0001841339,0.0004301528,0.0006967588,0.0007910997],"genre_scores_gemma":[0.33653647,0.0005125635,0.65809417,0.00010140152,0.000094959585,0.00041475554,0.0028697595,0.00013304417,0.0012428489],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99439764,0.0017170828,0.00039058377,0.001161525,0.0019965617,0.0003366333],"domain_scores_gemma":[0.99090004,0.0037489207,0.0009833293,0.0012555055,0.0027737212,0.00033842959],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038430674,0.0014589146,0.002320229,0.0058765067,0.0020845176,0.0034048478,0.0027619621,0.002166633,0.00091697136],"category_scores_gemma":[0.018570248,0.0010569788,0.001927939,0.0059831226,0.001610611,0.0034013514,0.00403104,0.001920173,0.00085037],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007366231,0.0005352815,0.02626669,0.00088613713,0.000890365,0.0015013831,0.0028139418,0.51516795,0.009665623,0.06391372,0.011021431,0.36660084],"study_design_scores_gemma":[0.00003637567,0.000078055295,0.0026737624,0.000046894023,0.000041927673,0.0002743098,0.00062768225,0.919227,0.0024768417,0.0724598,0.002007036,0.00005036467],"about_ca_topic_score_codex":0.0033761282,"about_ca_topic_score_gemma":0.0031436025,"teacher_disagreement_score":0.0058765067,"about_ca_system_score_codex":0.0009341603,"about_ca_system_score_gemma":0.0016228295,"threshold_uncertainty_score":0.02032435},"labels":[],"label_agreement":null},{"id":"W2131287878","doi":"10.5539/mas.v7n9p10","title":"A Novel Center Point Initialization Technique for K-means Clustering Algorithm","year":2013,"lang":"en","type":"article","venue":"Modern Applied Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Initialization; Cluster analysis; Computer science; Histogram; Algorithm; Point (geometry); k-means clustering; Center (category theory); Data mining; Pattern recognition (psychology); Artificial intelligence; Mathematics; Image (mathematics)","score_opus":0.027544671519506275,"score_gpt":0.29343813413077985,"score_spread":0.26589346261127356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131287878","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020980006,0.0003036871,0.99578744,0.00006527487,0.000112158355,0.000048739512,0.00003610384,0.0007999899,0.0007486367],"genre_scores_gemma":[0.05484778,0.00051739573,0.94072104,0.00012883292,0.00018050603,0.00024020683,0.00034134806,0.00029614635,0.002726765],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99787176,0.0003220223,0.000109023276,0.00049756427,0.0010523399,0.00014729594],"domain_scores_gemma":[0.9988809,0.00018259042,0.00009187721,0.00015109743,0.0006327919,0.00006069228],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010622223,0.0012087045,0.0015898613,0.0018143688,0.0015881956,0.0014148153,0.0025348486,0.0021658095,0.0030538696],"category_scores_gemma":[0.0027721848,0.00072302,0.0013689636,0.0032708908,0.0010396908,0.0019795876,0.0017788775,0.0025894444,0.0037490644],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047735983,0.00017035835,0.0023015954,0.00056592736,0.00025773217,0.0003604059,0.0006877821,0.09597251,0.07217255,0.028727448,0.021536147,0.7767701],"study_design_scores_gemma":[0.00011055402,0.00016555215,0.0015923887,0.000060272156,0.000089656176,0.00070047297,0.00012914503,0.89321136,0.04992017,0.013228046,0.04062136,0.00017101709],"about_ca_topic_score_codex":0.0042901468,"about_ca_topic_score_gemma":0.0036560318,"teacher_disagreement_score":0.0042901468,"about_ca_system_score_codex":0.000941839,"about_ca_system_score_gemma":0.0017928478,"threshold_uncertainty_score":0.010216236},"labels":[],"label_agreement":null},{"id":"W2138629976","doi":"10.5539/mas.v4n7p148","title":"Ontology Based Fuzzy Document Clustering Scheme","year":2010,"lang":"en","type":"article","venue":"Modern Applied Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Computer science; Document clustering; Fuzzy clustering; Vector space model; Semantic similarity; Information retrieval; Ontology; Data mining; Clustering high-dimensional data; Artificial intelligence","score_opus":0.015323686627275629,"score_gpt":0.29284564271317476,"score_spread":0.27752195608589914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138629976","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012294976,0.0013671269,0.9592092,0.0005444274,0.00024346956,0.0008940662,0.0011652439,0.0014956044,0.022785826],"genre_scores_gemma":[0.14492588,0.0017133683,0.81757236,0.0002152299,0.00010597176,0.0007974672,0.002812932,0.000105792016,0.031750917],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976992,0.00024463158,0.00028419847,0.00046568204,0.0011420249,0.0001642197],"domain_scores_gemma":[0.99888545,0.00012388296,0.00008285576,0.00015363368,0.00070351735,0.00005067301],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014423282,0.00056909694,0.0010785362,0.0034883737,0.0027065817,0.0027542058,0.0020887752,0.0014230012,0.005279831],"category_scores_gemma":[0.0025933033,0.00026322325,0.0012768966,0.0047029476,0.00078783516,0.0023895658,0.0012528325,0.00078542455,0.0025418256],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002530524,0.00016323713,0.0014452673,0.0007196407,0.0001640413,0.00040418224,0.0010139036,0.07018068,0.019322416,0.19350858,0.024928676,0.6878964],"study_design_scores_gemma":[0.00015416909,0.00020024674,0.0024103215,0.00020283784,0.00020133037,0.0017997012,0.0005863657,0.56102705,0.030688198,0.12128257,0.28120825,0.00023888027],"about_ca_topic_score_codex":0.0140766185,"about_ca_topic_score_gemma":0.009799498,"teacher_disagreement_score":0.0140766185,"about_ca_system_score_codex":0.003353635,"about_ca_system_score_gemma":0.004035384,"threshold_uncertainty_score":0.027989388},"labels":[],"label_agreement":null},{"id":"W2141277572","doi":"10.1109/tkde.2013.53","title":"Extended Subtree: A New Similarity Function for Tree Structured Data","year":2013,"lang":"en","type":"article","venue":"IEEE Transactions on Knowledge and Data Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Similarity (geometry); Edit distance; Cluster analysis; Data mining; Tree (set theory); Function (biology); Theoretical computer science; Artificial intelligence; Mathematics","score_opus":0.058334273774491116,"score_gpt":0.3147285319055442,"score_spread":0.25639425813105304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141277572","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018552322,0.00072548515,0.9778242,0.00013983005,0.000095705014,0.000098585326,0.00046011346,0.0012525616,0.0008512388],"genre_scores_gemma":[0.17310433,0.0006824738,0.821148,0.00019667827,0.00015021858,0.00029239597,0.0022966065,0.0004138959,0.0017154814],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965545,0.0006672787,0.00044956984,0.0004965391,0.0017033953,0.0001287311],"domain_scores_gemma":[0.9962134,0.001222057,0.00046824792,0.0008346188,0.0010454516,0.00021628893],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020865132,0.0006251882,0.0011114791,0.0041650156,0.00064342126,0.0016451104,0.0016604419,0.0012452323,0.0014690218],"category_scores_gemma":[0.009092638,0.00028146862,0.0011802017,0.0056081,0.00065390277,0.0056574373,0.0026038205,0.0011617058,0.001104382],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035406242,0.00020077499,0.005824598,0.0005695277,0.00021830612,0.00031622913,0.0008288116,0.040022794,0.033286124,0.05511758,0.009282807,0.85397846],"study_design_scores_gemma":[0.00009091879,0.0009738301,0.0061877873,0.00017894206,0.00014824962,0.0027839465,0.0005954216,0.7546619,0.032485835,0.13474983,0.066936545,0.00020683679],"about_ca_topic_score_codex":0.0010391814,"about_ca_topic_score_gemma":0.0014896074,"teacher_disagreement_score":0.0041650156,"about_ca_system_score_codex":0.0007460965,"about_ca_system_score_gemma":0.0010801341,"threshold_uncertainty_score":0.011034727},"labels":[],"label_agreement":null},{"id":"W2142989018","doi":"10.1109/icws.2008.120","title":"Adaptive Quality Recommendation Mechanism for Software Service Provisioning","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Dynamism; Cluster analysis; Granularity; Provisioning; Service provider; Feature (linguistics); Service (business); Quality of service; Data mining; Software; Index (typography); Quality (philosophy); Computer network; World Wide Web; Artificial intelligence","score_opus":0.12744566175716318,"score_gpt":0.3667250012835386,"score_spread":0.2392793395263754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142989018","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031534538,0.00059561,0.9581112,0.00046100558,0.0001742039,0.0002305737,0.000111714646,0.0065704165,0.0022107344],"genre_scores_gemma":[0.740847,0.00032880635,0.25479934,0.00036300236,0.00026328274,0.00022485665,0.00035636136,0.00013075446,0.002686625],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956267,0.0007446164,0.0003123774,0.0009797983,0.0020205323,0.0003159481],"domain_scores_gemma":[0.99061126,0.002355067,0.0011333786,0.0021877782,0.003315971,0.0003965408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047365464,0.0007521821,0.0012594254,0.0023976853,0.0011699883,0.0017027882,0.0039786045,0.0017262176,0.0019209974],"category_scores_gemma":[0.013223868,0.0005035143,0.0005788647,0.0019310244,0.0005882807,0.00293606,0.001225874,0.0018679749,0.0010427351],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010647193,0.0008725794,0.011667826,0.0002640988,0.0003546762,0.00031816502,0.00048946106,0.09365461,0.0354063,0.017526787,0.0139732715,0.8244075],"study_design_scores_gemma":[0.00010904244,0.00017809766,0.0032826837,0.000020987585,0.00010178356,0.000327612,0.00008141015,0.9719438,0.011710313,0.006355603,0.0057715424,0.00011717629],"about_ca_topic_score_codex":0.0069043217,"about_ca_topic_score_gemma":0.005216601,"teacher_disagreement_score":0.0069043217,"about_ca_system_score_codex":0.001519033,"about_ca_system_score_gemma":0.0010335738,"threshold_uncertainty_score":0.025049508},"labels":[],"label_agreement":null},{"id":"W2147809251","doi":"10.1007/978-0-387-36795-8_9","title":"Unsupervised Learning: Clustering","year":2007,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Unsupervised learning; Computer science; Artificial intelligence","score_opus":0.06366156814888656,"score_gpt":0.312241148465601,"score_spread":0.24857958031671445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147809251","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00054612115,0.008718099,0.9497231,0.0008864244,0.0006922381,0.000072142735,0.0003269716,0.002561601,0.03647329],"genre_scores_gemma":[0.01928845,0.012169848,0.8327802,0.0012271246,0.0017504896,0.00043078783,0.0024161204,0.0025329504,0.12740399],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99878806,0.000261031,0.000052320087,0.00033992532,0.0005241351,0.000034654196],"domain_scores_gemma":[0.9989863,0.00040821198,0.000035305893,0.0002920884,0.00024420986,0.000033875527],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00086005207,0.0014556715,0.001740233,0.0021907475,0.0008739871,0.0028185865,0.0024130186,0.0015709476,0.021316499],"category_scores_gemma":[0.0028800743,0.00097377895,0.0010204123,0.0039293277,0.0015684621,0.003472474,0.0021717004,0.0026269725,0.025212629],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017868811,0.000043560427,0.00014841814,0.000617217,0.000069849695,0.000046244768,0.00018765211,0.009792184,0.0018306591,0.11477351,0.18057781,0.69189507],"study_design_scores_gemma":[0.000010093951,0.000028172823,0.00064361945,0.0002525145,0.0000372032,0.00055616087,0.000110409455,0.0605303,0.0037813766,0.46368125,0.47029445,0.00007443646],"about_ca_topic_score_codex":0.0010419799,"about_ca_topic_score_gemma":0.001977012,"teacher_disagreement_score":0.021316499,"about_ca_system_score_codex":0.00086265785,"about_ca_system_score_gemma":0.00093363575,"threshold_uncertainty_score":0.07131082},"labels":[],"label_agreement":null},{"id":"W2148260866","doi":"10.4236/jsea.2014.78059","title":"D-IMPACT: A Data Preprocessing Algorithm to Improve the Performance of Clustering","year":2014,"lang":"en","type":"article","venue":"Journal of Software Engineering and Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; University of Tokyo; Institute of Medical Science, University of Tokyo; Research Organization of Information and Systems","keywords":"Cluster analysis; Computer science; Preprocessor; Data mining; Outlier; Data pre-processing; CURE data clustering algorithm; Noise (video); Algorithm; Canopy clustering algorithm; Correlation clustering; Pattern recognition (psychology); Artificial intelligence; Image (mathematics)","score_opus":0.013876194593148983,"score_gpt":0.2891263948725445,"score_spread":0.2752502002793955,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148260866","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0076201498,0.00031584446,0.985873,0.00018423886,0.00016513852,0.00021671265,0.00027020168,0.004459593,0.00089510035],"genre_scores_gemma":[0.0538119,0.00026907667,0.9417221,0.00017149874,0.00008851502,0.0003167941,0.0017555584,0.00072492973,0.0011396363],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960175,0.0005870814,0.00042368294,0.0009289716,0.0017416659,0.00030106306],"domain_scores_gemma":[0.992239,0.001925514,0.00048499834,0.0014320454,0.0036655185,0.0002529524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042669396,0.002350806,0.0020275754,0.007405015,0.0028394582,0.0026288668,0.004073218,0.00184546,0.0025986035],"category_scores_gemma":[0.015204263,0.00090997067,0.0020505239,0.007863746,0.0011119616,0.0034440872,0.0029473985,0.0028086237,0.0030965158],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000668858,0.00038564412,0.00872312,0.00063166604,0.00034070387,0.00019760434,0.00062801706,0.06885713,0.028397007,0.010293312,0.02080225,0.86007464],"study_design_scores_gemma":[0.00013502347,0.00032000602,0.0053882343,0.000078865676,0.00015288962,0.0004692323,0.0003780936,0.8883071,0.061032623,0.013553862,0.029979032,0.0002050323],"about_ca_topic_score_codex":0.006554496,"about_ca_topic_score_gemma":0.007236664,"teacher_disagreement_score":0.007405015,"about_ca_system_score_codex":0.0013527748,"about_ca_system_score_gemma":0.0029096112,"threshold_uncertainty_score":0.02256596},"labels":[],"label_agreement":null},{"id":"W2149433841","doi":"10.1109/wi.2003.1241276","title":"Incremental document clustering using cluster similarity histograms","year":2004,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Document clustering; Fuzzy clustering; Data mining; Information retrieval; Brown clustering; Conceptual clustering; Correlation clustering; Histogram; Consensus clustering; Similarity measure; Similarity (geometry); CURE data clustering algorithm; Artificial intelligence; Image (mathematics)","score_opus":0.03597026134903813,"score_gpt":0.32169177144347144,"score_spread":0.2857215100944333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2149433841","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013483188,0.0002937399,0.9828205,0.000062347644,0.000042025466,0.00011971118,0.00022143699,0.002164465,0.00079262134],"genre_scores_gemma":[0.18526582,0.00043955917,0.8103555,0.00005862902,0.000091960945,0.00023684843,0.001579177,0.00028262596,0.0016898825],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99879575,0.00019144632,0.00006439257,0.00027605786,0.0005875005,0.000084804036],"domain_scores_gemma":[0.9974897,0.0009636463,0.00020432303,0.00037638217,0.0008726912,0.00009333035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009997074,0.0005681967,0.0010505309,0.0044191293,0.00080632104,0.0017558794,0.0019335635,0.0007316202,0.0016030052],"category_scores_gemma":[0.005565453,0.00039812995,0.0007442594,0.0051321634,0.00047208183,0.0020650604,0.0012358616,0.00079378014,0.0011647863],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024234976,0.00013057976,0.0021398861,0.00019220726,0.00014821709,0.00011603619,0.0002426502,0.13646491,0.013978362,0.013731516,0.0067768907,0.8258365],"study_design_scores_gemma":[0.00004180888,0.000079940226,0.001955879,0.000015361424,0.0000603569,0.00020798293,0.00010186523,0.9615159,0.01178829,0.018432437,0.005727358,0.000072805495],"about_ca_topic_score_codex":0.0080055995,"about_ca_topic_score_gemma":0.0055647637,"teacher_disagreement_score":0.0080055995,"about_ca_system_score_codex":0.001050616,"about_ca_system_score_gemma":0.0013131048,"threshold_uncertainty_score":0.015917957},"labels":[],"label_agreement":null},{"id":"W2150062453","doi":"10.1186/1751-0473-9-5","title":"PFClust: an optimised implementation of a parameter-free clustering algorithm","year":2014,"lang":"en","type":"article","venue":"Source Code for Biology and Medicine","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"World Anti-Doping Agency","keywords":"Computer science; Cluster analysis; Data mining; Algorithm; Data science; Artificial intelligence","score_opus":0.031641960745721155,"score_gpt":0.38136154057600685,"score_spread":0.3497195798302857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150062453","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008077767,0.0002866008,0.9068923,0.0002067559,0.0001847165,0.00022884046,0.0016674772,0.07902111,0.003434341],"genre_scores_gemma":[0.03317592,0.0001519246,0.9501164,0.00020238067,0.000042525313,0.00057258125,0.00337213,0.00970684,0.002659324],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987029,0.0002572054,0.0001130341,0.00032727813,0.0004676455,0.00013194748],"domain_scores_gemma":[0.99823654,0.0008008821,0.00012158891,0.000261548,0.00048815858,0.00009128361],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021101164,0.0017593438,0.0014819532,0.0012544416,0.0010114956,0.001445064,0.004763534,0.0021248064,0.021948371],"category_scores_gemma":[0.007515911,0.0012593649,0.001710112,0.0017005823,0.00089888746,0.001710484,0.0015566146,0.0022593676,0.010780102],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023538056,0.00032162186,0.002978436,0.0013740617,0.0005320372,0.00069992186,0.0006417003,0.23934214,0.027260514,0.017473808,0.13760483,0.5694171],"study_design_scores_gemma":[0.0005043681,0.00013732084,0.00074616197,0.00006457734,0.000064279215,0.00031170092,0.000054280816,0.9331689,0.015567852,0.010357566,0.038933035,0.000089899055],"about_ca_topic_score_codex":0.006631343,"about_ca_topic_score_gemma":0.0060971775,"teacher_disagreement_score":0.021948371,"about_ca_system_score_codex":0.0012879444,"about_ca_system_score_gemma":0.0025553838,"threshold_uncertainty_score":0.07342458},"labels":[],"label_agreement":null},{"id":"W2155874464","doi":"10.1109/icpr.2008.4761242","title":"A novel validity measure for clusters of arbitrary shapes and densities","year":2008,"lang":"en","type":"article","venue":"Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Measure (data warehouse); Cluster analysis; Computer science; Cluster (spacecraft); Neighbourhood (mathematics); Mathematics; Algorithm; Data mining; Statistical physics; Artificial intelligence; Physics; Mathematical analysis","score_opus":0.19587218909234413,"score_gpt":0.32822098272933664,"score_spread":0.1323487936369925,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2155874464","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.061233163,0.0016517439,0.92765194,0.00043590323,0.00019727908,0.00032514875,0.0006714522,0.0006579928,0.0071752896],"genre_scores_gemma":[0.5401797,0.00060888234,0.45472935,0.00025375639,0.00035394894,0.0007342636,0.0015836349,0.00028841893,0.0012679945],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9796505,0.0053629545,0.0019196734,0.0023314483,0.009996975,0.0007383964],"domain_scores_gemma":[0.93432075,0.04005252,0.005655161,0.0050451,0.013383991,0.0015425121],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014380659,0.0012908427,0.0021341832,0.012518196,0.0022840633,0.0055297627,0.0029671395,0.002566425,0.0018237828],"category_scores_gemma":[0.08054037,0.00048255385,0.0017691565,0.0061250594,0.0041674254,0.008605648,0.004448221,0.0018381105,0.0006706377],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013486164,0.00042190187,0.07077503,0.0016325155,0.0009941419,0.0005841871,0.0024220361,0.11807131,0.018595304,0.2448669,0.010629637,0.5296585],"study_design_scores_gemma":[0.00013831377,0.0008614108,0.022550803,0.0005216345,0.00028501492,0.0015973601,0.0012875483,0.74515,0.018910224,0.1939777,0.014314964,0.0004049527],"about_ca_topic_score_codex":0.0017030565,"about_ca_topic_score_gemma":0.00092835876,"teacher_disagreement_score":0.014380659,"about_ca_system_score_codex":0.002854712,"about_ca_system_score_gemma":0.0023051098,"threshold_uncertainty_score":0.07605314},"labels":[],"label_agreement":null},{"id":"W2155880472","doi":"10.1109/ideas.2006.52","title":"Visualization of Web Usage Patterns","year":2006,"lang":"en","type":"article","venue":"Proceedings - International Database Engineering and Applications Symposium","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada; Concordia University","keywords":"Computer science; Cluster analysis; Data mining; Visualization; Multidimensional scaling; Rendering (computer graphics); Fidelity; Data visualization; Fuzzy logic; Fuzzy clustering; Relational database; Web mining; Machine learning; Artificial intelligence; Web page; World Wide Web","score_opus":0.007642618741154016,"score_gpt":0.25967164109081486,"score_spread":0.25202902234966085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2155880472","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.089061804,0.0006598233,0.8824657,0.0009936613,0.0001219833,0.00018926866,0.003761084,0.016132854,0.0066138385],"genre_scores_gemma":[0.41068614,0.00066993956,0.5821703,0.000118639546,0.00010882825,0.00021564747,0.0026223382,0.0009218608,0.002486309],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996717,0.000071787035,0.000032821274,0.000055394325,0.00013610852,0.00003232133],"domain_scores_gemma":[0.9982139,0.000674206,0.00023912104,0.00030556117,0.00045956334,0.000107735905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000658629,0.0007347734,0.0004169988,0.0054773213,0.00031991437,0.0022420369,0.00056590803,0.0004601175,0.003732905],"category_scores_gemma":[0.0033095353,0.00030466326,0.0004329543,0.0031778263,0.00021254581,0.0016164968,0.001083438,0.00089186826,0.00082605926],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00054000906,0.00021788468,0.01626654,0.0007123427,0.00018323895,0.0007339701,0.002794888,0.027369805,0.08792694,0.019317916,0.021340283,0.8225962],"study_design_scores_gemma":[0.000120271005,0.0002879865,0.042009823,0.00029528947,0.00015125165,0.0020388777,0.0022548696,0.73122126,0.0845424,0.05706593,0.07978776,0.00022428048],"about_ca_topic_score_codex":0.0014550921,"about_ca_topic_score_gemma":0.0016601852,"teacher_disagreement_score":0.0054773213,"about_ca_system_score_codex":0.0002277858,"about_ca_system_score_gemma":0.0003801655,"threshold_uncertainty_score":0.012487888},"labels":[],"label_agreement":null},{"id":"W2161173962","doi":"","title":"Towards Property-Based Classification of Clustering Paradigms","year":2010,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Constrained clustering; Correlation clustering; Impossibility; CURE data clustering algorithm; Data stream clustering; Canopy clustering algorithm; Data mining; Set (abstract data type); Consensus clustering; Fuzzy clustering; Conceptual clustering; Machine learning; Artificial intelligence; Theoretical computer science","score_opus":0.04395941345548841,"score_gpt":0.31657841053949193,"score_spread":0.2726189970840035,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161173962","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010243278,0.0006034542,0.979879,0.0025191123,0.00013916529,0.00010969675,0.00013915481,0.00029739802,0.0060697407],"genre_scores_gemma":[0.29596668,0.0019396397,0.6944435,0.001629216,0.00093770964,0.0008578159,0.0008077366,0.00042063784,0.0029970545],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9869414,0.0044519515,0.0009817523,0.0030657959,0.0038502424,0.0007088],"domain_scores_gemma":[0.9545797,0.021057807,0.0039685355,0.012537831,0.006595188,0.0012609892],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013517173,0.0014563916,0.0014017025,0.004559965,0.0025776657,0.009706125,0.0048640645,0.00557778,0.0035025314],"category_scores_gemma":[0.05339974,0.001076272,0.0031642052,0.0036535766,0.010521026,0.022203632,0.0065945336,0.009809836,0.0012570639],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019032344,0.000022663111,0.00043294101,0.00008098414,0.00001835959,0.000036024165,0.00016946495,0.005288686,0.00049147574,0.98529184,0.0006352215,0.0075132847],"study_design_scores_gemma":[0.0000110000765,0.000024286534,0.00019164094,0.000038259604,0.000008665563,0.00007140231,0.00006295508,0.060212024,0.00048827977,0.9354446,0.0034273048,0.00001960753],"about_ca_topic_score_codex":0.0008520883,"about_ca_topic_score_gemma":0.00050627434,"teacher_disagreement_score":0.013517173,"about_ca_system_score_codex":0.0042821732,"about_ca_system_score_gemma":0.0021615906,"threshold_uncertainty_score":0.07148647},"labels":[],"label_agreement":null},{"id":"W2161806858","doi":"10.1109/icsmc.2006.384571","title":"GRIDBSCAN: GRId Density-Based Spatial Clustering of Applications with Noise","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Innovation Council; National Research Council Canada; Simon Fraser University","funders":"","keywords":"DBSCAN; Cluster analysis; Computer science; Noise (video); Grid; Data mining; Benchmark (surveying); Set (abstract data type); Data set; Artificial intelligence; Pattern recognition (psychology); Correlation clustering; CURE data clustering algorithm; Image (mathematics); Mathematics","score_opus":0.00793827355241549,"score_gpt":0.24188464356674572,"score_spread":0.23394637001433022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161806858","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008476779,0.00030265888,0.9651187,0.00020518764,0.00007515899,0.00025512304,0.0008317244,0.022428093,0.002306537],"genre_scores_gemma":[0.14617752,0.00048877735,0.8422052,0.00022122335,0.000047499485,0.0006785296,0.0048692957,0.001695817,0.0036160767],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99803215,0.00035401937,0.00013813368,0.00033370176,0.0009813373,0.00016073592],"domain_scores_gemma":[0.9983765,0.00035301663,0.00014377387,0.00032910664,0.0007136996,0.00008391022],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013878372,0.0017636124,0.0015798889,0.002783326,0.0013978366,0.0021268264,0.0036977213,0.0012340418,0.003654566],"category_scores_gemma":[0.006118607,0.0009039565,0.0015865247,0.0038270587,0.00085636775,0.0016902802,0.0025940137,0.0013519047,0.002345431],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052892667,0.00018131944,0.005912362,0.0006153556,0.00028712425,0.0003819999,0.0005874755,0.44213971,0.0078367805,0.01989974,0.048414312,0.47321486],"study_design_scores_gemma":[0.00002868823,0.0000340484,0.0006908529,0.000018640587,0.000019056812,0.000079616075,0.00007339873,0.97339976,0.0053956364,0.006360181,0.013865481,0.00003458947],"about_ca_topic_score_codex":0.031656355,"about_ca_topic_score_gemma":0.029679596,"teacher_disagreement_score":0.031656355,"about_ca_system_score_codex":0.0017285016,"about_ca_system_score_gemma":0.0033691206,"threshold_uncertainty_score":0.062944174},"labels":[],"label_agreement":null},{"id":"W2162062774","doi":"10.1109/isvd.2006.18","title":"CRYSTAL - A new density-based fast and efficient clustering algorithm","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"University of Alberta","keywords":"Delaunay triangulation; Cluster analysis; CURE data clustering algorithm; Correlation clustering; Computer science; Canopy clustering algorithm; Bowyer–Watson algorithm; Fuzzy clustering; DBSCAN; Single-linkage clustering; Algorithm; Regular polygon; Cluster (spacecraft); Voronoi diagram; Data mining; Artificial intelligence; Mathematics","score_opus":0.011153836378406603,"score_gpt":0.25285739249769534,"score_spread":0.24170355611928873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162062774","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015601781,0.0002498115,0.99518967,0.00009857793,0.0000737061,0.00011143255,0.0001984614,0.0016466858,0.00087143027],"genre_scores_gemma":[0.014312128,0.00021436338,0.9817193,0.00009523509,0.000043521166,0.00029520286,0.0009989613,0.00044409963,0.00187711],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971117,0.0003962895,0.00015680543,0.00060868467,0.0015549223,0.00017168315],"domain_scores_gemma":[0.9967217,0.0006819733,0.0002691849,0.0005577494,0.0016122254,0.00015711585],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015316352,0.0016106082,0.0027048576,0.0051293517,0.0027125361,0.0025755027,0.004757999,0.002126097,0.0050820136],"category_scores_gemma":[0.0069221053,0.0015858443,0.0017050048,0.007119086,0.0011921126,0.0033856137,0.0039405506,0.002441155,0.0055961865],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029129407,0.00016334273,0.0019136143,0.0005910398,0.00032345994,0.00018672855,0.00035924933,0.21073477,0.015192746,0.045948554,0.043487016,0.6808081],"study_design_scores_gemma":[0.00008087018,0.000043637967,0.00043826,0.000045950826,0.000035351473,0.00024910885,0.00009523912,0.93572325,0.0067464844,0.025755227,0.03070213,0.00008459554],"about_ca_topic_score_codex":0.012301562,"about_ca_topic_score_gemma":0.012720821,"teacher_disagreement_score":0.012301562,"about_ca_system_score_codex":0.0018915307,"about_ca_system_score_gemma":0.004132981,"threshold_uncertainty_score":0.024459898},"labels":[],"label_agreement":null},{"id":"W2163067442","doi":"10.1109/icdm.2009.118","title":"A New MCA-Based Divisive Hierarchical Algorithm for Clustering Categorical Data","year":2009,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Similarity (geometry); Hierarchical clustering; Data mining; Pattern recognition (psychology); Algorithm; Canopy clustering algorithm; Single-linkage clustering; Subspace topology; Artificial intelligence; Linear subspace; CURE data clustering algorithm; Correlation clustering; Mathematics; Machine learning","score_opus":0.05942537267736089,"score_gpt":0.3551315825387727,"score_spread":0.29570620986141183,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163067442","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001774337,0.00023422397,0.99628973,0.00012614606,0.00006089365,0.00010041572,0.0001296212,0.00063662184,0.0006480192],"genre_scores_gemma":[0.023267906,0.00013026998,0.97436994,0.00011944844,0.00008131837,0.00025268432,0.00042406973,0.00015454255,0.0011998521],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965888,0.0007815353,0.00020209164,0.000878889,0.0013820853,0.00016660824],"domain_scores_gemma":[0.99680376,0.0010340586,0.00024078277,0.0004807247,0.0012867792,0.00015390181],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028440712,0.0010890814,0.0017056379,0.004326705,0.0021231358,0.002008655,0.0035303044,0.0015647752,0.002645461],"category_scores_gemma":[0.008628754,0.00070272095,0.001709836,0.007883797,0.0011180361,0.002544518,0.0019487641,0.0024631992,0.0017094088],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016306524,0.0001174293,0.0022466858,0.00027887343,0.00029247938,0.0001249478,0.00059437874,0.10821609,0.011413471,0.030597845,0.014841279,0.8311135],"study_design_scores_gemma":[0.000035382684,0.000052703148,0.00073971774,0.0000310579,0.00004082492,0.00020472312,0.00010966935,0.9433341,0.0053307083,0.032596827,0.017456375,0.000067882575],"about_ca_topic_score_codex":0.013089276,"about_ca_topic_score_gemma":0.015518763,"teacher_disagreement_score":0.013089276,"about_ca_system_score_codex":0.0017292992,"about_ca_system_score_gemma":0.003617902,"threshold_uncertainty_score":0.02602619},"labels":[],"label_agreement":null},{"id":"W2163097311","doi":"","title":"Proximity Graphs for Clustering and Manifold Learning","year":2004,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":120,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster analysis; Lattice graph; Dimensionality reduction; Computer science; Line graph; Mathematics; Graph; Pattern recognition (psychology); Voltage graph; Theoretical computer science; Artificial intelligence","score_opus":0.022688260320615685,"score_gpt":0.28182495072565483,"score_spread":0.25913669040503917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163097311","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002691486,0.0057069985,0.9813973,0.0012587173,0.0002403929,0.000091053524,0.0006579313,0.00081368163,0.007142448],"genre_scores_gemma":[0.16527304,0.0116236685,0.8044259,0.0010832947,0.001437319,0.0009134352,0.0037094918,0.000789062,0.010744755],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997566,0.001100298,0.0001153402,0.00059467304,0.0005298595,0.00009381171],"domain_scores_gemma":[0.99629647,0.0019429605,0.00039476273,0.00076705735,0.0004853928,0.00011335777],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017390795,0.0014824627,0.0017241752,0.0047938726,0.0014323866,0.0025442396,0.002152406,0.0022275373,0.0069081155],"category_scores_gemma":[0.0102829905,0.00066516077,0.00151673,0.0065422584,0.002750549,0.0044063525,0.0034181937,0.0034019414,0.0033705155],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003696386,0.000027032504,0.00038559487,0.00036736744,0.00008100349,0.00012319535,0.00025049894,0.051991325,0.0007312233,0.8337422,0.013694811,0.098568715],"study_design_scores_gemma":[0.000008880733,0.000015062169,0.00022569299,0.000054045056,0.000015090221,0.0001292093,0.00006400668,0.08582034,0.00033323496,0.879816,0.03349651,0.000021997723],"about_ca_topic_score_codex":0.0033616386,"about_ca_topic_score_gemma":0.0024667424,"teacher_disagreement_score":0.0069081155,"about_ca_system_score_codex":0.0017909842,"about_ca_system_score_gemma":0.00084526517,"threshold_uncertainty_score":0.023109972},"labels":[],"label_agreement":null},{"id":"W2164582339","doi":"10.1109/ijcnn.2006.247041","title":"PARTCAT: A Subspace Clustering Algorithm for High Dimensional Categorical Data","year":2006,"lang":"en","type":"article","venue":"The 2006 IEEE International Joint Conference on Neural Network Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Subspace topology; Data mining; Dimension (graph theory); Similarity (geometry); Clustering high-dimensional data; Artificial intelligence; Artificial neural network; Cluster (spacecraft); Pattern recognition (psychology); Algorithm; Mathematics; Machine learning","score_opus":0.08700548172302576,"score_gpt":0.3200420799469654,"score_spread":0.23303659822393963,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164582339","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026203897,0.00010034222,0.9952122,0.00006898131,0.00006902388,0.000083337334,0.00022366701,0.0011507218,0.0004713313],"genre_scores_gemma":[0.052662384,0.00020225348,0.9410489,0.00016958662,0.00008145364,0.0003997383,0.0022608703,0.00036960474,0.002805149],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99800104,0.00053419516,0.00013288866,0.00042737206,0.0007688428,0.0001356894],"domain_scores_gemma":[0.9972947,0.0004900104,0.00015215992,0.00069534325,0.0012142422,0.00015351208],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017113914,0.0014244093,0.0020038644,0.003046576,0.0017898543,0.0021024554,0.0025626656,0.0016448967,0.0045768395],"category_scores_gemma":[0.005865341,0.000669323,0.0019400923,0.005330152,0.0009726528,0.0032827272,0.0026178728,0.0022429428,0.0032051953],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031809212,0.00018058917,0.0022057493,0.00023115733,0.00035453957,0.0001184936,0.0002685559,0.13049655,0.011467299,0.028406847,0.025144782,0.8008074],"study_design_scores_gemma":[0.00003381917,0.00011290283,0.001023768,0.00002282879,0.000036681493,0.00020561829,0.00008129395,0.947122,0.004865467,0.03588368,0.010527435,0.000084492596],"about_ca_topic_score_codex":0.006587534,"about_ca_topic_score_gemma":0.008067107,"teacher_disagreement_score":0.006587534,"about_ca_system_score_codex":0.000864511,"about_ca_system_score_gemma":0.0026665016,"threshold_uncertainty_score":0.015311003},"labels":[],"label_agreement":null},{"id":"W2164665696","doi":"10.1109/ijcnn.2006.246962","title":"Refining Spherical K-Means for Clustering Documents","year":2006,"lang":"en","type":"article","venue":"The 2006 IEEE International Joint Conference on Neural Network Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Convexity; Cluster analysis; Mathematical optimization; Constraint (computer-aided design); Set (abstract data type); Computer science; Maximization; Algorithm; Linear programming; Mathematics; Function (biology); Optimization problem; Artificial intelligence","score_opus":0.06047978020347221,"score_gpt":0.318466311927053,"score_spread":0.25798653172358077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164665696","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022348964,0.000097657874,0.9971603,0.000042500767,0.000010101809,0.000017331118,0.000023580324,0.00012694315,0.0002866741],"genre_scores_gemma":[0.09857825,0.00046227235,0.89841205,0.00009794325,0.000051412084,0.00015086084,0.00043785307,0.00028930648,0.0015200528],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99694234,0.0012487528,0.00020433891,0.00055041915,0.000892537,0.00016167808],"domain_scores_gemma":[0.9976076,0.0008458296,0.00020916582,0.0005454546,0.0007191937,0.00007266495],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029301543,0.0011914093,0.0014749722,0.0016184826,0.00085911265,0.0015950135,0.0021185966,0.001580626,0.0014867017],"category_scores_gemma":[0.0100784,0.0007767088,0.0017314035,0.0028608225,0.0013414135,0.0026288296,0.0020456978,0.0019858368,0.0017717715],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014221591,0.00004830197,0.0006750656,0.00023410012,0.00009328615,0.00005921,0.00021850811,0.78665257,0.011003809,0.059474263,0.0033472287,0.13805138],"study_design_scores_gemma":[0.00000829499,0.000023441642,0.00009578014,0.0000067049054,0.000007839583,0.00002958735,0.000023547984,0.98185235,0.0025201684,0.013597857,0.0018165471,0.000017897319],"about_ca_topic_score_codex":0.009426391,"about_ca_topic_score_gemma":0.010029726,"teacher_disagreement_score":0.009426391,"about_ca_system_score_codex":0.0013698832,"about_ca_system_score_gemma":0.0018801257,"threshold_uncertainty_score":0.018743038},"labels":[],"label_agreement":null},{"id":"W2165944636","doi":"10.1109/icsmc.2007.4413785","title":"Image segmentation using joint clustering analysis of attribute data and relationship data","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Cluster analysis; Segmentation-based object categorization; Computer science; Image segmentation; Segmentation; Artificial intelligence; Pattern recognition (psychology); Scale-space segmentation; Data mining; Market segmentation","score_opus":0.2598599314233456,"score_gpt":0.4275597758651129,"score_spread":0.16769984444176728,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165944636","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028713888,0.00018763838,0.96899736,0.000093947005,0.000026055046,0.000066988745,0.00009591082,0.000770405,0.0010479046],"genre_scores_gemma":[0.23040523,0.00022366675,0.76635396,0.00005160437,0.000052652995,0.00011418115,0.00081481406,0.00021176625,0.0017721158],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99755317,0.0004997657,0.00016005806,0.0006699046,0.00090077193,0.00021626713],"domain_scores_gemma":[0.9975345,0.00064600323,0.00027486339,0.00055336684,0.00089077605,0.00010052401],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022860267,0.00087288395,0.0015406323,0.005078182,0.0009741542,0.002510034,0.0013698114,0.0013318198,0.0010302926],"category_scores_gemma":[0.00459234,0.0005999557,0.0020751434,0.005094172,0.00090991124,0.002647403,0.0014308256,0.0010839636,0.0010028204],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046697663,0.00031672313,0.0090084495,0.00023673894,0.0005239198,0.00017422355,0.00080920197,0.118337035,0.065528855,0.016715022,0.0035670095,0.78431576],"study_design_scores_gemma":[0.000020622301,0.000118922064,0.0059192795,0.000019474946,0.00011778152,0.00021086163,0.00023068466,0.94932383,0.024811313,0.014921918,0.004236839,0.00006848973],"about_ca_topic_score_codex":0.004498115,"about_ca_topic_score_gemma":0.0045415503,"teacher_disagreement_score":0.005078182,"about_ca_system_score_codex":0.001021512,"about_ca_system_score_gemma":0.0010832698,"threshold_uncertainty_score":0.0120898485},"labels":[],"label_agreement":null},{"id":"W2169629580","doi":"10.1109/ictai.2008.128","title":"Oracle Clustering: Dynamic Partitioning Based on Random Observations","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"","keywords":"Cluster analysis; Computer science; Dynamism; Oracle; Stability (learning theory); Data mining; Function (biology); Sampling (signal processing); Algorithm; Artificial intelligence; Machine learning","score_opus":0.060515682179626516,"score_gpt":0.3047927912108204,"score_spread":0.2442771090311939,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2169629580","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042699534,0.00021009018,0.9943627,0.00006141217,0.00003060144,0.0000489697,0.000049220474,0.00061710377,0.00034993517],"genre_scores_gemma":[0.2178556,0.00048736387,0.7776068,0.00017721848,0.00016226292,0.00024790017,0.0010902357,0.00040441178,0.0019682606],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976586,0.0007065453,0.000108776105,0.00067368604,0.0006836698,0.00016877355],"domain_scores_gemma":[0.9980161,0.0008348416,0.00022802128,0.00042376574,0.00038564982,0.00011157333],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00198628,0.0010530946,0.0018516787,0.001948432,0.0007931518,0.0016139076,0.002726131,0.0011700965,0.0014742939],"category_scores_gemma":[0.0057410174,0.0005802885,0.0008185964,0.0017778794,0.0010774656,0.002521741,0.001693072,0.0012613775,0.00094950525],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006803223,0.00013758511,0.0035547821,0.00028908235,0.00020647119,0.00016762155,0.00034443729,0.43855965,0.012671269,0.043700565,0.0066259615,0.49306226],"study_design_scores_gemma":[0.000021501639,0.00006264663,0.0005706024,0.000013839316,0.000017684692,0.00012612749,0.000029625691,0.98166406,0.0030372848,0.01155178,0.0028762456,0.000028633125],"about_ca_topic_score_codex":0.0035165092,"about_ca_topic_score_gemma":0.0030401538,"teacher_disagreement_score":0.0035165092,"about_ca_system_score_codex":0.0008733852,"about_ca_system_score_gemma":0.000973101,"threshold_uncertainty_score":0.010504603},"labels":[],"label_agreement":null},{"id":"W2170006979","doi":"10.1007/s10107-010-0349-7","title":"An improved column generation algorithm for minimum sum-of-squares clustering","year":2010,"lang":"en","type":"article","venue":"Mathematical Programming","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":99,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Group for Research in Decision Analysis; HEC Montréal","funders":"","keywords":"Mathematics; Column (typography); Cluster analysis; Explained sum of squares; Bottleneck; Algorithm; Column generation; Set (abstract data type); Centroid; Point (geometry); Combinatorics; Square (algebra); Euclidean space; Mathematical optimization; Computer science; Statistics; Geometry","score_opus":0.03302216249232391,"score_gpt":0.33510104580939204,"score_spread":0.3020788833170681,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170006979","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002446406,0.00010286463,0.99561584,0.000085379164,0.000089536086,0.00005068462,0.00007600191,0.0009261004,0.0006072658],"genre_scores_gemma":[0.024196478,0.000070656555,0.9729873,0.00012250869,0.00006430586,0.00015566954,0.00043215076,0.00027366437,0.0016972758],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998771,0.0003660323,0.000069495145,0.00020406737,0.0004929037,0.00009645069],"domain_scores_gemma":[0.9978067,0.000796723,0.000086723434,0.00030544557,0.0009109185,0.00009346567],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011596014,0.0013974,0.00162508,0.0014646518,0.0010075124,0.0011973989,0.0023563437,0.0014822658,0.0066803824],"category_scores_gemma":[0.0042822566,0.00075826055,0.0011211177,0.0022480781,0.00057506183,0.0013053338,0.0016525398,0.0019461911,0.003642429],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041767547,0.0002183435,0.00056996994,0.0002733566,0.00013238478,0.00013674302,0.00015188794,0.23885815,0.018860107,0.014912972,0.021306999,0.70416147],"study_design_scores_gemma":[0.00005099477,0.00005515674,0.00017217253,0.000008438884,0.00002013929,0.00008388034,0.000017319924,0.9870062,0.004079745,0.0054138014,0.0030675498,0.000024566858],"about_ca_topic_score_codex":0.0041980976,"about_ca_topic_score_gemma":0.006472728,"teacher_disagreement_score":0.0066803824,"about_ca_system_score_codex":0.0005814551,"about_ca_system_score_gemma":0.0018345201,"threshold_uncertainty_score":0.022348046},"labels":[],"label_agreement":null},{"id":"W2170896428","doi":"10.1109/icdm.2009.143","title":"Semi-supervised Density-Based Clustering","year":2009,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":100,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence; Set (abstract data type); Pattern recognition (psychology); Cluster (spacecraft); Data mining; Noise (video); Feature (linguistics); Image (mathematics)","score_opus":0.022955954467767742,"score_gpt":0.2907169793540355,"score_spread":0.26776102488626774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170896428","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01033242,0.00015608974,0.98587763,0.00006387182,0.000018721616,0.00014444218,0.00016141421,0.0012515392,0.0019938913],"genre_scores_gemma":[0.30554217,0.00027365002,0.688697,0.00010704777,0.000064220614,0.00039665267,0.0015851278,0.00035339769,0.0029808348],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959282,0.001230485,0.000237686,0.0009117886,0.0015099895,0.00018184974],"domain_scores_gemma":[0.99171996,0.0027590115,0.00066837075,0.0014370037,0.0032570201,0.00015859459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003218754,0.0011539585,0.0017337844,0.0037773708,0.0013163923,0.0020400467,0.0031620928,0.00153009,0.0030351328],"category_scores_gemma":[0.012671683,0.0007026571,0.0013246313,0.0023352548,0.0013883574,0.0021592479,0.0021120594,0.0010932642,0.002587748],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042880647,0.0002715062,0.0073117637,0.0007989884,0.00031626652,0.00024181815,0.0011156644,0.33792818,0.015272713,0.0271291,0.011724422,0.59746075],"study_design_scores_gemma":[0.000014435625,0.000027549364,0.0010712158,0.00003769795,0.000016717295,0.00017826867,0.00012190209,0.97897696,0.004680795,0.012080084,0.0027629666,0.000031361884],"about_ca_topic_score_codex":0.0047892975,"about_ca_topic_score_gemma":0.0062321746,"teacher_disagreement_score":0.0047892975,"about_ca_system_score_codex":0.0011543994,"about_ca_system_score_gemma":0.0018666035,"threshold_uncertainty_score":0.01702261},"labels":[],"label_agreement":null},{"id":"W2174016580","doi":"","title":"Estimation and Selection in Regression Clustering","year":2011,"lang":"en","type":"article","venue":"European Journal of Pure and Applied Mathematics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Mathematics; Correlation clustering; CURE data clustering algorithm; Regression analysis; Regression; Single-linkage clustering; Clustering high-dimensional data; Data mining; Statistics; Artificial intelligence; Computer science","score_opus":0.03784475837396073,"score_gpt":0.2723698911089335,"score_spread":0.23452513273497277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2174016580","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019068675,0.0004077313,0.9964301,0.0001648023,0.000032244803,0.000033092485,0.000039458624,0.00019069202,0.00079511065],"genre_scores_gemma":[0.19036156,0.0025967392,0.79938316,0.00032552241,0.0004670671,0.0005270545,0.00082944543,0.00039583616,0.0051136003],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9933076,0.004033079,0.00020264281,0.0010040529,0.0012173734,0.00023527701],"domain_scores_gemma":[0.9890998,0.0071925325,0.0008367129,0.0010157641,0.00167238,0.00018271124],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007279413,0.001200627,0.0022720583,0.0029186793,0.0011891831,0.0018889739,0.002720403,0.0020114847,0.002632941],"category_scores_gemma":[0.030801306,0.00080523686,0.0016849915,0.0038247854,0.0018878604,0.0023558696,0.002305838,0.0020837896,0.0020051217],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012423945,0.00007633526,0.0023999016,0.00039218366,0.00022704009,0.00020644396,0.000260132,0.6203718,0.0020934623,0.168208,0.0066243983,0.19901602],"study_design_scores_gemma":[0.000015015288,0.000030261246,0.0005270228,0.000035397068,0.000019436156,0.00006386121,0.000032454867,0.9216482,0.0009998711,0.07254945,0.0040540486,0.00002499203],"about_ca_topic_score_codex":0.0039920546,"about_ca_topic_score_gemma":0.0024317398,"teacher_disagreement_score":0.007279413,"about_ca_system_score_codex":0.0016170061,"about_ca_system_score_gemma":0.0014878468,"threshold_uncertainty_score":0.038497686},"labels":[],"label_agreement":null},{"id":"W2185184584","doi":"","title":"A characterization of linkage-based hierarchical clustering","year":2016,"lang":"en","type":"article","venue":"Journal of Machine Learning Research","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Linkage (software); Hierarchical clustering of networks; Hierarchical clustering; Computer science; Class (philosophy); Cluster analysis; Complete linkage; Locality; Hierarchical database model; Mathematics; Artificial intelligence; Algorithm; Data mining; Canopy clustering algorithm; Correlation clustering","score_opus":0.04192133453023034,"score_gpt":0.3573990556538818,"score_spread":0.31547772112365147,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2185184584","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008078279,0.00039904908,0.9850124,0.00038335953,0.000034314515,0.00008463912,0.00015441643,0.0002731132,0.005580434],"genre_scores_gemma":[0.39553988,0.0012317589,0.5911429,0.0010167824,0.00050399173,0.00066767895,0.0013008008,0.0005275722,0.00806861],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9951126,0.0012607281,0.00022447263,0.0010436588,0.002071574,0.00028703397],"domain_scores_gemma":[0.9856655,0.005817783,0.0018264215,0.0025013196,0.0035807064,0.00060817716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004536599,0.000711996,0.00089122227,0.003547776,0.002222597,0.0031734717,0.002721125,0.0020261623,0.0043216357],"category_scores_gemma":[0.023411307,0.0005683326,0.0009671164,0.0038049226,0.0031712218,0.0055297106,0.0040653297,0.002317522,0.0018637681],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000840675,0.00008691362,0.0032289233,0.00028111046,0.00010818184,0.0001877139,0.0009206968,0.06947049,0.005950893,0.8064811,0.008557828,0.10464202],"study_design_scores_gemma":[0.000033508768,0.00013236301,0.0018386095,0.00009583523,0.00004501681,0.0005751984,0.00038141763,0.31308493,0.004435547,0.6485292,0.030782267,0.00006617315],"about_ca_topic_score_codex":0.0012510361,"about_ca_topic_score_gemma":0.0011288632,"teacher_disagreement_score":0.004536599,"about_ca_system_score_codex":0.0017985703,"about_ca_system_score_gemma":0.0017266654,"threshold_uncertainty_score":0.023992121},"labels":[],"label_agreement":null},{"id":"W2234615877","doi":"10.1080/10618600.2018.1546593","title":"A General Hybrid Clustering Technique","year":2018,"lang":"en","type":"preprint","venue":"Journal of Computational and Graphical Statistics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"Division of Mathematical Sciences; Defense Threat Reduction Agency","keywords":"Cluster analysis; Single-linkage clustering; Complete-linkage clustering; Dendrogram; Complete linkage; Correlation clustering; CURE data clustering algorithm; Series (stratigraphy); Regular polygon; Mathematics; k-medians clustering; Computer science; Joins; Hamming distance; Combinatorics; Data mining; Artificial intelligence; Geometry","score_opus":0.020955277286332633,"score_gpt":0.3180252457379465,"score_spread":0.2970699684516139,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2234615877","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011830915,0.00021328706,0.99537,0.00012938931,0.000087933775,0.00007220484,0.000127779,0.00062893884,0.0021873089],"genre_scores_gemma":[0.029387128,0.00035313875,0.95944524,0.00035710144,0.00017093473,0.00030716474,0.0006518535,0.00047653707,0.008850901],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9962698,0.0008281509,0.00017242353,0.0012352328,0.0012620088,0.00023233121],"domain_scores_gemma":[0.9973985,0.00062984956,0.00016026034,0.0009471321,0.0007520552,0.0001121712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026842246,0.0014745438,0.0016619278,0.004145661,0.0021045443,0.0025019643,0.004132249,0.0030596722,0.010447522],"category_scores_gemma":[0.005259081,0.00091406435,0.0028360363,0.004998524,0.0014780986,0.0030148418,0.004327227,0.0029973865,0.008916982],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002681467,0.00017054734,0.0022555368,0.0006158429,0.00062410545,0.00041114667,0.00090054143,0.098976046,0.033486072,0.22793344,0.035228007,0.5991305],"study_design_scores_gemma":[0.00008939945,0.00017260057,0.0017146631,0.000118422606,0.00018856557,0.0013441677,0.00033148,0.70768297,0.011993401,0.16560227,0.110597,0.00016511415],"about_ca_topic_score_codex":0.003092081,"about_ca_topic_score_gemma":0.0041752015,"teacher_disagreement_score":0.010447522,"about_ca_system_score_codex":0.0010770428,"about_ca_system_score_gemma":0.00140674,"threshold_uncertainty_score":0.034950495},"labels":[],"label_agreement":null},{"id":"W2237415976","doi":"","title":"Using Pivots to Speed-Up k-Medoids Clustering","year":2011,"lang":"en","type":"article","venue":"Cadernos de Linguística e Teoria da Literatura (Universidade Federal de Minas Gerais)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Medoid; Cluster analysis; Computer science; k-medoids; Data mining; Canopy clustering algorithm; Correlation clustering; CURE data clustering algorithm; Algorithm; Set (abstract data type); Artificial intelligence","score_opus":0.07390989392732218,"score_gpt":0.31530947885642335,"score_spread":0.24139958492910119,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2237415976","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0075911595,0.00078416354,0.9876094,0.00016832801,0.00025798127,0.00018475056,0.00016024105,0.0021870786,0.0010569744],"genre_scores_gemma":[0.057072543,0.00035528478,0.93934846,0.00012786922,0.000099123405,0.0002150932,0.0007786119,0.00044671836,0.0015563583],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99673295,0.0011510258,0.00033569147,0.0006957457,0.000833056,0.0002515737],"domain_scores_gemma":[0.99118584,0.004656603,0.00040498129,0.001351294,0.00207466,0.0003267139],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004921719,0.00213385,0.0034708886,0.00491431,0.0028471726,0.002744749,0.003888726,0.002879056,0.009150805],"category_scores_gemma":[0.022134185,0.0015803515,0.0024201286,0.0055543375,0.0015072447,0.0029740992,0.0044617713,0.0034989684,0.009383251],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012125643,0.00024352748,0.002818388,0.0007843031,0.0004851196,0.0002171567,0.001298913,0.24084872,0.0073878854,0.020011978,0.014525807,0.7101656],"study_design_scores_gemma":[0.00016220276,0.00017493482,0.0008069499,0.00011109803,0.00008733493,0.00020739145,0.0004965028,0.95156795,0.0061276364,0.028529175,0.011650005,0.00007893357],"about_ca_topic_score_codex":0.016104484,"about_ca_topic_score_gemma":0.022931712,"teacher_disagreement_score":0.016104484,"about_ca_system_score_codex":0.0012763519,"about_ca_system_score_gemma":0.0033549867,"threshold_uncertainty_score":0.032021523},"labels":[],"label_agreement":null},{"id":"W2270192120","doi":"10.1371/journal.pone.0144059","title":"A Comparison Study on Similarity and Dissimilarity Measures in Clustering Continuous Data","year":2015,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":408,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"Universiti Malaya","keywords":"Similarity (geometry); Cluster analysis; Computer science; Data mining; Benchmark (surveying); Distance measures; Similarity measure; Measure (data warehouse); Cluster (spacecraft); Pattern recognition (psychology); Artificial intelligence; Geography","score_opus":0.43152004598405463,"score_gpt":0.4023074041110184,"score_spread":0.029212641873036238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2270192120","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4486104,0.043214757,0.48831856,0.0016390567,0.0009752794,0.0005503662,0.0009025484,0.0005570646,0.0152319055],"genre_scores_gemma":[0.7592854,0.0055803084,0.23240362,0.00014035757,0.0002876577,0.00024097173,0.0010880061,0.000105022766,0.0008685301],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98033524,0.008104173,0.0020931084,0.0023807585,0.0066950535,0.00039175613],"domain_scores_gemma":[0.91955173,0.05883326,0.0032926213,0.0050297105,0.012283258,0.0010094392],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01797167,0.00089089153,0.0011962884,0.008430123,0.0010103418,0.0030404602,0.0013975857,0.0014546175,0.0011197905],"category_scores_gemma":[0.09682115,0.00030370158,0.0012412741,0.0106320325,0.0014880805,0.005809433,0.0017067232,0.001202843,0.00030544383],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021762268,0.00057169265,0.08183413,0.004355904,0.0021924628,0.00042260523,0.0024949755,0.056666903,0.011429341,0.06501206,0.005745935,0.76709783],"study_design_scores_gemma":[0.00035228892,0.006583017,0.14981754,0.0015852208,0.0016804399,0.0032301021,0.006019058,0.68544436,0.030977085,0.071633704,0.042134453,0.00054273667],"about_ca_topic_score_codex":0.0017728733,"about_ca_topic_score_gemma":0.0015281041,"teacher_disagreement_score":0.01797167,"about_ca_system_score_codex":0.0017193551,"about_ca_system_score_gemma":0.0010920449,"threshold_uncertainty_score":0.095044374},"labels":[],"label_agreement":null},{"id":"W2276374992","doi":"","title":"A novel evolutionary clustering algorithm based on Gaussian mixture model","year":2006,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Cluster analysis; Determining the number of clusters in a data set; Correlation clustering; CURE data clustering algorithm; Single-linkage clustering; Canopy clustering algorithm; Computer science; Fuzzy clustering; Benchmark (surveying); Evolutionary algorithm; Data mining; Algorithm; Entropy (arrow of time); Clustering high-dimensional data; Artificial intelligence","score_opus":0.01715317738749668,"score_gpt":0.2689564433357263,"score_spread":0.25180326594822966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2276374992","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011717116,0.00014744645,0.98668927,0.00010585881,0.000042302698,0.000048499307,0.000020459172,0.0003188378,0.00091012125],"genre_scores_gemma":[0.16448419,0.00017171404,0.83151674,0.000129535,0.00003910628,0.00023481989,0.00019998779,0.00013410549,0.0030897344],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988274,0.00024135283,0.00006447222,0.00030573583,0.00044879614,0.00011236383],"domain_scores_gemma":[0.99899083,0.00032728977,0.00009325836,0.00009844166,0.00043144546,0.000058717997],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018597238,0.0009826345,0.0015756434,0.002482528,0.0012142881,0.0012668853,0.0025557263,0.0021067495,0.0018670931],"category_scores_gemma":[0.0032831484,0.00069252326,0.0014011833,0.002114853,0.0007845661,0.0015289896,0.0013793211,0.001091515,0.0005875357],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014414491,0.00013578236,0.002386738,0.00008933336,0.00021749297,0.00012906178,0.0002036613,0.67178965,0.009379765,0.017267484,0.0026918326,0.29556507],"study_design_scores_gemma":[0.00001086696,0.000013234394,0.00015400095,0.000004941127,0.000010601832,0.000034345874,0.000008822408,0.9968135,0.0008184812,0.0015131921,0.0006095641,0.000008330986],"about_ca_topic_score_codex":0.007454791,"about_ca_topic_score_gemma":0.005076671,"teacher_disagreement_score":0.007454791,"about_ca_system_score_codex":0.0013748268,"about_ca_system_score_gemma":0.0014394191,"threshold_uncertainty_score":0.014822781},"labels":[],"label_agreement":null},{"id":"W2278576066","doi":"10.3166/isi.20.5.77-101","title":"Matérialisation de cubes de données par des partitions d’ensembles de n-uplets","year":2015,"lang":"fr","type":"article","venue":"Ingénierie des systèmes d information","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; Philosophy","score_opus":0.08654406474758126,"score_gpt":0.3013027100652112,"score_spread":0.21475864531762995,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2278576066","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12698446,0.0008932144,0.8635934,0.0009991027,0.00030678068,0.00023082012,0.0015269384,0.0015565276,0.0039086603],"genre_scores_gemma":[0.38267025,0.0012477493,0.598857,0.00037842494,0.00028994336,0.0007353723,0.0052311965,0.00074587215,0.009844179],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9927884,0.0014215225,0.00038922383,0.0012354766,0.00369669,0.00046870505],"domain_scores_gemma":[0.9870062,0.0089001665,0.00055051356,0.00096943247,0.0020399182,0.00053383736],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003561947,0.0034888461,0.0025035536,0.0045815087,0.0022709463,0.008600284,0.0020221937,0.0041845185,0.0070905313],"category_scores_gemma":[0.016355403,0.0019975614,0.005143309,0.0032321494,0.0034947162,0.0037148937,0.003158324,0.0049081063,0.001936914],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020945864,0.00041227505,0.006746754,0.0014840155,0.00090145477,0.0026212437,0.0022343246,0.6629183,0.06316581,0.062163465,0.009502198,0.18575563],"study_design_scores_gemma":[0.00012626825,0.00015730981,0.003527594,0.00011711283,0.00005972464,0.00062259496,0.00082618196,0.95140743,0.021557817,0.011384921,0.010138453,0.000074595235],"about_ca_topic_score_codex":0.03832358,"about_ca_topic_score_gemma":0.026867991,"teacher_disagreement_score":0.03832358,"about_ca_system_score_codex":0.0040013287,"about_ca_system_score_gemma":0.0026103908,"threshold_uncertainty_score":0.07620102},"labels":[],"label_agreement":null},{"id":"W2286286503","doi":"10.1007/978-3-319-49487-6_3","title":"Theoretical Analysis of the k-Means Algorithm – A Survey","year":2016,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Heuristics; Algorithm; Cluster analysis; Extension (predicate logic); Simplicity; k-means clustering; Artificial intelligence","score_opus":0.018668256590251276,"score_gpt":0.28678644875811515,"score_spread":0.26811819216786387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2286286503","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035645673,0.1635648,0.79161525,0.004567097,0.001208053,0.00006718446,0.0004296143,0.00033935605,0.034644097],"genre_scores_gemma":[0.10652284,0.30794555,0.55424565,0.0036665692,0.010294799,0.00041365012,0.0017209644,0.0008169499,0.014372993],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99667716,0.00085582247,0.00021239439,0.00063440704,0.0014415407,0.00017861059],"domain_scores_gemma":[0.9941472,0.0041103377,0.00019613447,0.0005489387,0.00088350166,0.00011388064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032235826,0.0017918054,0.0026452346,0.0038696926,0.0016721331,0.0056586605,0.00467318,0.0030106625,0.0074631115],"category_scores_gemma":[0.012415451,0.0015304369,0.0018542866,0.010025845,0.003334873,0.00754407,0.0029901438,0.005484617,0.0045348783],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006667871,0.0001413233,0.00079959416,0.0024856103,0.000118087024,0.000079882935,0.00022294138,0.042336185,0.00073436176,0.62500966,0.025873395,0.30213237],"study_design_scores_gemma":[0.000017150815,0.00005349837,0.00051221263,0.00051094947,0.000053506134,0.00032907745,0.00010400654,0.12743454,0.0007870167,0.78221893,0.08792611,0.000052992116],"about_ca_topic_score_codex":0.0032456703,"about_ca_topic_score_gemma":0.002310704,"teacher_disagreement_score":0.0074631115,"about_ca_system_score_codex":0.0036344668,"about_ca_system_score_gemma":0.0032596937,"threshold_uncertainty_score":0.026370049},"labels":[],"label_agreement":null},{"id":"W2320187719","doi":"10.1515/jisys.2004.13.4.329","title":"Evolutionary, Neural, and Statistical Approaches to Interval Clustering for Web Mining","year":2004,"lang":"en","type":"article","venue":"Journal of Intelligent Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Cluster analysis; Data mining; Web mining; Artificial intelligence; Machine learning; World Wide Web; Web service","score_opus":0.15315841354830215,"score_gpt":0.33149553045794183,"score_spread":0.1783371169096397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2320187719","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0057628066,0.0017889147,0.9894187,0.00029743806,0.00004900289,0.00002735844,0.00004961964,0.00009804341,0.0025080987],"genre_scores_gemma":[0.15775351,0.0037724606,0.83528227,0.00011956742,0.00023451727,0.00014581282,0.00023042453,0.00006237542,0.0023990464],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986376,0.000567498,0.00009699565,0.00015102417,0.000496154,0.000050617553],"domain_scores_gemma":[0.9966455,0.0021757735,0.00027631284,0.00030872758,0.000529195,0.00006440806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030292848,0.00041511867,0.00080239325,0.0026702567,0.00068821147,0.0013576787,0.0013694051,0.0006843836,0.0013527392],"category_scores_gemma":[0.010422261,0.00022562212,0.0008854466,0.0039510475,0.0010239093,0.002003749,0.0008606574,0.001335687,0.00040872366],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000047832844,0.00009885556,0.0026073789,0.00031783985,0.00017100351,0.00010293861,0.00034945295,0.25416648,0.0016423868,0.2776573,0.0026453268,0.46019316],"study_design_scores_gemma":[0.000011027082,0.000055466295,0.002213968,0.00006938684,0.00005040342,0.00017587036,0.00012313788,0.6995608,0.001556012,0.2875912,0.008542945,0.000049739294],"about_ca_topic_score_codex":0.0018782287,"about_ca_topic_score_gemma":0.0026940308,"teacher_disagreement_score":0.0030292848,"about_ca_system_score_codex":0.0008278764,"about_ca_system_score_gemma":0.0007104036,"threshold_uncertainty_score":0.016020596},"labels":[],"label_agreement":null},{"id":"W2355122750","doi":"","title":"Multilevel Clustering Algorithm Using Core-Sets Coarsening","year":2013,"lang":"en","type":"article","venue":"Jisuanji kexue yu tansuo","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Cluster analysis; Hierarchical clustering; Vertex (graph theory); Algorithm; Core (optical fiber); Enhanced Data Rates for GSM Evolution; Data mining; Theoretical computer science; Graph; Artificial intelligence","score_opus":0.07943376252303293,"score_gpt":0.34899215693311125,"score_spread":0.26955839441007834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2355122750","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006187209,0.00016512738,0.99150026,0.00007878473,0.000051069015,0.00006335405,0.00007927395,0.0005268427,0.0013479977],"genre_scores_gemma":[0.14767511,0.0002706985,0.8474107,0.00011738692,0.00007131993,0.00029995528,0.0007366107,0.00022536406,0.0031928462],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987381,0.00015820461,0.00008459567,0.0003440968,0.00055664446,0.00011836983],"domain_scores_gemma":[0.99897003,0.00022524502,0.000089947804,0.00022092526,0.0004372485,0.000056622408],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00084089977,0.00070280174,0.0014446808,0.0025732338,0.0013740616,0.0014197166,0.0018079154,0.0011104335,0.0032006237],"category_scores_gemma":[0.0030117691,0.00048076743,0.0015696753,0.002471766,0.00063569634,0.0016589471,0.0022259578,0.0014259218,0.0009849478],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017934159,0.00008978224,0.002338317,0.0002595677,0.00020658101,0.00012813225,0.00043236758,0.38379297,0.019637771,0.046988975,0.012987829,0.5329584],"study_design_scores_gemma":[0.000019761992,0.00003710154,0.00052227365,0.000014972278,0.00002333151,0.00006201216,0.00005073705,0.9784209,0.003998828,0.012160639,0.0046616364,0.000027779028],"about_ca_topic_score_codex":0.009554615,"about_ca_topic_score_gemma":0.008499931,"teacher_disagreement_score":0.009554615,"about_ca_system_score_codex":0.0011547761,"about_ca_system_score_gemma":0.0016272042,"threshold_uncertainty_score":0.018998027},"labels":[],"label_agreement":null},{"id":"W2370156110","doi":"","title":"Cluster Analysis and Visualization Enhanced Genetic Algorithm——II. Analysis of Cases and Validation","year":2004,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"CAE (Canada)","funders":"","keywords":"Visualization; Computer science; Consistency (knowledge bases); Cluster (spacecraft); Robustness (evolution); Data mining; Dimension (graph theory); Algorithm; Process (computing); Cluster analysis; Genetic algorithm; Machine learning; Artificial intelligence; Mathematics; Chemistry","score_opus":0.015945932551580253,"score_gpt":0.3218239109678345,"score_spread":0.30587797841625425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2370156110","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60973924,0.0001508503,0.37766033,0.00032395343,0.00005110079,0.00058536686,0.00043288473,0.0010272376,0.010029055],"genre_scores_gemma":[0.8429736,0.000068365094,0.1545025,0.000024471146,0.0000059192625,0.00036922417,0.00031682322,0.0000944815,0.0016446224],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975268,0.0011717944,0.00010248829,0.0002764103,0.0007500639,0.00017233446],"domain_scores_gemma":[0.9957902,0.002206975,0.00017479176,0.0006894818,0.0010926506,0.00004596665],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029882623,0.00076486106,0.00060454453,0.0018218351,0.0006880839,0.00096503843,0.0009756116,0.000964855,0.0024364975],"category_scores_gemma":[0.008780972,0.00025206438,0.0008335031,0.0014274789,0.0009754335,0.00060871063,0.00093816826,0.00060603843,0.00030095302],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006978114,0.0004057179,0.015620823,0.000485135,0.00012802363,0.0005697625,0.0015098305,0.731888,0.035505395,0.020364275,0.0030068748,0.1898184],"study_design_scores_gemma":[0.00005336598,0.00024814563,0.007181278,0.000057584963,0.00003554249,0.00017519906,0.00054225273,0.9435306,0.03983894,0.0043347673,0.003955393,0.000046867794],"about_ca_topic_score_codex":0.012555935,"about_ca_topic_score_gemma":0.004892579,"teacher_disagreement_score":0.012555935,"about_ca_system_score_codex":0.0011870482,"about_ca_system_score_gemma":0.0009824721,"threshold_uncertainty_score":0.024965703},"labels":[],"label_agreement":null},{"id":"W23725892","doi":"10.1007/978-3-642-37186-8_4","title":"Spectral Clustering: An Explorative Study of Proximity Measures","year":2013,"lang":"en","type":"book-chapter","venue":"Communications in computer and information science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Cluster analysis; Spectral clustering; Measure (data warehouse); Computer science; Euclidean distance; Similarity (geometry); Data mining; Similarity measure; Distance measures; Distance matrix; Usability; Boundary (topology); Artificial intelligence; Pattern recognition (psychology); Mathematics; Algorithm","score_opus":0.1191461492858851,"score_gpt":0.3554448074291799,"score_spread":0.2362986581432948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W23725892","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06928739,0.010504384,0.89532197,0.001256468,0.00014551757,0.00009426373,0.00044686714,0.00031727614,0.022625826],"genre_scores_gemma":[0.504238,0.008972731,0.47873878,0.00023881665,0.0003882041,0.00019456638,0.00073853333,0.00028746822,0.006202966],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99816906,0.00077542255,0.00007197526,0.00034738937,0.00057195436,0.00006408668],"domain_scores_gemma":[0.992721,0.005596976,0.00037994533,0.0006790533,0.00047961142,0.00014333746],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002135654,0.000774844,0.00088396476,0.004135643,0.0011590525,0.004420644,0.0017911,0.0009907235,0.0027938345],"category_scores_gemma":[0.014537137,0.00054825214,0.0008092217,0.009239113,0.002298512,0.0065150135,0.0023636231,0.0014849805,0.0005577549],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000528982,0.000062207524,0.0033199405,0.00063977786,0.00011808044,0.00013740444,0.0029717595,0.022058068,0.0024206706,0.8242055,0.0051982603,0.13881549],"study_design_scores_gemma":[0.000009377695,0.000052482614,0.0023858955,0.00015758733,0.00005467372,0.0005393065,0.0016297204,0.11963363,0.0017450223,0.85030085,0.023442643,0.000048703278],"about_ca_topic_score_codex":0.0012263674,"about_ca_topic_score_gemma":0.0012569011,"teacher_disagreement_score":0.004420644,"about_ca_system_score_codex":0.0011758383,"about_ca_system_score_gemma":0.0008019399,"threshold_uncertainty_score":0.011294544},"labels":[],"label_agreement":null},{"id":"W2373927542","doi":"10.1142/s0219622016500267","title":"A Hybrid Artificial Fish Swarm Simulated Annealing Optimization Algorithm for Automatic Identification of Clusters","year":2016,"lang":"en","type":"article","venue":"International Journal of Information Technology & Decision Making","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Crandall University; National Research Council Canada; Université de Moncton","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Computer science; Simulated annealing; Swarm behaviour; Fuzzy logic; Fuzzy clustering; A priori and a posteriori; Robustness (evolution); Algorithm; Rate of convergence; Data mining; Convergence (economics); Artificial intelligence","score_opus":0.013617359455804386,"score_gpt":0.3266986201948819,"score_spread":0.3130812607390775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2373927542","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007712593,0.00012002668,0.99059767,0.000052469037,0.0000305814,0.000035362125,0.00001959279,0.00030905305,0.0011226899],"genre_scores_gemma":[0.13249938,0.00012484842,0.86465645,0.000061490035,0.000018830715,0.00022936852,0.0001065399,0.00010170113,0.0022013846],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99959105,0.000113243856,0.000022025502,0.00008172299,0.00016605417,0.000025885003],"domain_scores_gemma":[0.9996388,0.00015492484,0.000035568337,0.000038069218,0.000115721035,0.000016958487],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007327638,0.0006397387,0.00071096775,0.0007659745,0.0005328483,0.0005474438,0.0011611461,0.0008737914,0.0012510779],"category_scores_gemma":[0.0014436604,0.0004002318,0.00082097785,0.0006789617,0.0004324107,0.00059295265,0.0006950791,0.0006176605,0.00038713237],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013817045,0.000054719934,0.0012337791,0.0001476969,0.00017635229,0.00007402759,0.00018022936,0.7798945,0.0150713995,0.011346746,0.002360917,0.18932155],"study_design_scores_gemma":[0.000009363521,0.000019239467,0.00013390188,0.0000047602853,0.00000817346,0.00001891586,0.0000079709225,0.99601537,0.0013176177,0.0010995854,0.0013577213,0.0000073130777],"about_ca_topic_score_codex":0.0048675975,"about_ca_topic_score_gemma":0.00592284,"teacher_disagreement_score":0.0048675975,"about_ca_system_score_codex":0.0005598,"about_ca_system_score_gemma":0.00094446604,"threshold_uncertainty_score":0.009678483},"labels":[],"label_agreement":null},{"id":"W2390764","doi":"10.1007/978-3-642-20841-6_22","title":"Semi-supervised Parameter-Free Divisive Hierarchical Clustering of Categorical Data","year":2011,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Computer science; Categorical variable; Hierarchical clustering; Artificial intelligence; Canopy clustering algorithm; CURE data clustering algorithm; Data mining; Single-linkage clustering; Correlation clustering; Constrained clustering; Pattern recognition (psychology); Data stream clustering; Constraint (computer-aided design); Dendrogram; Machine learning; Mathematics","score_opus":0.06073154317108151,"score_gpt":0.3008922138382473,"score_spread":0.24016067066716582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2390764","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004204945,0.00012027474,0.99396837,0.000057626512,0.000020539133,0.000053496133,0.0001702854,0.0008610856,0.0005434073],"genre_scores_gemma":[0.14276037,0.00018442766,0.84936595,0.00015267267,0.00009861859,0.00031668652,0.0024208217,0.0006076072,0.004092796],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99627656,0.0013234849,0.00026292933,0.0009053106,0.0010440175,0.00018758482],"domain_scores_gemma":[0.99380314,0.0019464873,0.00037425553,0.0023052127,0.0014028206,0.00016811765],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034254899,0.000775138,0.002051535,0.0013675204,0.0010507411,0.0016143111,0.005583869,0.0016063,0.0025876397],"category_scores_gemma":[0.011323871,0.0010400608,0.0019019608,0.0022429226,0.0012442191,0.002459833,0.0034609116,0.0020518661,0.0027931426],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037386475,0.000179125,0.0015852585,0.00059589616,0.0003008919,0.00008322812,0.0005429014,0.22660038,0.015921988,0.020728374,0.014598472,0.7184896],"study_design_scores_gemma":[0.000016186228,0.000049488546,0.00053284084,0.000028794595,0.000027258926,0.00010549877,0.000066077635,0.9611602,0.0044053798,0.030613473,0.0029663607,0.000028467763],"about_ca_topic_score_codex":0.0028067026,"about_ca_topic_score_gemma":0.006609463,"teacher_disagreement_score":0.005583869,"about_ca_system_score_codex":0.0009814505,"about_ca_system_score_gemma":0.0019179033,"threshold_uncertainty_score":0.018115938},"labels":[],"label_agreement":null},{"id":"W2394938095","doi":"","title":"Hybrid hierarchical clustering: cluster assessment via cluster validation indices.","year":2012,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Cluster analysis; Hierarchical clustering; Single-linkage clustering; Computer science; Data mining; Benchmark (surveying); Cluster (spacecraft); Brown clustering; Correlation clustering; Fuzzy clustering; Consensus clustering; Hierarchical clustering of networks; CURE data clustering algorithm; Pattern recognition (psychology); Artificial intelligence; Geography","score_opus":0.025519508021927914,"score_gpt":0.33101121721370524,"score_spread":0.3054917091917773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2394938095","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025792994,0.00061599805,0.9684004,0.00015473425,0.000089858055,0.0003148257,0.00065134064,0.0019768896,0.0020029687],"genre_scores_gemma":[0.242356,0.00015578252,0.7541404,0.000069444046,0.000036606238,0.00032245967,0.0012203675,0.00025037114,0.0014485756],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9938065,0.0025613292,0.000330259,0.0009289975,0.0021717534,0.00020111215],"domain_scores_gemma":[0.9912479,0.003121782,0.0010604431,0.0014861706,0.0028610376,0.00022272847],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00752733,0.00093804044,0.0008603124,0.004045506,0.0010753747,0.002447818,0.0021032437,0.0010042905,0.001532366],"category_scores_gemma":[0.021646991,0.0005492527,0.0009123816,0.003465618,0.0008857142,0.0021560797,0.0017571579,0.0013141861,0.0009918398],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008258258,0.00029336396,0.021221234,0.00077401055,0.0010993452,0.00016305204,0.0009190156,0.2302343,0.022495486,0.025672546,0.016065262,0.6802365],"study_design_scores_gemma":[0.00004430948,0.00013460846,0.0076020644,0.00008849232,0.00012465655,0.0001287261,0.00020367767,0.9529191,0.016277533,0.016950557,0.005425236,0.00010101814],"about_ca_topic_score_codex":0.005222459,"about_ca_topic_score_gemma":0.0076878327,"teacher_disagreement_score":0.00752733,"about_ca_system_score_codex":0.0017058436,"about_ca_system_score_gemma":0.0020101469,"threshold_uncertainty_score":0.03980881},"labels":[],"label_agreement":null},{"id":"W2398976774","doi":"10.5591/978-1-57735-516-8/ijcai11-195","title":"Discerning linkage-based algorithms among hierarchical clustering methods","year":2011,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Algorithm; Hierarchical clustering; Linkage (software); Class (philosophy); Canopy clustering algorithm; Hierarchical clustering of networks; Complete linkage; Focus (optics); Data mining; Correlation clustering; Artificial intelligence","score_opus":0.07758011686625375,"score_gpt":0.3613604285279193,"score_spread":0.2837803116616655,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2398976774","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008662745,0.00076683605,0.9874536,0.00029577402,0.000029559144,0.00014931132,0.000054573764,0.0002324149,0.0023551094],"genre_scores_gemma":[0.14451733,0.00086404075,0.8522754,0.00024721658,0.00016731706,0.00039777177,0.0003069527,0.00016224847,0.0010617246],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9880427,0.004885233,0.00081220694,0.001965216,0.0039126114,0.00038192558],"domain_scores_gemma":[0.96394026,0.022588367,0.0036497163,0.003968555,0.005013261,0.0008399693],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01853662,0.0012033983,0.0015496438,0.006197934,0.0024825553,0.006026309,0.002818229,0.0033189608,0.0023557935],"category_scores_gemma":[0.05263599,0.00087211153,0.0013338914,0.0048168106,0.0036800941,0.010155306,0.0051480415,0.002591322,0.0013114173],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003508891,0.00023960436,0.008887834,0.0008002297,0.00024479054,0.0001867449,0.0014825811,0.08063184,0.004365872,0.60588765,0.0033581129,0.29356384],"study_design_scores_gemma":[0.00008260279,0.00030168777,0.002190312,0.00019064207,0.00010541819,0.00032064796,0.0005384673,0.35228983,0.0053069456,0.6251859,0.013402302,0.00008528505],"about_ca_topic_score_codex":0.0007453482,"about_ca_topic_score_gemma":0.000936982,"teacher_disagreement_score":0.01853662,"about_ca_system_score_codex":0.002093133,"about_ca_system_score_gemma":0.0021643816,"threshold_uncertainty_score":0.09803218},"labels":[],"label_agreement":null},{"id":"W2399784609","doi":"","title":"Bruno Simeone's Work in Clustering.","year":2011,"lang":"en","type":"article","venue":"Cologne Twente Workshop on Graphs and Combinatorial Optimization","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"HEC Montréal","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.025207815627535973,"score_gpt":0.25765075609501553,"score_spread":0.23244294046747957,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2399784609","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00657455,0.13790607,0.39000934,0.16901939,0.049527563,0.00016966194,0.0018387234,0.0032829894,0.24167167],"genre_scores_gemma":[0.15964238,0.06747879,0.22887781,0.04751739,0.019517617,0.0003684805,0.00277827,0.0027220221,0.47109723],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971601,0.00079732365,0.00006873703,0.0010429029,0.00072354334,0.00020747149],"domain_scores_gemma":[0.9955754,0.002176858,0.00011624279,0.00078224903,0.0008038295,0.0005454933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024353792,0.0012732914,0.0012977349,0.0025470303,0.0021026004,0.0032458294,0.0019960108,0.0027264524,0.017840315],"category_scores_gemma":[0.00760157,0.0007887201,0.00097277004,0.004534915,0.002227568,0.0053015556,0.0025715455,0.003975452,0.0118417265],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017556238,0.000073216914,0.00079616066,0.00055163377,0.00007939523,0.00016209668,0.00064496725,0.003933679,0.0014108556,0.23947896,0.5795699,0.1731236],"study_design_scores_gemma":[0.00002010965,0.000015109109,0.0005093004,0.00016525768,0.000025590178,0.0002973172,0.00017623055,0.003804513,0.0019573208,0.13995947,0.85302156,0.00004824202],"about_ca_topic_score_codex":0.006864415,"about_ca_topic_score_gemma":0.0067895427,"teacher_disagreement_score":0.017840315,"about_ca_system_score_codex":0.0031139767,"about_ca_system_score_gemma":0.0013249335,"threshold_uncertainty_score":0.059681833},"labels":[],"label_agreement":null},{"id":"W2408186052","doi":"10.1137/1.9781611973440.96","title":"Density-Based Clustering Validation","year":2014,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":262,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Cluster analysis; Computer science; Data mining; Kernel density estimation; Cluster (spacecraft); Kernel (algebra); Mathematics; Artificial intelligence; Pattern recognition (psychology); Statistics; Combinatorics","score_opus":0.02015969811375551,"score_gpt":0.284899446381844,"score_spread":0.2647397482680885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2408186052","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.054793216,0.00065779896,0.93786794,0.00019069546,0.00011006849,0.00041864373,0.0005272026,0.0018251556,0.0036092945],"genre_scores_gemma":[0.5693581,0.00041795947,0.42430273,0.0001879687,0.000054417917,0.000550926,0.0028245621,0.00062874716,0.0016745417],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9777729,0.008688192,0.0018155223,0.002564231,0.008373225,0.000786015],"domain_scores_gemma":[0.9445189,0.021760114,0.0034373004,0.009269884,0.020437168,0.00057659246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.027637724,0.0016807712,0.0021567722,0.007147281,0.0020771301,0.003911127,0.0036774501,0.0026953327,0.0021116813],"category_scores_gemma":[0.11357451,0.0006569583,0.0018560021,0.0045190286,0.0021703588,0.004201065,0.0043409783,0.002017509,0.0013710721],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010165012,0.00033637814,0.037478395,0.0013396969,0.0008505393,0.00034578957,0.0014406287,0.52743524,0.021506475,0.057390742,0.009205127,0.34165442],"study_design_scores_gemma":[0.00003130325,0.00011592177,0.0059313634,0.00013255091,0.000068431815,0.00033909548,0.00025931824,0.95920056,0.015676528,0.014836225,0.0033190486,0.000089670815],"about_ca_topic_score_codex":0.004832457,"about_ca_topic_score_gemma":0.0033343819,"teacher_disagreement_score":0.027637724,"about_ca_system_score_codex":0.002412501,"about_ca_system_score_gemma":0.0024967717,"threshold_uncertainty_score":0.146164},"labels":[],"label_agreement":null},{"id":"W2417989999","doi":"","title":"Hierarchical Information-Theoretic Co-clustering for High Dimensional Data","year":2011,"lang":"en","type":"article","venue":"HKBU Institutional Repository (Hong Kong Baptist University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Cluster analysis; Hierarchical clustering; Data mining; Biclustering; Data science; Artificial intelligence; Fuzzy clustering; CURE data clustering algorithm","score_opus":0.03773163137134344,"score_gpt":0.2488227367389772,"score_spread":0.21109110536763376,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2417989999","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0079986295,0.0009114015,0.9877,0.00030964197,0.000052992542,0.00011816282,0.0010545094,0.0009229653,0.0009316469],"genre_scores_gemma":[0.27701074,0.0012296295,0.7047287,0.00023744612,0.00019879347,0.000619122,0.0114021385,0.0005457496,0.0040277317],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9932302,0.0027851642,0.0004984278,0.0011790709,0.0019011945,0.0004058158],"domain_scores_gemma":[0.9788735,0.010008165,0.001221061,0.006368727,0.0030004657,0.0005280628],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065048733,0.0011805951,0.0024995832,0.007998151,0.0028925391,0.0038955682,0.004399373,0.0023331072,0.0033789047],"category_scores_gemma":[0.039856967,0.0011861908,0.0028264364,0.0103227,0.0020068772,0.004042963,0.0043030353,0.002736989,0.0024681285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000681656,0.00040582914,0.0076001193,0.0011216144,0.0007941688,0.00042059974,0.0010923452,0.42092666,0.004755552,0.15356913,0.028439743,0.38019267],"study_design_scores_gemma":[0.000023646742,0.00004414668,0.0015747395,0.000051132767,0.00006753189,0.00014533817,0.00013841786,0.85014504,0.0014725387,0.14293125,0.0033587492,0.000047513848],"about_ca_topic_score_codex":0.009538276,"about_ca_topic_score_gemma":0.016707014,"teacher_disagreement_score":0.009538276,"about_ca_system_score_codex":0.0028169213,"about_ca_system_score_gemma":0.0043052607,"threshold_uncertainty_score":0.034401476},"labels":[],"label_agreement":null},{"id":"W2494099403","doi":"10.1145/2851613.2851643","title":"PatchWork, a scalable density-grid clustering algorithm","year":2016,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Computer Research Institute of Montréal","funders":"","keywords":"Cluster analysis; Computer science; Scalability; SPARK (programming language); CURE data clustering algorithm; Data mining; Canopy clustering algorithm; Correlation clustering; Computation; Knowledge extraction; Data stream clustering; Anomaly detection; Outlier; Algorithm; Artificial intelligence; Database","score_opus":0.017057390880455377,"score_gpt":0.2673199606908087,"score_spread":0.25026256981035333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2494099403","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009038986,0.00052345905,0.97837764,0.00036569513,0.00018883194,0.00022384714,0.00044218023,0.0075597274,0.0032796084],"genre_scores_gemma":[0.1057995,0.00042794115,0.88371414,0.00025959755,0.00009339097,0.0004071084,0.0021948065,0.0011284698,0.005975002],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992417,0.000110435416,0.000040603172,0.00020091778,0.00032425727,0.00008210119],"domain_scores_gemma":[0.99887997,0.0002703341,0.000082558334,0.00028769835,0.00036872353,0.00011064122],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009268355,0.0009773015,0.0014617048,0.0015687774,0.0015023893,0.0012959279,0.0028640002,0.0012707277,0.004240085],"category_scores_gemma":[0.0052315285,0.00073591917,0.0010760552,0.0023570657,0.00071508426,0.0019563017,0.0025489829,0.0013639372,0.0025318037],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050104636,0.00017178211,0.0040256046,0.00035015424,0.0002896523,0.000267279,0.00047540793,0.39589062,0.007381363,0.022801135,0.080058224,0.48778763],"study_design_scores_gemma":[0.0001156505,0.000044327004,0.00040296643,0.000014538181,0.000024546722,0.00014453693,0.000082164974,0.96391064,0.0020831598,0.017633142,0.015519398,0.000025002637],"about_ca_topic_score_codex":0.015492568,"about_ca_topic_score_gemma":0.016959414,"teacher_disagreement_score":0.015492568,"about_ca_system_score_codex":0.001017294,"about_ca_system_score_gemma":0.002319952,"threshold_uncertainty_score":0.030804753},"labels":[],"label_agreement":null},{"id":"W2512875827","doi":"10.18637/jss.v072.i05","title":"<b>RSKC</b>: An<i>R</i>Package for a Robust and Sparse K-Means Clustering Algorithm","year":2016,"lang":"en","type":"article","venue":"Journal of Statistical Software","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Outlier; Cluster analysis; Computer science; Data mining; R package; Identification (biology); Algorithm; Monte Carlo method; Artificial intelligence; Mathematics; Statistics","score_opus":0.033844006603445616,"score_gpt":0.3084657004614797,"score_spread":0.2746216938580341,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2512875827","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019352706,0.00039641163,0.7657343,0.0005254048,0.00039996774,0.00030770662,0.03339776,0.1938272,0.0034759757],"genre_scores_gemma":[0.011701428,0.0003384122,0.86581796,0.0005724712,0.00011267564,0.0019829024,0.027909156,0.0869103,0.004654733],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974026,0.0008536855,0.00026243992,0.0005392933,0.00075646414,0.00018556468],"domain_scores_gemma":[0.9885315,0.0062944326,0.001022668,0.0018289811,0.002079226,0.00024328224],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044858395,0.004450302,0.002675449,0.0036181735,0.0013874857,0.0027290052,0.0051713213,0.00230444,0.10193569],"category_scores_gemma":[0.029946018,0.0025826269,0.0029655811,0.0039844266,0.0012330262,0.003420288,0.0036006114,0.004842449,0.09441852],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006202791,0.00019717778,0.0025959658,0.0028712342,0.00085495313,0.0003623073,0.0006157177,0.02128648,0.009308229,0.021134142,0.7948718,0.14528172],"study_design_scores_gemma":[0.00069133064,0.0002138728,0.0055825664,0.00070645195,0.00038924866,0.0011108036,0.00019053342,0.22754563,0.029824585,0.066330105,0.6668005,0.0006144092],"about_ca_topic_score_codex":0.006139835,"about_ca_topic_score_gemma":0.007782201,"teacher_disagreement_score":0.10193569,"about_ca_system_score_codex":0.0010575907,"about_ca_system_score_gemma":0.0033869445,"threshold_uncertainty_score":0.34100884},"labels":[],"label_agreement":null},{"id":"W2515705452","doi":"10.18637/jss.v073.i07","title":"My Early Interactions with Jan and Some of His Lost Papers","year":2016,"lang":"en","type":"article","venue":"Journal of Statistical Software","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Period (music); Scaling; Multidimensional scaling; Computer science; History; Mathematics; Art; Machine learning","score_opus":0.012775728765790148,"score_gpt":0.2838603270710328,"score_spread":0.2710845983052427,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2515705452","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013908357,0.03767016,0.015297783,0.7570931,0.14356121,0.00010837948,0.0006232682,0.0006968112,0.03104098],"genre_scores_gemma":[0.2297581,0.030430248,0.025682118,0.21791609,0.17133355,0.00051359774,0.00087186764,0.004381706,0.3191128],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97608733,0.010319716,0.0009216042,0.0023239877,0.009088988,0.0012583982],"domain_scores_gemma":[0.88747436,0.031112704,0.006480469,0.00621169,0.0342914,0.03442943],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02159546,0.0014505072,0.0016898907,0.003239064,0.010360385,0.015621328,0.002647136,0.00262353,0.017063998],"category_scores_gemma":[0.11842648,0.0010646909,0.0011197419,0.0033294049,0.007926658,0.009696931,0.00676886,0.01742462,0.014555188],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013404101,0.00012714417,0.0015426459,0.00023909823,0.00006604361,0.0010381023,0.017000841,0.00024273912,0.0008778372,0.029160118,0.90236336,0.04720797],"study_design_scores_gemma":[0.000013558423,0.000048876773,0.0009138462,0.00038699975,0.000017481065,0.0014205306,0.010366067,0.00026291475,0.00055640686,0.014261067,0.97166663,0.00008555746],"about_ca_topic_score_codex":0.0012396052,"about_ca_topic_score_gemma":0.0018491133,"teacher_disagreement_score":0.02159546,"about_ca_system_score_codex":0.0047917524,"about_ca_system_score_gemma":0.004040276,"threshold_uncertainty_score":0.114209116},"labels":[],"label_agreement":null},{"id":"W2520601089","doi":"10.1145/2938503.2938520","title":"A new scalable aggregation scheme for fuzzy clustering Taking unstructured textual resources as a case","year":2016,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Rimouski","funders":"","keywords":"Cluster analysis; Computer science; Scalability; Scheme (mathematics); Fuzzy clustering; Data mining; Artificial intelligence; Fuzzy logic; Correlation clustering; CURE data clustering algorithm; Machine learning; Database; Mathematics","score_opus":0.026825741286833293,"score_gpt":0.3094472715279418,"score_spread":0.28262153024110853,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2520601089","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009679362,0.00018015418,0.9870837,0.00014591933,0.00008056771,0.0001027858,0.00019167257,0.0011426719,0.0013933177],"genre_scores_gemma":[0.25283372,0.00023067807,0.74066156,0.00013600625,0.00014643588,0.00021860268,0.0009269966,0.00014252598,0.004703503],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860495,0.0001854669,0.00015344779,0.00027678217,0.0006387667,0.00014055321],"domain_scores_gemma":[0.9984491,0.00023518,0.000113193804,0.0005426141,0.00053515873,0.0001247565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014786561,0.0007451334,0.0011914241,0.001829677,0.0015309775,0.0016709237,0.0020172074,0.00086061,0.0026655267],"category_scores_gemma":[0.0035160247,0.00035676907,0.0008393652,0.002849549,0.0005727734,0.0024948355,0.0026636552,0.0011171518,0.0011206276],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053353753,0.00028625515,0.001802469,0.00022312641,0.00018375051,0.00041108715,0.0008835945,0.1592759,0.04460514,0.06620958,0.020559574,0.705026],"study_design_scores_gemma":[0.000025100859,0.000063411455,0.000573545,0.000014845523,0.00004186316,0.00015509466,0.00013396837,0.95699626,0.0072235586,0.02715191,0.007580805,0.00003966856],"about_ca_topic_score_codex":0.0063735587,"about_ca_topic_score_gemma":0.008609994,"teacher_disagreement_score":0.0063735587,"about_ca_system_score_codex":0.0014973732,"about_ca_system_score_gemma":0.0014993252,"threshold_uncertainty_score":0.012672901},"labels":[],"label_agreement":null},{"id":"W2525029050","doi":"10.1007/978-3-319-47160-0_1","title":"Advances in Rough and Soft Clustering: Meta-Clustering, Dynamic Clustering, Data-Stream Clustering","year":2016,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Fuzzy clustering; Data stream clustering; CURE data clustering algorithm; Canopy clustering algorithm; Correlation clustering; Consensus clustering; Conceptual clustering; Clustering high-dimensional data; Artificial intelligence; Machine learning","score_opus":0.04102602574981118,"score_gpt":0.3191668127387236,"score_spread":0.2781407869889124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2525029050","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002995321,0.048507523,0.93363357,0.0018424245,0.0015398277,0.000048541184,0.00023811283,0.0009087652,0.01028596],"genre_scores_gemma":[0.08529799,0.055124905,0.8335283,0.0008723324,0.0053931195,0.00014054129,0.0011828953,0.00082670205,0.0176332],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9980008,0.00039049666,0.000118875236,0.0004923665,0.00092884793,0.000068674446],"domain_scores_gemma":[0.9974026,0.0011317036,0.00012512447,0.00049231667,0.0007261626,0.000122174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025144431,0.0012517197,0.0021450887,0.002886254,0.0007411636,0.0038170624,0.002148704,0.0015894845,0.0034818952],"category_scores_gemma":[0.0056833546,0.0007885582,0.001515129,0.008076201,0.0014922724,0.004178528,0.0019799701,0.0040293573,0.0030890736],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000094761555,0.00009720982,0.0006767838,0.0015149259,0.00023853195,0.00008117491,0.00033435973,0.048607126,0.0044150115,0.18723518,0.054073945,0.70263106],"study_design_scores_gemma":[0.000021106223,0.00009620766,0.0013746361,0.0002631486,0.000114873146,0.0005644874,0.00019487366,0.3656381,0.0045428257,0.46139207,0.16566293,0.00013475025],"about_ca_topic_score_codex":0.0016249624,"about_ca_topic_score_gemma":0.001593249,"teacher_disagreement_score":0.0038170624,"about_ca_system_score_codex":0.0014273089,"about_ca_system_score_gemma":0.0012579999,"threshold_uncertainty_score":0.013297796},"labels":[],"label_agreement":null},{"id":"W2529045002","doi":"10.1007/s10115-016-0998-9","title":"Finding multiple stable clusterings","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Office of Naval Research; National Science Foundation","keywords":"Cluster analysis; Computer science; Data mining; Constrained clustering; Clustering high-dimensional data; Spectral clustering; Fuzzy clustering; Correlation clustering; Consensus clustering; Partition (number theory); CURE data clustering algorithm; Stability (learning theory); Machine learning; Artificial intelligence; Mathematics","score_opus":0.022548493315349666,"score_gpt":0.2736265916603705,"score_spread":0.2510780983450208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2529045002","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09634908,0.00083355507,0.89593244,0.00069746934,0.00015460992,0.00024807695,0.00047462658,0.00082346995,0.0044867117],"genre_scores_gemma":[0.52128667,0.00039900202,0.46768528,0.00019248897,0.00015686163,0.00032219544,0.002068403,0.0004952815,0.00739385],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99759054,0.00049359846,0.00016607149,0.0009756435,0.0005511167,0.00022312562],"domain_scores_gemma":[0.9901957,0.0045851986,0.00080731587,0.0017480607,0.002191603,0.00047207653],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025129118,0.0013754332,0.002053639,0.005706573,0.002981475,0.0030873942,0.0037526211,0.0029413984,0.0069274306],"category_scores_gemma":[0.02120949,0.0016158427,0.0024828482,0.0044459263,0.0015468694,0.0042069415,0.0035249642,0.00195933,0.0024900981],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019980937,0.00042924215,0.01437215,0.0011769116,0.001545148,0.0010015037,0.0016489188,0.23896158,0.036051728,0.1283021,0.0234632,0.5510494],"study_design_scores_gemma":[0.00014907985,0.00021964451,0.0032760978,0.00006886735,0.00025073,0.0005280762,0.00077238487,0.8088738,0.00887589,0.17208548,0.0048139766,0.00008601944],"about_ca_topic_score_codex":0.0019931165,"about_ca_topic_score_gemma":0.003676966,"teacher_disagreement_score":0.0069274306,"about_ca_system_score_codex":0.0014967852,"about_ca_system_score_gemma":0.001204108,"threshold_uncertainty_score":0.023174584},"labels":[],"label_agreement":null},{"id":"W2532543454","doi":"10.1109/is.2006.348479","title":"A Method for Fuzzy Clustering with Ordinal Attributes Replaced by Fuzzy Set Parameters","year":2006,"lang":"en","type":"article","venue":"2006 3rd International IEEE Conference Intelligent Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Mathematics; Ordinal data; Fuzzy set; Ordinal regression; Cluster analysis; Fuzzy clustering; Fuzzy logic; Ordinal optimization; Data mining; Set (abstract data type); Artificial intelligence; Fuzzy classification; Pattern recognition (psychology); Computer science; Statistics","score_opus":0.05653600829032689,"score_gpt":0.33658797172013016,"score_spread":0.28005196342980326,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2532543454","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00036849812,0.000092746086,0.99865013,0.000044802993,0.000067781795,0.00005314711,0.00003734858,0.00019790852,0.00048759556],"genre_scores_gemma":[0.00596599,0.00007929087,0.9929952,0.00003325673,0.000032923577,0.00013792196,0.000049174352,0.000048676087,0.00065759243],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996332,0.0011097625,0.00027399373,0.00073215313,0.0014500048,0.000102177575],"domain_scores_gemma":[0.9976603,0.001030298,0.00014697739,0.0003475471,0.00075491524,0.000059937607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038149413,0.001158951,0.0014630397,0.003903379,0.0018467903,0.0022248377,0.0018153465,0.001769732,0.0032690351],"category_scores_gemma":[0.007937637,0.0007882456,0.0016672878,0.004527366,0.0018125562,0.0022439247,0.0014550166,0.0031673065,0.002258794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019525712,0.00007902888,0.00089988974,0.00067087,0.00033203457,0.00024828548,0.001074015,0.051926233,0.015921008,0.22619464,0.013746504,0.68871224],"study_design_scores_gemma":[0.000116287585,0.0001822017,0.0016699666,0.00030658918,0.00018742813,0.0011925542,0.00032944203,0.567403,0.015423191,0.25959608,0.15321341,0.00037981902],"about_ca_topic_score_codex":0.0032778257,"about_ca_topic_score_gemma":0.0033743025,"teacher_disagreement_score":0.003903379,"about_ca_system_score_codex":0.0013151905,"about_ca_system_score_gemma":0.0018160277,"threshold_uncertainty_score":0.020175576},"labels":[],"label_agreement":null},{"id":"W2554498112","doi":"10.1109/ijcnn.2016.7727579","title":"LVC: Local Variance-based Clustering","year":2016,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; DBSCAN; Computer science; Spectral clustering; Variance (accounting); Data mining; Artificial intelligence; Outlier; Pattern recognition (psychology); CURE data clustering algorithm; Correlation clustering; Clustering high-dimensional data; Single-linkage clustering","score_opus":0.021927587469671387,"score_gpt":0.28175319326366105,"score_spread":0.25982560579398967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2554498112","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005362712,0.0002844217,0.9885937,0.00013204124,0.000058455375,0.00012361202,0.00030638475,0.0042597638,0.00087884534],"genre_scores_gemma":[0.148289,0.0003290064,0.84295845,0.00028411992,0.00012450336,0.00048459633,0.0031133797,0.0014749109,0.0029421966],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960445,0.00072322896,0.00024015694,0.00090961857,0.0017868041,0.000295639],"domain_scores_gemma":[0.99566597,0.0011087129,0.00038090211,0.0007442333,0.0019439692,0.00015633473],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025113453,0.0014045675,0.0018904047,0.0058280434,0.0018926365,0.0023375067,0.00410115,0.0018968513,0.0025229284],"category_scores_gemma":[0.009220328,0.0006323026,0.0020242156,0.005781011,0.0013722093,0.0016644114,0.0024798245,0.0020756244,0.002565373],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003631223,0.00025551696,0.005687775,0.00042169428,0.00042395684,0.00019669304,0.00073988235,0.2121103,0.016516587,0.01797787,0.029129017,0.7161776],"study_design_scores_gemma":[0.000034907614,0.00006120841,0.0016496183,0.000037030793,0.000044834702,0.0001678939,0.0001494828,0.97168994,0.00699597,0.012880817,0.0062121693,0.000076182725],"about_ca_topic_score_codex":0.021551361,"about_ca_topic_score_gemma":0.0231206,"teacher_disagreement_score":0.021551361,"about_ca_system_score_codex":0.0023150777,"about_ca_system_score_gemma":0.0030053842,"threshold_uncertainty_score":0.042851865},"labels":[],"label_agreement":null},{"id":"W2556640145","doi":"10.3233/jifs-169172","title":"Double fuzzy C-means model and its application in the technology innovation of China","year":2016,"lang":"en","type":"article","venue":"Journal of Intelligent & Fuzzy Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Department of Family and Community Medicine, University of Toronto; National Social Science Fund of China; National Natural Science Foundation of China","keywords":"Fuzzy logic; Data mining; Computer science; Class (philosophy); Fuzzy clustering; Cluster analysis; Index (typography); Cluster (spacecraft); Artificial intelligence; Set (abstract data type); Fuzzy set; Pattern recognition (psychology)","score_opus":0.03422056838834272,"score_gpt":0.3118910415095601,"score_spread":0.2776704731212174,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2556640145","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13273856,0.0014003614,0.85584724,0.0006198524,0.00017959446,0.00019332941,0.00028067446,0.00033716363,0.008403211],"genre_scores_gemma":[0.84132105,0.0008679706,0.15354142,0.000072719624,0.000052827218,0.00023619832,0.00027673037,0.000039130613,0.003591903],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99853003,0.00027970457,0.00010489682,0.000392504,0.0005332609,0.00015959807],"domain_scores_gemma":[0.9988714,0.00038018016,0.00012747059,0.000083327985,0.00047307895,0.00006441378],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00207135,0.00083173544,0.0008563432,0.0028573328,0.0010518789,0.002005591,0.0016487993,0.0012045881,0.0012577208],"category_scores_gemma":[0.0041666557,0.00032284745,0.0011139208,0.00360743,0.0007826631,0.0014601495,0.0011735058,0.0008251003,0.00017890664],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019238754,0.00008673058,0.009515127,0.00033898026,0.00017778453,0.00029425707,0.000500071,0.7810181,0.0031547283,0.039880224,0.0020136468,0.16282792],"study_design_scores_gemma":[0.000008511547,0.000030550375,0.0015511108,0.000018272898,0.000018599283,0.000044856108,0.000066452165,0.9890681,0.00070707884,0.0072423895,0.001216735,0.00002721774],"about_ca_topic_score_codex":0.04593819,"about_ca_topic_score_gemma":0.020707324,"teacher_disagreement_score":0.04593819,"about_ca_system_score_codex":0.0027208715,"about_ca_system_score_gemma":0.003522943,"threshold_uncertainty_score":0.091341615},"labels":[],"label_agreement":null},{"id":"W2580505187","doi":"","title":"A New Measure For Clustering Model Selection","year":2008,"lang":"en","type":"dissertation","venue":"UWSpace (University of Waterloo)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of Waterloo","keywords":"Measure (data warehouse); Cluster analysis; Selection (genetic algorithm); Computer science; Data mining; Artificial intelligence","score_opus":0.024037401848323613,"score_gpt":0.24678027070918093,"score_spread":0.2227428688608573,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2580505187","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017666154,0.000526881,0.9951278,0.00021735295,0.00023267342,0.00006396938,0.00015170094,0.0003420626,0.0015709104],"genre_scores_gemma":[0.0693946,0.00085131713,0.92386174,0.00044439128,0.00083456817,0.00061121024,0.0011117058,0.0004418713,0.0024486831],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.984053,0.004532183,0.00092183193,0.0024437704,0.0076671746,0.0003821237],"domain_scores_gemma":[0.98396945,0.008728974,0.0012245316,0.00224318,0.0033893078,0.0004445518],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009846266,0.0020223553,0.0024471262,0.0095834965,0.0019133532,0.0057992293,0.0032385697,0.002392811,0.0042722262],"category_scores_gemma":[0.04372887,0.0006418166,0.0023181052,0.007140446,0.0024682437,0.006056483,0.004011019,0.003949494,0.0019977572],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002344711,0.00019003556,0.006690397,0.00086925045,0.0009863998,0.00027993447,0.0005676726,0.09564032,0.009581033,0.2604311,0.024714971,0.59981436],"study_design_scores_gemma":[0.00009112907,0.00033336398,0.004092426,0.00027409612,0.00028658562,0.0009975062,0.00029223523,0.6776112,0.008094152,0.26943338,0.038297627,0.0001963206],"about_ca_topic_score_codex":0.0014686028,"about_ca_topic_score_gemma":0.0017188011,"teacher_disagreement_score":0.009846266,"about_ca_system_score_codex":0.0026404094,"about_ca_system_score_gemma":0.0021898013,"threshold_uncertainty_score":0.052072704},"labels":[],"label_agreement":null},{"id":"W2582229329","doi":"10.3390/ijgi6010030","title":"A Novel Divisive Hierarchical Clustering Algorithm for Geospatial Analysis","year":2017,"lang":"en","type":"article","venue":"ISPRS International Journal of Geo-Information","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Geospatial analysis; Computer science; Hierarchical clustering; Single-linkage clustering; Correlation clustering; CURE data clustering algorithm; Robustness (evolution); Pattern recognition (psychology); Context (archaeology); Data mining; Artificial intelligence; Algorithm; Geography; Cartography","score_opus":0.01706036113338718,"score_gpt":0.33289478348858553,"score_spread":0.3158344223551984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2582229329","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029977018,0.00031928907,0.9944101,0.000119953525,0.00005321637,0.00011128613,0.00012726898,0.0008390838,0.0010221567],"genre_scores_gemma":[0.053749982,0.0003312928,0.9414211,0.00012831228,0.00006908094,0.0002779534,0.0007957498,0.000207121,0.003019329],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985013,0.0002180985,0.00007409694,0.0003890137,0.00072644337,0.00009094655],"domain_scores_gemma":[0.99929285,0.00015169913,0.000054636534,0.00011303315,0.0003476525,0.000040049945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001021241,0.0011618559,0.0013709062,0.0024843249,0.001617007,0.0012679851,0.0031410588,0.0012131175,0.002362482],"category_scores_gemma":[0.002258568,0.0006429116,0.0013613487,0.004005818,0.0007891535,0.001980187,0.0017611487,0.0014663243,0.0016384239],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014715946,0.00010560814,0.001715873,0.0003224408,0.00023400443,0.00020388167,0.00068768027,0.21059102,0.019873332,0.025538484,0.017812535,0.722768],"study_design_scores_gemma":[0.000021600912,0.000031641463,0.0005314285,0.000016367558,0.00002904991,0.00013164358,0.00010576093,0.9682484,0.0044362983,0.011728291,0.014678468,0.000040993946],"about_ca_topic_score_codex":0.018441197,"about_ca_topic_score_gemma":0.021883456,"teacher_disagreement_score":0.018441197,"about_ca_system_score_codex":0.001253609,"about_ca_system_score_gemma":0.0025364284,"threshold_uncertainty_score":0.036667705},"labels":[],"label_agreement":null},{"id":"W2583764785","doi":"10.1109/icmla.2016.0089","title":"A Parallel K-Medoids Algorithm for Clustering based on MapReduce","year":2016,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Cluster analysis; Computer science; CURE data clustering algorithm; Data stream clustering; Medoid; Canopy clustering algorithm; Correlation clustering; Data mining; Big data; Algorithm; Scale (ratio); Parallel computing; Machine learning","score_opus":0.028868289273536685,"score_gpt":0.3075938819899316,"score_spread":0.2787255927163949,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2583764785","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031061547,0.000219272,0.99323386,0.00017874915,0.00011713269,0.00012627747,0.00015290246,0.0014410652,0.0014246295],"genre_scores_gemma":[0.06728834,0.0002400875,0.928133,0.0001323124,0.000073186595,0.0002832861,0.0006436181,0.00032936124,0.0028767877],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984572,0.00030134147,0.00010518688,0.0003778364,0.0006337189,0.00012474823],"domain_scores_gemma":[0.999172,0.00019648242,0.000056837183,0.00017681284,0.00032707385,0.000070816895],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011211081,0.0011494451,0.0018167987,0.001499812,0.0018315084,0.0014511317,0.0026722588,0.0012266261,0.0029727987],"category_scores_gemma":[0.0027759336,0.00073719665,0.0019280711,0.0020534298,0.0008759779,0.0017028498,0.0018822362,0.0017979607,0.0024753744],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00061718666,0.00024929785,0.0015450879,0.00065155025,0.00037995042,0.0003121257,0.00072283746,0.42268226,0.020532759,0.027161662,0.029968748,0.49517658],"study_design_scores_gemma":[0.00009454657,0.00011745401,0.0005354789,0.000028810175,0.00003975705,0.00037517096,0.000269601,0.92973703,0.0102489665,0.029272031,0.029187182,0.00009402801],"about_ca_topic_score_codex":0.010252322,"about_ca_topic_score_gemma":0.010731137,"teacher_disagreement_score":0.010252322,"about_ca_system_score_codex":0.0009724286,"about_ca_system_score_gemma":0.0024867153,"threshold_uncertainty_score":0.020385325},"labels":[],"label_agreement":null},{"id":"W2597766798","doi":"","title":"NP-Hardness of Euclidean Sum-of-Squares Clustering","year":2008,"lang":"fr","type":"article","venue":"PolyPublie (École Polytechnique de Montréal)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Group for Research in Decision Analysis; HEC Montréal; Polytechnique Montréal","funders":"","keywords":"Cluster analysis; Mathematics; Combinatorics; Euclidean distance; Explained sum of squares; Computer science; Artificial intelligence; Statistics","score_opus":0.03416253621654149,"score_gpt":0.2831926804146098,"score_spread":0.24903014419806832,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2597766798","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.076472804,0.0058124107,0.82307166,0.026668904,0.0012685528,0.00029988986,0.0064631104,0.0028405718,0.05710213],"genre_scores_gemma":[0.65345734,0.004492495,0.29108632,0.006800981,0.00169983,0.0007988383,0.010754605,0.0019261892,0.0289835],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9934708,0.001703542,0.00033540858,0.002121936,0.0016577182,0.0007106921],"domain_scores_gemma":[0.97835577,0.016268842,0.00067259005,0.0026442634,0.0014213971,0.00063724496],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002720662,0.001322334,0.0030365991,0.0007700932,0.0022588735,0.004727173,0.0045100837,0.0036755793,0.008047415],"category_scores_gemma":[0.025190758,0.0012695151,0.0022134485,0.0023233425,0.0031863553,0.009561331,0.0040646074,0.0068504736,0.002167073],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013313451,0.00035624378,0.0028857894,0.0021343534,0.000478064,0.00087214337,0.000991522,0.29655522,0.0069910707,0.33559433,0.16342182,0.18838815],"study_design_scores_gemma":[0.00011881091,0.000046027828,0.0007001035,0.00008088812,0.000051355542,0.00042233986,0.00023474831,0.24777511,0.0027539048,0.7332753,0.014499961,0.00004132299],"about_ca_topic_score_codex":0.00355538,"about_ca_topic_score_gemma":0.0036827773,"teacher_disagreement_score":0.008047415,"about_ca_system_score_codex":0.002733021,"about_ca_system_score_gemma":0.0024947869,"threshold_uncertainty_score":0.026921213},"labels":[],"label_agreement":null},{"id":"W2599038115","doi":"10.5430/air.v6n2p10","title":"Bio-inspired multiobjective clustering optimization: A survey and a proposal","year":2017,"lang":"en","type":"article","venue":"Artificial Intelligence Research","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Cluster analysis; Multi-objective optimization; Computer science; Data mining; Quality (philosophy); Mathematical optimization; Machine learning; Mathematics","score_opus":0.2304565960268817,"score_gpt":0.45474251321250253,"score_spread":0.22428591718562083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2599038115","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005748317,0.48102677,0.49397916,0.0017321621,0.0007685473,0.0001320262,0.00015412935,0.00029277607,0.016166087],"genre_scores_gemma":[0.08768076,0.60058534,0.29808867,0.0011909707,0.0020262313,0.0003950262,0.0006635329,0.00021568255,0.009153802],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99955493,0.00009332099,0.000044975885,0.000107985325,0.00017116136,0.000027745093],"domain_scores_gemma":[0.99963975,0.00015386425,0.000028882292,0.000027948943,0.00012524163,0.000024330842],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000912979,0.0012064725,0.0012637591,0.0022969532,0.0004841658,0.0018038925,0.0012554678,0.0014733515,0.0014836794],"category_scores_gemma":[0.0010978711,0.00054185325,0.0011223386,0.0048917476,0.000637814,0.0014580354,0.00089701446,0.0012561801,0.0010532832],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000093498165,0.00021967449,0.0016631695,0.005161556,0.00022678675,0.00024770756,0.0002438186,0.085192904,0.0052531715,0.056921396,0.013967837,0.83080846],"study_design_scores_gemma":[0.00004324075,0.00044494055,0.004022249,0.0025668603,0.00030689684,0.00159558,0.0005107921,0.41106915,0.007938749,0.09596424,0.4752775,0.00025991356],"about_ca_topic_score_codex":0.0015821726,"about_ca_topic_score_gemma":0.001051695,"teacher_disagreement_score":0.0022969532,"about_ca_system_score_codex":0.00066408515,"about_ca_system_score_gemma":0.0008525151,"threshold_uncertainty_score":0.004963398},"labels":[],"label_agreement":null},{"id":"W2606056073","doi":"10.1016/j.ijar.2017.04.004","title":"Collaborative fuzzy clustering algorithm: Some refinements","year":2017,"lang":"en","type":"article","venue":"International Journal of Approximate Reasoning","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Cluster analysis; Partition (number theory); Computer science; Data mining; Fuzzy clustering; Fuzzy logic; Algorithm; Granular computing; Artificial intelligence; Mathematics; Rough set","score_opus":0.017989009181289577,"score_gpt":0.3372030557690087,"score_spread":0.3192140465877191,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2606056073","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033045306,0.0008676832,0.9932099,0.00033144897,0.00016405694,0.000048630092,0.000062061554,0.000098798926,0.0019129023],"genre_scores_gemma":[0.09565048,0.0017909594,0.8965848,0.00023290169,0.0005253028,0.00018552133,0.0002737047,0.00011769288,0.004638556],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9951415,0.0015481819,0.00039925144,0.0011282876,0.0016018312,0.00018082687],"domain_scores_gemma":[0.9905309,0.003295589,0.0002622847,0.0021366214,0.0035221959,0.00025240637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0067800223,0.0009117055,0.0021034991,0.0023184381,0.0018862177,0.0023832868,0.0047929217,0.00272255,0.0040769936],"category_scores_gemma":[0.022434935,0.000756767,0.002526049,0.00489074,0.0019311605,0.0057634125,0.0028247072,0.0030451398,0.0019507377],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046678886,0.00028036517,0.003046378,0.00063702825,0.0003294644,0.0002215928,0.0009965189,0.17276525,0.0068945033,0.2914388,0.01274453,0.51017874],"study_design_scores_gemma":[0.000052801828,0.00013748281,0.00092383276,0.00005792215,0.00012320142,0.00042204518,0.00017122188,0.8823673,0.0033280775,0.098967336,0.013370276,0.00007845586],"about_ca_topic_score_codex":0.011991781,"about_ca_topic_score_gemma":0.009310338,"teacher_disagreement_score":0.011991781,"about_ca_system_score_codex":0.0015163204,"about_ca_system_score_gemma":0.0024995368,"threshold_uncertainty_score":0.035856664},"labels":[],"label_agreement":null},{"id":"W2612181456","doi":"10.1007/978-1-4899-7502-7_35-1","title":"Categorical Data Clustering","year":2016,"lang":"en","type":"book-chapter","venue":"Encyclopedia of Machine Learning and Data Mining","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Data mining; Domain (mathematical analysis); Cluster (spacecraft); Artificial intelligence; Machine learning; Mathematics","score_opus":0.05026195322463737,"score_gpt":0.32755562043522746,"score_spread":0.27729366721059007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2612181456","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016212984,0.004119958,0.92517805,0.000792303,0.0009177334,0.0002614647,0.0036722347,0.007820287,0.055616718],"genre_scores_gemma":[0.022615138,0.004992927,0.87993354,0.0006020788,0.0005653283,0.00041848962,0.015864637,0.0014571063,0.07355078],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99862075,0.00015928231,0.00009331089,0.00035501042,0.0007098349,0.00006170809],"domain_scores_gemma":[0.9985606,0.00024224706,0.000044603225,0.00045593793,0.00063126703,0.00006529984],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001025443,0.0010042867,0.0011080117,0.0036361215,0.0011845957,0.0026522384,0.002135462,0.00083062745,0.030745767],"category_scores_gemma":[0.0033891262,0.0005975033,0.0012637507,0.007035077,0.00059963384,0.0018073767,0.002015249,0.0016367369,0.034055024],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040357852,0.000046952795,0.00047714825,0.00034360884,0.00004593951,0.0000486726,0.00012483879,0.0037150565,0.0033813415,0.051755603,0.15095434,0.78906614],"study_design_scores_gemma":[0.000014799488,0.00006364826,0.0018806714,0.00022168864,0.000051673032,0.0009043564,0.00019129919,0.05505541,0.009964566,0.17415443,0.75741136,0.000086107306],"about_ca_topic_score_codex":0.0016597337,"about_ca_topic_score_gemma":0.0025212108,"teacher_disagreement_score":0.030745767,"about_ca_system_score_codex":0.0010128991,"about_ca_system_score_gemma":0.0014899572,"threshold_uncertainty_score":0.10285485},"labels":[],"label_agreement":null},{"id":"W2623118547","doi":"10.5120/ijca2016912481","title":"A Survey on Unsupervised Clustering Algorithm based on K-Means Clustering","year":2016,"lang":"en","type":"article","venue":"International Journal of Computer Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Optech (Canada)","funders":"","keywords":"Computer science; Cluster analysis; Data mining; Artificial intelligence","score_opus":0.030654955753429088,"score_gpt":0.32357217122990406,"score_spread":0.292917215476475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2623118547","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005302323,0.03083392,0.94351673,0.0007116323,0.00076125335,0.00023072207,0.000530137,0.0017372994,0.016376037],"genre_scores_gemma":[0.09154984,0.064084835,0.81995684,0.0007468493,0.0013413525,0.00072172843,0.0035274343,0.0007219074,0.017349133],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9978339,0.00035930268,0.00020731887,0.0005563293,0.00095468824,0.00008841885],"domain_scores_gemma":[0.9987212,0.00028041683,0.000075187665,0.00013984925,0.0007445695,0.00003865851],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011241778,0.0013386989,0.001817572,0.003861995,0.0016087252,0.002127215,0.002020127,0.0016072752,0.004124583],"category_scores_gemma":[0.0028795404,0.0005950672,0.0017090711,0.007583594,0.000763842,0.0028089,0.0010071343,0.0010657259,0.0037784884],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014086216,0.00014333922,0.0035951824,0.0020466521,0.00031251204,0.00020143061,0.00033561714,0.041296866,0.005107487,0.03471466,0.03442912,0.8776763],"study_design_scores_gemma":[0.00008064125,0.0002653818,0.008121895,0.00096863956,0.00032466394,0.0021989124,0.00075199903,0.39768478,0.020151818,0.07512599,0.49391297,0.00041223495],"about_ca_topic_score_codex":0.0064226226,"about_ca_topic_score_gemma":0.00431178,"teacher_disagreement_score":0.0064226226,"about_ca_system_score_codex":0.001106117,"about_ca_system_score_gemma":0.0020623293,"threshold_uncertainty_score":0.013798118},"labels":[],"label_agreement":null},{"id":"W27100606","doi":"10.1111/jgs.14068","title":"New data induced metric for density based clustering","year":2004,"lang":"en","type":"article","venue":"Journal of the American Geriatrics Society","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Morris Kerzner Memorial Fund","keywords":"Cluster analysis; Metric (unit); Computer science; Data mining; Mathematics; Artificial intelligence; Business","score_opus":0.056767522146321144,"score_gpt":0.3413801901686271,"score_spread":0.28461266802230595,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W27100606","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038944208,0.0003272828,0.99416625,0.00020683142,0.0001242178,0.0000844749,0.00031980162,0.00030479586,0.00057197484],"genre_scores_gemma":[0.15213819,0.00078799686,0.8393022,0.00023402448,0.00031875062,0.0007859326,0.0033258018,0.00037061982,0.0027364944],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9921934,0.0024337291,0.0006343109,0.0009773021,0.0035352749,0.00022602196],"domain_scores_gemma":[0.99060297,0.0026584072,0.00068176014,0.0018245247,0.0037517305,0.00048068148],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034790027,0.0011730024,0.0021626183,0.0041969772,0.0014397495,0.003093535,0.0036553005,0.0018796374,0.0018062071],"category_scores_gemma":[0.023682166,0.00071261206,0.0013533818,0.005851228,0.0011856057,0.004673546,0.004352845,0.0028864888,0.0010643551],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005404849,0.00034801697,0.0059343306,0.0007073351,0.00037458853,0.00025799262,0.00042337726,0.23825645,0.012154799,0.19976225,0.02382281,0.51741755],"study_design_scores_gemma":[0.000023832406,0.000113892194,0.0011728875,0.00003557621,0.000036805595,0.00023091769,0.00006560124,0.9377287,0.0033650852,0.04776582,0.009407157,0.000053577212],"about_ca_topic_score_codex":0.0031778975,"about_ca_topic_score_gemma":0.0037635665,"teacher_disagreement_score":0.0041969772,"about_ca_system_score_codex":0.0032028197,"about_ca_system_score_gemma":0.002455879,"threshold_uncertainty_score":0.023238242},"labels":[],"label_agreement":null},{"id":"W2745587906","doi":"10.1007/978-3-319-65636-6_27","title":"Enhancing Ant Brood Clustering with Adaptive Radius of Perception and Non-parametric Estimation on Multi-core Architectures","year":2017,"lang":"en","type":"book-chapter","venue":"Lecture notes on data engineering and communications technologies","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Cluster analysis; Computer science; CUDA; Benchmark (surveying); Convergence (economics); Ant colony; Ant colony optimization algorithms; Data mining; Artificial intelligence; Machine learning; Parallel computing","score_opus":0.0583825477258629,"score_gpt":0.30550924447221084,"score_spread":0.24712669674634793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2745587906","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015133201,0.00022640842,0.98189396,0.0000528848,0.00006063827,0.000022813336,0.000015064982,0.00064896897,0.0019460409],"genre_scores_gemma":[0.480203,0.00021605485,0.5138882,0.00007892353,0.000056523648,0.0000745738,0.00010632483,0.00026096095,0.005115523],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99940825,0.00012826304,0.000026199277,0.00014393232,0.00022659631,0.00006679357],"domain_scores_gemma":[0.9990219,0.00027461717,0.00006898185,0.00018792276,0.00039482204,0.000051877174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000579086,0.0008099716,0.0009223955,0.000627962,0.0005602767,0.0008918861,0.0016293725,0.0008966244,0.001610533],"category_scores_gemma":[0.002369404,0.00040240062,0.00058128027,0.0008091662,0.000424653,0.0012250212,0.0011687401,0.00082845485,0.0007283974],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032533568,0.00016308048,0.00070974446,0.00013764536,0.00008060468,0.00009861181,0.0001836582,0.56869495,0.040629342,0.01291563,0.0035764172,0.37248498],"study_design_scores_gemma":[0.00000506178,0.000028372375,0.00011889243,0.0000023425478,0.0000060956077,0.00002757582,0.000010398388,0.99495846,0.0026695994,0.0015772558,0.0005886208,0.000007260948],"about_ca_topic_score_codex":0.0036985928,"about_ca_topic_score_gemma":0.005780394,"teacher_disagreement_score":0.0036985928,"about_ca_system_score_codex":0.0006324412,"about_ca_system_score_gemma":0.00085122127,"threshold_uncertainty_score":0.0073541403},"labels":[],"label_agreement":null},{"id":"W2750557327","doi":"10.1016/j.ijar.2017.08.008","title":"Fuzzy C-Means clustering based on dual expression between cluster prototypes and reconstructed data","year":2017,"lang":"en","type":"article","venue":"International Journal of Approximate Reasoning","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Dual (grammatical number); Cluster analysis; Cluster (spacecraft); Fuzzy logic; Fuzzy clustering; Computer science; Pattern recognition (psychology); Artificial intelligence; Data mining; Expression (computer science); Computer network; Art","score_opus":0.049615045357977505,"score_gpt":0.3513245074083753,"score_spread":0.3017094620503978,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2750557327","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016844545,0.0001664225,0.98177123,0.000094711526,0.000036825684,0.00003865343,0.00003138611,0.00012793124,0.000888256],"genre_scores_gemma":[0.3627993,0.00026468895,0.6340062,0.000085268155,0.00006504203,0.00016409617,0.0002842548,0.00009300135,0.0022381616],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976555,0.00045776786,0.00017727756,0.00059565366,0.0009348248,0.00017897494],"domain_scores_gemma":[0.997928,0.00048603612,0.00015840559,0.00022967656,0.0011283184,0.00006940921],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026551322,0.00069599017,0.0012140505,0.0017498594,0.0012873417,0.0023548454,0.0023935328,0.0017028878,0.0012315671],"category_scores_gemma":[0.008324402,0.00047371883,0.0013128307,0.001867942,0.0014359957,0.0022280016,0.001793401,0.0013332621,0.0004054039],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00095133815,0.00015034952,0.002821021,0.0004709299,0.00026632953,0.0003017532,0.00092907797,0.37634024,0.03529386,0.10204025,0.0036862097,0.47674865],"study_design_scores_gemma":[0.000013077873,0.00003339284,0.0003456055,0.000021366852,0.000032408414,0.00009595961,0.000060853872,0.98359364,0.004674264,0.01028448,0.00081777753,0.000027174088],"about_ca_topic_score_codex":0.00608891,"about_ca_topic_score_gemma":0.0036084037,"teacher_disagreement_score":0.00608891,"about_ca_system_score_codex":0.0014616109,"about_ca_system_score_gemma":0.0020125543,"threshold_uncertainty_score":0.014041841},"labels":[],"label_agreement":null},{"id":"W2754977413","doi":"","title":"Standardization and Weighting of Variables for the Fuzzy K-Means Clustering of Discontinuity Data","year":2000,"lang":"en","type":"article","venue":"4th North American Rock Mechanics Symposium","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Golder Associates (Canada)","funders":"","keywords":"Standardization; Weighting; Cluster analysis; Data mining; Computer science; Fuzzy logic; Discontinuity (linguistics); Fuzzy clustering; Mathematics; Artificial intelligence; Medicine","score_opus":0.018446712484574013,"score_gpt":0.2747637271969266,"score_spread":0.2563170147123526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2754977413","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019874945,0.00032284993,0.9790141,0.00011787828,0.00005610167,0.00008609203,0.00009536638,0.00010958017,0.0003231327],"genre_scores_gemma":[0.1756906,0.00037600083,0.82227945,0.000040571023,0.000070556045,0.00029091994,0.00046495433,0.00015937304,0.0006276549],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9922621,0.003693508,0.0009767911,0.0011045876,0.0016252171,0.00033762894],"domain_scores_gemma":[0.9860282,0.005292679,0.0011784459,0.0031744207,0.003984996,0.00034122687],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018805785,0.0008305229,0.0015150266,0.0038650872,0.0018801839,0.0031946946,0.0020975173,0.0014724454,0.0012741624],"category_scores_gemma":[0.052729454,0.0005480941,0.0020409836,0.005585299,0.0020732493,0.0029693565,0.0029475316,0.0019700862,0.00044355806],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005779908,0.00015526991,0.0068096984,0.0005497651,0.00030143606,0.00007183351,0.0011755308,0.12819889,0.009711582,0.17477399,0.0030038736,0.67467],"study_design_scores_gemma":[0.000087340304,0.00022511906,0.00815026,0.00024386529,0.00021905312,0.00014551256,0.0005814624,0.77892774,0.008643315,0.19396801,0.008638087,0.00017022691],"about_ca_topic_score_codex":0.0065070246,"about_ca_topic_score_gemma":0.006974019,"teacher_disagreement_score":0.018805785,"about_ca_system_score_codex":0.002018497,"about_ca_system_score_gemma":0.0036245342,"threshold_uncertainty_score":0.099455714},"labels":[],"label_agreement":null},{"id":"W2755594639","doi":"10.1109/trustcom/bigdatase/icess.2017.332","title":"A Data Science and Engineering Solution for Fast K-Means Clustering of Big Data","year":2017,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Big data; Scalability; Computer science; Cluster analysis; Data mining; Science and engineering; Heuristic; Task (project management); Variety (cybernetics); Centroid; Data science; Database; Artificial intelligence; Engineering","score_opus":0.19239062046661298,"score_gpt":0.3746471718461937,"score_spread":0.18225655137958074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2755594639","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019409481,0.00016472721,0.99524885,0.00029315517,0.00011130853,0.0000827675,0.00006964979,0.0009806826,0.0011079727],"genre_scores_gemma":[0.027150765,0.0001558642,0.9703569,0.00011065727,0.00005752513,0.00017755576,0.00027491694,0.00011786405,0.0015979569],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984863,0.00022746394,0.00009318033,0.00036432012,0.0007191102,0.00010967437],"domain_scores_gemma":[0.99876034,0.00022159962,0.000099912875,0.0002938673,0.00054103485,0.00008330196],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012134289,0.0009381813,0.0010133268,0.0016755046,0.0017379682,0.0015791157,0.002551228,0.0027541472,0.004297047],"category_scores_gemma":[0.0055623106,0.0006937099,0.0012746261,0.002898339,0.0008775573,0.0023611113,0.0018816116,0.0018083458,0.0028924032],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027647367,0.00026287502,0.0013294806,0.00054710376,0.00014744911,0.00031764258,0.0003263994,0.19343469,0.014098807,0.070952415,0.031146584,0.6871601],"study_design_scores_gemma":[0.00008419108,0.00012468174,0.00042325354,0.000039212573,0.000031257947,0.00053904107,0.00016171926,0.91175985,0.0070505138,0.050153088,0.029587261,0.000046003122],"about_ca_topic_score_codex":0.0039534504,"about_ca_topic_score_gemma":0.0052280133,"teacher_disagreement_score":0.004297047,"about_ca_system_score_codex":0.0009470125,"about_ca_system_score_gemma":0.002505592,"threshold_uncertainty_score":0.014375031},"labels":[],"label_agreement":null},{"id":"W2762595523","doi":"10.1007/s10115-017-1110-9","title":"Subspace multi-clustering: a review","year":2017,"lang":"en","type":"review","venue":"Knowledge and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":41,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Cluster analysis; Computer science; Clustering high-dimensional data; Correlation clustering; CURE data clustering algorithm; Data mining; Consensus clustering; Canopy clustering algorithm; Data stream clustering; Fuzzy clustering; Constrained clustering; Artificial intelligence; Brown clustering; Pattern recognition (psychology); Machine learning","score_opus":0.16548597611151802,"score_gpt":0.43143692051603333,"score_spread":0.2659509444045153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2762595523","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012459,0.9975314,0.0010698198,0.00038354215,0.00026050772,0.000008283933,0.000044054985,0.000015643063,0.0005620815],"genre_scores_gemma":[0.0010731893,0.9965216,0.0013963103,0.00027713957,0.00039439995,0.000012545137,0.00008359187,0.000006058863,0.00023510505],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.999483,0.000100562575,0.00007846512,0.00013007176,0.00017494925,0.000032862583],"domain_scores_gemma":[0.9982779,0.0008915665,0.00016673138,0.000069716414,0.0005021219,0.00009201267],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014667364,0.0013249909,0.00258246,0.0039566266,0.00043218918,0.0021893368,0.0017622445,0.0016470506,0.00374628],"category_scores_gemma":[0.0037692508,0.0005089733,0.00094803446,0.007532904,0.00085279206,0.0028034453,0.001205236,0.0013606728,0.0023152572],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007994848,0.00007318317,0.00048668394,0.026126184,0.0002792771,0.00011063687,0.000058251633,0.00086409185,0.000543758,0.004970846,0.04801681,0.9183903],"study_design_scores_gemma":[0.000058530568,0.00015717057,0.002022639,0.012765466,0.000872759,0.0019722397,0.00020357333,0.0013656837,0.00079625036,0.01287257,0.9667932,0.00011999435],"about_ca_topic_score_codex":0.0023639032,"about_ca_topic_score_gemma":0.0035834287,"teacher_disagreement_score":0.0039566266,"about_ca_system_score_codex":0.0008517377,"about_ca_system_score_gemma":0.0026316326,"threshold_uncertainty_score":0.012532532},"labels":[],"label_agreement":null},{"id":"W2772904413","doi":"10.1109/tnnls.2017.2770167","title":"An Algorithm for Clustering Categorical Data With Set-Valued Features","year":2017,"lang":"en","type":"article","venue":"IEEE Transactions on Neural Networks and Learning Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Science Foundation of Shanxi Province; Simon Fraser University; Kungliga Tekniska Högskolan; Shanxi Scholarship Council of China; National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Categorical variable; Data mining; Initialization; Algorithm; Set (abstract data type); CURE data clustering algorithm; Canopy clustering algorithm; Data set; Determining the number of clusters in a data set; Data stream clustering; Heuristic; Correlation clustering; Pattern recognition (psychology); Artificial intelligence; Machine learning","score_opus":0.04585140131797229,"score_gpt":0.32887373130401953,"score_spread":0.28302232998604726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2772904413","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025879142,0.00007127575,0.996289,0.000063820575,0.000029127239,0.00009949491,0.00008512141,0.00050792,0.0002662652],"genre_scores_gemma":[0.028558092,0.000060094815,0.9700918,0.0000508367,0.000021029862,0.00028194103,0.0003301852,0.000067603156,0.00053839915],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980597,0.0004785909,0.00018745016,0.00051481964,0.00065456494,0.00010474557],"domain_scores_gemma":[0.9975873,0.00082766835,0.00022657379,0.0003346443,0.0009291058,0.00009469551],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023433943,0.0010079416,0.001203255,0.0032767914,0.0014287512,0.001349667,0.0029249527,0.0015624779,0.0019666548],"category_scores_gemma":[0.0076325033,0.00061802915,0.0013575488,0.0041925753,0.0009660269,0.0019326509,0.0018889029,0.0019344747,0.0012939163],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028523616,0.00013144065,0.003544597,0.00019250432,0.0001560286,0.00011453464,0.00054137997,0.1414619,0.006703022,0.027957022,0.009582253,0.8093301],"study_design_scores_gemma":[0.000069064685,0.00010194083,0.0010881064,0.000038595856,0.00002901592,0.00027557803,0.0001527266,0.95369935,0.004045152,0.03313155,0.0073145116,0.000054402775],"about_ca_topic_score_codex":0.0057319477,"about_ca_topic_score_gemma":0.0061858366,"teacher_disagreement_score":0.0057319477,"about_ca_system_score_codex":0.0012028081,"about_ca_system_score_gemma":0.0022491512,"threshold_uncertainty_score":0.012393236},"labels":[],"label_agreement":null},{"id":"W2778728785","doi":"10.5430/air.v7n1p15","title":"Estimating the number of clusters using diversity","year":2017,"lang":"en","type":"article","venue":"Artificial Intelligence Research","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Outlier; Silhouette; Computer science; Pattern recognition (psychology); Entropy (arrow of time); Single-linkage clustering; Cluster (spacecraft); Ground truth; Artificial intelligence; Statistic; Mathematics; Data mining; Fuzzy clustering; Statistics; CURE data clustering algorithm; Physics","score_opus":0.36841764293722823,"score_gpt":0.505386192201522,"score_spread":0.1369685492642938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2778728785","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04885091,0.0011991546,0.94624096,0.00042192958,0.00005654639,0.00011030344,0.00037401533,0.0006102608,0.0021359085],"genre_scores_gemma":[0.5717149,0.0006481977,0.4236878,0.0002672323,0.00023841643,0.00027359524,0.0015592161,0.0002827335,0.0013278943],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9933484,0.0017989469,0.0004785802,0.0021410892,0.0018670228,0.0003658032],"domain_scores_gemma":[0.9785373,0.012902022,0.0022992357,0.0023483906,0.0032950395,0.00061804004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006803065,0.0013756996,0.0024409462,0.00727645,0.0026267448,0.003362668,0.0026935325,0.0026303513,0.0008944196],"category_scores_gemma":[0.033861555,0.0008780399,0.0014765644,0.0041326117,0.0022708457,0.0042095953,0.0041977847,0.0021730578,0.0007740223],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008462392,0.00018876235,0.04351984,0.0005925383,0.0008096377,0.0004250587,0.0017404568,0.47747526,0.012769797,0.042860847,0.0117483195,0.40702325],"study_design_scores_gemma":[0.00006073234,0.00008985945,0.0055908696,0.00012247766,0.00007779424,0.00036235157,0.0004365485,0.90423745,0.005376069,0.07912745,0.004394055,0.00012434935],"about_ca_topic_score_codex":0.0040104818,"about_ca_topic_score_gemma":0.005037246,"teacher_disagreement_score":0.00727645,"about_ca_system_score_codex":0.001955168,"about_ca_system_score_gemma":0.0017798117,"threshold_uncertainty_score":0.035978496},"labels":[],"label_agreement":null},{"id":"W2783873704","doi":"10.1186/s12859-017-1996-y","title":"diceR: an R package for class discovery using an ensemble driven approach","year":2018,"lang":"en","type":"article","venue":"BMC Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"BC Cancer Foundation","keywords":"Computer science; Dicer; Cluster analysis; Context (archaeology); Data mining; Set (abstract data type); Class (philosophy); Permutation (music); Generalization; Cluster (spacecraft); Machine learning; Artificial intelligence; Data science; Biology; Mathematics","score_opus":0.08579926171176273,"score_gpt":0.34100829396308385,"score_spread":0.25520903225132113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2783873704","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013707225,0.0010630309,0.8547183,0.00067263906,0.00055472355,0.00081600685,0.03219429,0.10616066,0.0024496529],"genre_scores_gemma":[0.013646309,0.0006818832,0.92482424,0.0007313746,0.0002745494,0.0047552823,0.021339044,0.031765163,0.0019821783],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98375136,0.0075297314,0.001727452,0.0034001975,0.0031583044,0.0004330793],"domain_scores_gemma":[0.9297916,0.048568424,0.0048346166,0.008724692,0.0070000384,0.0010805819],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.023153238,0.003914279,0.0047044144,0.009180763,0.0022751638,0.0063226903,0.007729058,0.002847473,0.061650783],"category_scores_gemma":[0.10757555,0.003794423,0.0059009735,0.0066715917,0.0023406758,0.004112719,0.0063874405,0.0076840254,0.03254007],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010742848,0.00014672539,0.0074288487,0.0066146837,0.004286401,0.00059772795,0.0011298931,0.02779114,0.004328352,0.04227517,0.63740057,0.26692635],"study_design_scores_gemma":[0.0009468817,0.00023107047,0.005912778,0.0010664475,0.0014004378,0.001232554,0.0003029339,0.15832351,0.01158805,0.19554465,0.6228955,0.0005551748],"about_ca_topic_score_codex":0.004472909,"about_ca_topic_score_gemma":0.0065453704,"teacher_disagreement_score":0.061650783,"about_ca_system_score_codex":0.0016863173,"about_ca_system_score_gemma":0.007401756,"threshold_uncertainty_score":0.20624244},"labels":[],"label_agreement":null},{"id":"W2786121513","doi":"10.5539/mas.v12n2p116","title":"An Improved Version of K-medoid Algorithm using CRO","year":2018,"lang":"en","type":"article","venue":"Modern Applied Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Medoid; Cluster analysis; Computer science; Algorithm; k-means clustering; Disjoint sets; Data mining; DBSCAN; Cluster (spacecraft); Outlier; Correlation clustering; Artificial intelligence; CURE data clustering algorithm; Mathematics","score_opus":0.02065885472170116,"score_gpt":0.3194668025855861,"score_spread":0.2988079478638849,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2786121513","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023974774,0.00044432993,0.99452233,0.00015558761,0.00009795359,0.00011784793,0.00015278606,0.0007605142,0.0013511854],"genre_scores_gemma":[0.056681287,0.00043543262,0.93761826,0.00022951073,0.00007586823,0.00033489702,0.0008608922,0.00035434403,0.003409452],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9957514,0.0013008446,0.00035735156,0.0011992361,0.0011532165,0.00023803009],"domain_scores_gemma":[0.9963219,0.0011163367,0.00030491455,0.00055972993,0.0015800917,0.0001170495],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030173396,0.0017316658,0.003102316,0.003329354,0.0021744033,0.0024216764,0.0039103488,0.003183449,0.0050557246],"category_scores_gemma":[0.009453474,0.00089748856,0.0029369364,0.004291497,0.0012123667,0.0034843446,0.0025410773,0.002809779,0.0042919456],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006212068,0.00017900612,0.0026713521,0.000778951,0.00043427176,0.00021833535,0.0006240636,0.3726374,0.012952268,0.029732669,0.019969516,0.5591809],"study_design_scores_gemma":[0.000109238346,0.00014629729,0.0006110801,0.00010043163,0.00006737049,0.0005009256,0.00024038683,0.93892086,0.0076604206,0.025020067,0.02649593,0.00012716831],"about_ca_topic_score_codex":0.009041762,"about_ca_topic_score_gemma":0.007525991,"teacher_disagreement_score":0.009041762,"about_ca_system_score_codex":0.0012517412,"about_ca_system_score_gemma":0.002956792,"threshold_uncertainty_score":0.017978251},"labels":[],"label_agreement":null},{"id":"W2788096725","doi":"10.1145/3178876.3185993","title":"ROSC","year":2018,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Spectral clustering; Similarity (geometry); Cluster analysis; Reachability; Data mining; Scale (ratio); Feature (linguistics); Competitor analysis; Artificial intelligence; Algorithm","score_opus":0.019395380790563112,"score_gpt":0.3199427341705084,"score_spread":0.30054735337994526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2788096725","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10139687,0.0029677397,0.79061246,0.0016285797,0.0014667668,0.0010929961,0.009836008,0.034056973,0.05694152],"genre_scores_gemma":[0.32492533,0.0009487765,0.61517,0.0014242427,0.00035819446,0.00063433533,0.028924502,0.0040365765,0.023577938],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974113,0.00047305148,0.00012294768,0.0007337694,0.0010458722,0.00021313867],"domain_scores_gemma":[0.99480915,0.0012691704,0.00032437552,0.0017477991,0.0016322117,0.00021730427],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0025880544,0.0012392824,0.0008659327,0.0030821136,0.0013724283,0.0017962005,0.0030170598,0.0015508662,0.013668132],"category_scores_gemma":[0.009219609,0.0003184112,0.00093753386,0.002593957,0.0007401832,0.0025549498,0.001967503,0.0010524562,0.0105007235],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009334458,0.00059835805,0.009471256,0.00084503944,0.00047466153,0.00022945121,0.00030918673,0.17409556,0.011844083,0.051844038,0.16494048,0.5844145],"study_design_scores_gemma":[0.00017444392,0.0003609512,0.0034912615,0.00007501552,0.00007419376,0.0004856949,0.000279794,0.84481984,0.021806693,0.0254157,0.10293741,0.000079108955],"about_ca_topic_score_codex":0.0077866567,"about_ca_topic_score_gemma":0.010456275,"teacher_disagreement_score":0.9863319,"about_ca_system_score_codex":0.0013049913,"about_ca_system_score_gemma":0.001248676,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2792821451","doi":"10.1007/s10898-018-0634-1","title":"A sampling-based exact algorithm for the solution of the minimax diameter clustering problem","year":2018,"lang":"en","type":"article","venue":"Journal of Global Optimization","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"École de Technologie Supérieure; Université du Québec à Montréal; Polytechnique Montréal","funders":"Fonds Québécois de la Recherche sur la Nature et les Technologies; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Mathematics; Cluster analysis; Minimax; Heuristic; Mathematical optimization; Algorithm; Set (abstract data type); Sampling (signal processing); Computer science; Statistics","score_opus":0.0290985115856251,"score_gpt":0.31785168437011047,"score_spread":0.28875317278448537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792821451","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026878251,0.00020176056,0.9954535,0.0001086869,0.00006587474,0.00005299609,0.000030159943,0.00029409648,0.0011051445],"genre_scores_gemma":[0.087860204,0.00025396293,0.9084973,0.0001744791,0.00011116541,0.00036485295,0.00021588197,0.00018783593,0.002334351],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99891126,0.00031083435,0.00005190265,0.00018579292,0.000447191,0.000093012546],"domain_scores_gemma":[0.9972754,0.0016789001,0.00013734156,0.00025252404,0.000515147,0.00014067913],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021441989,0.0011562648,0.0018733124,0.0013261916,0.0009340379,0.0011909532,0.002759333,0.0023993899,0.005263855],"category_scores_gemma":[0.01045594,0.0007797606,0.00088589813,0.0018074465,0.0011020913,0.0017588959,0.0023371754,0.0023391184,0.0014479554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036803348,0.00015325073,0.00067768776,0.00024133072,0.00007347618,0.000076609984,0.00020480009,0.6940311,0.0040987018,0.039481275,0.008168845,0.25242484],"study_design_scores_gemma":[0.000038623708,0.000032591866,0.00008649044,0.000011256853,0.000006047996,0.000031223597,0.000015294332,0.98911893,0.00036969795,0.00924309,0.0010387235,0.000007983797],"about_ca_topic_score_codex":0.0060515874,"about_ca_topic_score_gemma":0.007410464,"teacher_disagreement_score":0.0060515874,"about_ca_system_score_codex":0.0014492713,"about_ca_system_score_gemma":0.002775816,"threshold_uncertainty_score":0.017609358},"labels":[],"label_agreement":null},{"id":"W2801522993","doi":"10.1155/2018/6164534","title":"Application of Data Clustering to Railway Delay Pattern Recognition","year":2018,"lang":"en","type":"article","venue":"Journal of Advanced Transportation","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Danmarks Tekniske Universitet","keywords":"Cluster analysis; Computer science; Data mining; Pattern recognition (psychology); Artificial intelligence; Transport engineering; Engineering","score_opus":0.04256832023106765,"score_gpt":0.33834660307763686,"score_spread":0.2957782828465692,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2801522993","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.086416334,0.00036861876,0.9080159,0.00025896332,0.0000894606,0.00020620952,0.00074865593,0.002412456,0.0014834023],"genre_scores_gemma":[0.5536823,0.00020761232,0.44370303,0.000040299332,0.00002768341,0.00013020675,0.0010390511,0.00013476674,0.0010350377],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986493,0.0003273748,0.0001407804,0.00038047158,0.00040354073,0.00009847044],"domain_scores_gemma":[0.99812,0.00068150525,0.00018323303,0.00032036018,0.00063085876,0.00006398172],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010944798,0.0008472541,0.00083988556,0.0030074553,0.0008556642,0.0011886157,0.00095535413,0.0008918246,0.0006225767],"category_scores_gemma":[0.004331325,0.0003753375,0.0008586572,0.0034391857,0.00042214757,0.0005639884,0.0009907802,0.0006010311,0.0004085581],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028846765,0.00024272378,0.013361664,0.00034561823,0.0003404998,0.00041029384,0.00067980273,0.4840815,0.02239933,0.007782089,0.004783569,0.4652844],"study_design_scores_gemma":[0.00001033275,0.000040843446,0.0041664066,0.000014606659,0.000028698993,0.00011437787,0.00018418196,0.9774335,0.010241119,0.0049680807,0.0027634094,0.000034463825],"about_ca_topic_score_codex":0.012533324,"about_ca_topic_score_gemma":0.0077980226,"teacher_disagreement_score":0.012533324,"about_ca_system_score_codex":0.0010869302,"about_ca_system_score_gemma":0.0012605289,"threshold_uncertainty_score":0.024920762},"labels":[],"label_agreement":null},{"id":"W2804364458","doi":"10.1609/aaai.v32i1.12221","title":"Clustering - What Both Theoreticians and Practitioners Are Doing Wrong","year":2018,"lang":"en","type":"preprint","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Selection (genetic algorithm); Task (project management); Conceptual clustering; Correlation clustering; Machine learning; Implementation; Consensus clustering; Constrained clustering; CURE data clustering algorithm; Artificial intelligence; Data mining; Engineering","score_opus":0.0757605411064104,"score_gpt":0.3368907167482257,"score_spread":0.2611301756418153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2804364458","genre_codex":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020607195,0.051469218,0.10604784,0.8104351,0.01007587,0.000042456242,0.00029580668,0.0007326796,0.018840343],"genre_scores_gemma":[0.19983816,0.13873835,0.28540242,0.3163682,0.03495739,0.00047584335,0.001282011,0.0023444323,0.020593215],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9596301,0.021094264,0.0017661551,0.00478134,0.01160472,0.0011233862],"domain_scores_gemma":[0.86469203,0.09028243,0.0034638692,0.021631006,0.016668355,0.0032623906],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0436604,0.0013717723,0.0024761434,0.0055859243,0.007707458,0.015652617,0.0063417633,0.0153981345,0.005899038],"category_scores_gemma":[0.12677279,0.0011133262,0.0015767599,0.008810697,0.03102647,0.036055278,0.0056636883,0.012619315,0.0055359732],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055237895,0.000034407738,0.0014798352,0.0010527181,0.00013862959,0.00010118853,0.0025439193,0.0023622261,0.00020060166,0.6217492,0.2329999,0.13728218],"study_design_scores_gemma":[0.000016152459,0.000010641258,0.00033033092,0.00078131445,0.000014790629,0.00013696334,0.0018271597,0.0017018643,0.00025078465,0.8910473,0.10382327,0.000059503956],"about_ca_topic_score_codex":0.007537201,"about_ca_topic_score_gemma":0.0056511746,"teacher_disagreement_score":0.9563396,"about_ca_system_score_codex":0.005627858,"about_ca_system_score_gemma":0.0054233233,"threshold_uncertainty_score":0.23090106},"labels":[],"label_agreement":null},{"id":"W2805783523","doi":"10.1007/978-3-319-92058-0_8","title":"A Comparison of Knee Strategies for Hierarchical Spatial Clustering","year":2018,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Hierarchical clustering; Computer science; Cluster analysis; Medoid; Centroid; Dendrogram; Cluster (spacecraft); Data mining; Artificial intelligence; Complete linkage; Pattern recognition (psychology); Population","score_opus":0.04336293284831568,"score_gpt":0.3482684205343744,"score_spread":0.30490548768605874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2805783523","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031462397,0.004652618,0.94933695,0.00035753733,0.00025562788,0.00028426741,0.00033223012,0.002379099,0.0109393485],"genre_scores_gemma":[0.26853022,0.0032735374,0.7156395,0.00036442024,0.00016154765,0.0003166662,0.0015769145,0.001436971,0.008700261],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99731785,0.0009249962,0.00018987808,0.00029294685,0.0009646457,0.0003096872],"domain_scores_gemma":[0.99179417,0.0043471646,0.0003053912,0.0010909158,0.0019771212,0.0004851742],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004828253,0.0016252649,0.002472989,0.0030749692,0.0015105106,0.0038859101,0.004135366,0.0021726794,0.008729295],"category_scores_gemma":[0.015601587,0.00074919115,0.0014724188,0.0046829004,0.0011122534,0.004668908,0.0033470225,0.0015549102,0.002481801],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016397301,0.0005319232,0.0012759299,0.00080995366,0.00033846943,0.00006997692,0.0005322985,0.13476722,0.0035402686,0.071995966,0.01522378,0.7692745],"study_design_scores_gemma":[0.00030434117,0.0006770859,0.0014833687,0.00016729119,0.00024228446,0.00022739149,0.0009835409,0.9176179,0.0037892635,0.066953585,0.007457228,0.00009667918],"about_ca_topic_score_codex":0.01039831,"about_ca_topic_score_gemma":0.016109833,"teacher_disagreement_score":0.01039831,"about_ca_system_score_codex":0.0018356108,"about_ca_system_score_gemma":0.0029427777,"threshold_uncertainty_score":0.029202461},"labels":[],"label_agreement":null},{"id":"W2885148117","doi":"10.1142/s0219649218500338","title":"High-Dimensional Text Datasets Clustering Algorithm Based on Cuckoo Search and Latent Semantic Indexing","year":2018,"lang":"en","type":"article","venue":"Journal of Information & Knowledge Management","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Concordia University","keywords":"Computer science; Cluster analysis; Cuckoo search; Data mining; Search engine indexing; Document clustering; Clustering high-dimensional data; Curse of dimensionality; Spectral clustering; Artificial intelligence; Machine learning; Particle swarm optimization","score_opus":0.016741765429655938,"score_gpt":0.2882689026699501,"score_spread":0.27152713724029415,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2885148117","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04222537,0.0012186291,0.9481145,0.0005352135,0.0001310392,0.000539427,0.00073004264,0.0019100424,0.0045956797],"genre_scores_gemma":[0.4007259,0.00097999,0.58409,0.00025055392,0.0001349544,0.001335292,0.004190778,0.00029225193,0.008000265],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998511,0.00027576683,0.00013558975,0.0003796444,0.0005607132,0.00013723894],"domain_scores_gemma":[0.9990414,0.0002294008,0.00012302511,0.00010548013,0.00043484778,0.00006584978],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011387733,0.0013130008,0.0017112496,0.0038011884,0.0019592275,0.0015168958,0.002921881,0.0014567496,0.001817919],"category_scores_gemma":[0.0030201424,0.00041338525,0.001277393,0.0057695555,0.0007138736,0.002228553,0.0012092653,0.0010279876,0.00065771135],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031191303,0.00025059175,0.0039559463,0.00057266245,0.000263109,0.00018358372,0.00046321825,0.60247487,0.008982547,0.036446545,0.017614854,0.3284802],"study_design_scores_gemma":[0.00003279612,0.00003195679,0.0005220553,0.000012772242,0.000017107895,0.000056049503,0.000047138088,0.988887,0.0014709401,0.0067961114,0.0020997226,0.000026378953],"about_ca_topic_score_codex":0.026111681,"about_ca_topic_score_gemma":0.018222239,"teacher_disagreement_score":0.026111681,"about_ca_system_score_codex":0.0027272138,"about_ca_system_score_gemma":0.0029801868,"threshold_uncertainty_score":0.0519194},"labels":[],"label_agreement":null},{"id":"W2885153817","doi":"10.5740/jaoacint.18-0161","title":"Log Transformation and the Effect on Estimation, Implication, and Interpretation of Mean and Measurement Uncertainty in Microbial Enumeration","year":2018,"lang":"en","type":"article","venue":"Journal of AOAC International","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Enumeration; Transformation (genetics); Interpretation (philosophy); Estimation; Statistics; Mathematics; Biological system; Applied mathematics; Econometrics; Biology; Computer science; Combinatorics; Economics; Genetics","score_opus":0.010492868194594974,"score_gpt":0.2946327197836856,"score_spread":0.2841398515890906,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2885153817","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05065761,0.017224802,0.8981076,0.007302704,0.0065086368,0.0006351458,0.0020073266,0.0033626908,0.014193417],"genre_scores_gemma":[0.45569187,0.005336114,0.52217686,0.0049102856,0.0018817224,0.001891409,0.0012443127,0.002788435,0.004079018],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8036887,0.13997039,0.0130669875,0.011075138,0.031160973,0.0010378258],"domain_scores_gemma":[0.4462702,0.48107034,0.019617036,0.03343835,0.019020611,0.00058363704],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1027159,0.0017258009,0.0018739135,0.0034496658,0.0018686845,0.004903444,0.0027195287,0.0027024688,0.005766162],"category_scores_gemma":[0.41190547,0.0009298991,0.0026372832,0.005832488,0.0076562646,0.004278556,0.0044809636,0.0062416024,0.001510676],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027580305,0.00037517797,0.044279832,0.009552866,0.0014575899,0.0017255256,0.008718367,0.014716035,0.025905987,0.16055231,0.036229156,0.69372904],"study_design_scores_gemma":[0.00028080525,0.0026064874,0.08155416,0.007938881,0.0018802475,0.0051036025,0.0038531234,0.08035182,0.12540947,0.38657597,0.3031255,0.0013199502],"about_ca_topic_score_codex":0.0015910092,"about_ca_topic_score_gemma":0.0015524242,"teacher_disagreement_score":0.1027159,"about_ca_system_score_codex":0.0021855035,"about_ca_system_score_gemma":0.0027466174,"threshold_uncertainty_score":0.5432201},"labels":[],"label_agreement":null},{"id":"W2889601666","doi":"10.5555/3327757.3327943","title":"Clustering Redemption–Beyond the Impossibility of Kleinberg’s Axioms","year":2018,"lang":"en","type":"article","venue":"Research Portal (King's College London)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Axiom; Cluster analysis; Impossibility; Consistency (knowledge bases); Mathematics; Theoretical computer science; Computer science; Algorithm; Discrete mathematics; Mathematical economics; Artificial intelligence","score_opus":0.04832968473082041,"score_gpt":0.3718895973347923,"score_spread":0.32355991260397193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889601666","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017770356,0.00072755927,0.9624813,0.006068539,0.00023701036,0.00026264726,0.00029457567,0.0005647989,0.011593214],"genre_scores_gemma":[0.19855662,0.0004853088,0.7928641,0.0024375229,0.0003130301,0.00066169485,0.0004445028,0.0004562495,0.0037809575],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.94569844,0.020774461,0.004645216,0.013296112,0.013683955,0.0019019681],"domain_scores_gemma":[0.85102,0.07427755,0.0072654304,0.046356156,0.019131592,0.0019492871],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.052926656,0.0016192743,0.0037324755,0.002650347,0.0050678616,0.0083444305,0.009129402,0.006372561,0.0039695934],"category_scores_gemma":[0.16789372,0.00214637,0.0042728125,0.0030418106,0.018100759,0.022078212,0.010550365,0.01390049,0.002003708],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001406997,0.000044293647,0.0011606851,0.00021931955,0.000120388846,0.00008767709,0.00069682265,0.012670887,0.0010912522,0.947589,0.004074709,0.03210438],"study_design_scores_gemma":[0.00007015183,0.000040556453,0.00035762542,0.00008047144,0.000032383432,0.00014284751,0.00011049859,0.043821666,0.0019199543,0.9460087,0.0073534125,0.0000616483],"about_ca_topic_score_codex":0.0044451584,"about_ca_topic_score_gemma":0.0032131104,"teacher_disagreement_score":0.052926656,"about_ca_system_score_codex":0.005144436,"about_ca_system_score_gemma":0.0066008815,"threshold_uncertainty_score":0.27990627},"labels":[],"label_agreement":null},{"id":"W2896348621","doi":"10.1007/978-3-030-01298-4_2","title":"A Self-representation Model for Robust Clustering of Categorical Sequences","year":2018,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Categorical variable; Computer science; Pattern recognition (psychology); Artificial intelligence; Pairwise comparison; Representation (politics); CURE data clustering algorithm; Similarity (geometry); Correlation clustering; Noise (video); Data mining; Subspace topology; Machine learning; Image (mathematics)","score_opus":0.05873108898476034,"score_gpt":0.32340882607728344,"score_spread":0.26467773709252307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2896348621","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0051925937,0.00008932473,0.9937156,0.00009866059,0.000025702493,0.000024473598,0.00012060615,0.00038854807,0.0003445414],"genre_scores_gemma":[0.356146,0.00047579227,0.62792283,0.00044816957,0.00027012234,0.00046133026,0.002625607,0.0005757905,0.011074325],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978885,0.0006116253,0.00012753173,0.0006913842,0.00048088952,0.00019996602],"domain_scores_gemma":[0.9961766,0.001379515,0.00037992134,0.0010882339,0.0007946101,0.00018111269],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027064274,0.0006549809,0.0014914767,0.0019240772,0.0007518224,0.0020153576,0.004383781,0.0026087663,0.0032772273],"category_scores_gemma":[0.008507132,0.0007121093,0.0020297698,0.0025241855,0.0014057531,0.0036227042,0.0027906362,0.0023902773,0.0019518574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034872373,0.00021314075,0.0014579042,0.00019217697,0.00018572976,0.00010560287,0.00034616786,0.51428276,0.0074761496,0.19512792,0.00815564,0.27210802],"study_design_scores_gemma":[0.0000051630022,0.000023049968,0.00010898262,0.0000068416603,0.000009885081,0.000022856462,0.000011282455,0.96954376,0.0004853727,0.029172461,0.00059973204,0.00001064392],"about_ca_topic_score_codex":0.005420919,"about_ca_topic_score_gemma":0.0052226344,"teacher_disagreement_score":0.005420919,"about_ca_system_score_codex":0.0018266307,"about_ca_system_score_gemma":0.0013078457,"threshold_uncertainty_score":0.014313102},"labels":[],"label_agreement":null},{"id":"W2905920056","doi":"10.32920/ryerson.14661465","title":"Kernel k-MACE: hypercube unsupervised clustering method","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; University of Toronto","funders":"","keywords":"Cluster analysis; Kernel (algebra); Pattern recognition (psychology); Computer science; Artificial intelligence; Kernel method; Initialization; Kernel embedding of distributions; Feature vector; Mace; Variable kernel density estimation; Mathematics; Support vector machine; Combinatorics","score_opus":0.0463775866907028,"score_gpt":0.3578731729497401,"score_spread":0.3114955862590373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905920056","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019232426,0.00018549289,0.9961661,0.00009882346,0.000037025096,0.00007733586,0.000108486565,0.0006501373,0.0007532873],"genre_scores_gemma":[0.093520924,0.0002859973,0.89893126,0.00023170818,0.00008538761,0.00073937816,0.0010950877,0.0004739115,0.004636334],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99717915,0.0011645763,0.00015063732,0.00062146405,0.000750656,0.00013352768],"domain_scores_gemma":[0.99748343,0.00078934856,0.0001975614,0.00055349746,0.0008840815,0.000092114795],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022759647,0.0010449219,0.0018528363,0.001984247,0.0011459488,0.0015661438,0.003051281,0.001811291,0.003725919],"category_scores_gemma":[0.007206761,0.0006844355,0.0016064988,0.002559613,0.0009874932,0.0019345576,0.0024400835,0.0021184254,0.002543226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031307587,0.0001273818,0.0013279476,0.0003569545,0.00038125739,0.00014019056,0.00032880926,0.56922734,0.007284055,0.04242767,0.017847857,0.3602375],"study_design_scores_gemma":[0.000014446242,0.000017756749,0.00023340149,0.000012957669,0.000008972169,0.00006916317,0.000025967971,0.98060834,0.0021771537,0.012381157,0.004428669,0.000022029653],"about_ca_topic_score_codex":0.005193781,"about_ca_topic_score_gemma":0.005115139,"teacher_disagreement_score":0.005193781,"about_ca_system_score_codex":0.0011162391,"about_ca_system_score_gemma":0.0022887199,"threshold_uncertainty_score":0.012464404},"labels":[],"label_agreement":null},{"id":"W2913182958","doi":"10.1109/hpcc/smartcity/dss.2018.00044","title":"Towards a New Approach for Empowering the MR-DBSCAN Clustering for Massive Data Using Quadtree","year":2018,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"DBSCAN; Computer science; Quadtree; Cluster analysis; Scalability; Data mining; Noise (video); Node (physics); Unsupervised learning; Artificial intelligence; CURE data clustering algorithm; Correlation clustering; Database","score_opus":0.20291280341509466,"score_gpt":0.4207903255328411,"score_spread":0.21787752211774647,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2913182958","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026573357,0.00018882539,0.9942954,0.0002753279,0.000044271717,0.000049640497,0.00017287218,0.0014447503,0.0008715545],"genre_scores_gemma":[0.032937825,0.00028368115,0.9643422,0.00019036153,0.00004318278,0.00010516393,0.000672228,0.0002435632,0.0011818318],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979614,0.00032575888,0.00014334559,0.0003614569,0.0010712944,0.00013671743],"domain_scores_gemma":[0.9977793,0.0003145929,0.00012503787,0.0005829993,0.001087499,0.000110540816],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001462947,0.0008776188,0.0012539291,0.002647355,0.0014401458,0.0029523799,0.0036599091,0.0012686314,0.0023040902],"category_scores_gemma":[0.0048312345,0.0007391826,0.0013376182,0.0050511817,0.0009912467,0.0034592294,0.0029058133,0.0018187705,0.0018117118],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036058653,0.00027312708,0.0037659144,0.00041456736,0.00035913484,0.0004246949,0.0009290984,0.26644966,0.021055883,0.10972405,0.037329227,0.558914],"study_design_scores_gemma":[0.000026037686,0.000050906245,0.0005288961,0.000025638361,0.000027122618,0.00026241265,0.00025488177,0.9054516,0.0074436413,0.056184005,0.0296948,0.0000500449],"about_ca_topic_score_codex":0.015534647,"about_ca_topic_score_gemma":0.0192226,"teacher_disagreement_score":0.015534647,"about_ca_system_score_codex":0.0012529494,"about_ca_system_score_gemma":0.0025053548,"threshold_uncertainty_score":0.030888438},"labels":[],"label_agreement":null},{"id":"W2913529671","doi":"10.1109/ssci.2018.8628863","title":"Hybrid Metaheuristic Algorithm for Clustering","year":2018,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Metaheuristic; Cluster analysis; Tabu search; Computer science; Parallel metaheuristic; Heuristics; Artificial intelligence; Data mining; Machine learning; Meta-optimization","score_opus":0.03140047396787493,"score_gpt":0.3250039390918261,"score_spread":0.2936034651239512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2913529671","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040000044,0.0012370466,0.9882571,0.00026011092,0.00011862951,0.00009007866,0.00008066133,0.00044362084,0.0055127908],"genre_scores_gemma":[0.14985652,0.0017369948,0.8381295,0.00045734155,0.00013207096,0.000610084,0.00047084817,0.00022559764,0.008380975],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999164,0.00029988596,0.000051006016,0.00014050426,0.00026033135,0.00008424272],"domain_scores_gemma":[0.99939024,0.00029862297,0.00006027527,0.00007062165,0.00014875353,0.000031418364],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012465061,0.0014520253,0.0016310356,0.0020470757,0.0010816224,0.0017608451,0.0025375045,0.0025762916,0.0030782276],"category_scores_gemma":[0.0021998044,0.0005973453,0.0019440361,0.0027966485,0.0008624335,0.0014567628,0.0011540123,0.0015953901,0.0011151306],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000075898075,0.00007135195,0.00037935394,0.0002187471,0.00021148326,0.00007718046,0.00008528429,0.85441595,0.0025087579,0.043467935,0.0043528704,0.09413528],"study_design_scores_gemma":[0.000021954405,0.00002313118,0.000057815683,0.000018377597,0.000017781882,0.00003454262,0.000014713425,0.98193365,0.00059761526,0.011882915,0.005385937,0.000011628559],"about_ca_topic_score_codex":0.004794848,"about_ca_topic_score_gemma":0.0044684755,"teacher_disagreement_score":0.004794848,"about_ca_system_score_codex":0.0017924425,"about_ca_system_score_gemma":0.0017545501,"threshold_uncertainty_score":0.013005137},"labels":[],"label_agreement":null},{"id":"W2914448888","doi":"","title":"Proceedings of the 8th Conference on 8th WSEAS International Conference on Neural Networks - Volume 8","year":2007,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Volume (thermodynamics); Artificial neural network; Library science; Political science; Computer science; Artificial intelligence","score_opus":0.046595973343882394,"score_gpt":0.31942001435816136,"score_spread":0.272824041014279,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914448888","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10469909,0.04665603,0.49181637,0.019097406,0.10463739,0.0008256482,0.0041229883,0.009827985,0.21831717],"genre_scores_gemma":[0.16141702,0.017860398,0.15271443,0.0019080006,0.005181569,0.00044320623,0.011098548,0.0023035994,0.64707327],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99909747,0.00020323588,0.000074131065,0.00011771771,0.0003713633,0.00013621533],"domain_scores_gemma":[0.99844545,0.00013876853,0.000039347906,0.00017653253,0.0010342348,0.0001657233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021803738,0.0010164654,0.0016065799,0.0013990785,0.0009243151,0.0032608714,0.0014156745,0.0013222535,0.060728222],"category_scores_gemma":[0.002536906,0.00027220705,0.00085286127,0.0012461162,0.0006156769,0.001805798,0.001408435,0.0014262206,0.019589866],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006554318,0.0004674982,0.002674127,0.0004274044,0.00023606798,0.00017850555,0.00017820994,0.007393394,0.008693758,0.007014918,0.28916815,0.6829125],"study_design_scores_gemma":[0.000105771374,0.00041745947,0.00968686,0.00058866787,0.0004026008,0.00059528806,0.00068035565,0.13671188,0.027216746,0.015553158,0.807929,0.00011223037],"about_ca_topic_score_codex":0.010171391,"about_ca_topic_score_gemma":0.020807803,"teacher_disagreement_score":0.060728222,"about_ca_system_score_codex":0.00095175544,"about_ca_system_score_gemma":0.0018286181,"threshold_uncertainty_score":0.20315611},"labels":[],"label_agreement":null},{"id":"W2915001123","doi":"10.1609/aaai.v26i1.8282","title":"Weighted Clustering","year":2021,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Strategic Research Council; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; National Research Foundation; Sino-Danish Center; Danmarks Grundforskningsfond; National Science Foundation","keywords":"Cluster analysis; Computer science; Correlation clustering; CURE data clustering algorithm; Generalization; Single-linkage clustering; Data mining; Canopy clustering algorithm; Artificial intelligence; Fuzzy clustering; Consensus clustering; Hierarchical clustering; Selection (genetic algorithm); Pattern recognition (psychology); Mathematics","score_opus":0.08640250646346599,"score_gpt":0.32819258052314715,"score_spread":0.24179007405968117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915001123","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006224465,0.00051622046,0.9872662,0.0002992848,0.00011894264,0.00014922164,0.00032562274,0.00024826918,0.0048517766],"genre_scores_gemma":[0.20498632,0.0013573525,0.7757124,0.000664753,0.00041540584,0.00060553645,0.0027024571,0.0004965924,0.013059147],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99022734,0.0031150442,0.0004975833,0.0026566684,0.0030273814,0.0004760414],"domain_scores_gemma":[0.98735726,0.004378977,0.001104795,0.0041349195,0.0026356492,0.0003883849],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0063957223,0.0014604193,0.0014150544,0.0038929603,0.0015368639,0.0039715394,0.004711466,0.0019436327,0.0056130616],"category_scores_gemma":[0.024892882,0.00070178445,0.0015643418,0.005369802,0.0019649877,0.0058946167,0.0041006575,0.0019462891,0.00259355],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016699122,0.00014768873,0.0027045652,0.00060953514,0.00042089744,0.0001656261,0.00040310164,0.16794086,0.004662048,0.5719521,0.016879112,0.23394758],"study_design_scores_gemma":[0.000031811167,0.00009386971,0.0009229043,0.000082447,0.00008840288,0.0002602985,0.00016773367,0.49116886,0.0041030236,0.46747822,0.035551284,0.000051226933],"about_ca_topic_score_codex":0.0024637347,"about_ca_topic_score_gemma":0.0034539765,"teacher_disagreement_score":0.0063957223,"about_ca_system_score_codex":0.0020057096,"about_ca_system_score_gemma":0.0020188054,"threshold_uncertainty_score":0.033824265},"labels":[],"label_agreement":null},{"id":"W2918582631","doi":"10.1109/acssc.2018.8645076","title":"Kernel K-Mace Clustering","year":2018,"lang":"en","type":"article","venue":"2018 52nd Asilomar Conference on Signals, Systems, and Computers","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Kernel (algebra); Cluster analysis; Variable kernel density estimation; Initialization; Mathematics; Kernel method; Computer science; Gaussian function; Kernel embedding of distributions; Algorithm; Pattern recognition (psychology); Artificial intelligence; Gaussian; Discrete mathematics","score_opus":0.05666554003204609,"score_gpt":0.3049090948370821,"score_spread":0.24824355480503602,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2918582631","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028743905,0.00020214848,0.9954052,0.000068589434,0.000032762047,0.000051508032,0.00008022199,0.0005483658,0.00073678873],"genre_scores_gemma":[0.12192122,0.00038435083,0.8708016,0.00018764145,0.00007753966,0.00034036435,0.0012242622,0.00045023338,0.004612714],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996926,0.000720453,0.00021500539,0.0007870081,0.0011258142,0.00022569449],"domain_scores_gemma":[0.995933,0.00092876,0.00033665376,0.00080642063,0.001880696,0.00011446293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021276341,0.0012715723,0.0021235733,0.0032752715,0.0012707852,0.0023072613,0.0038235353,0.0021556087,0.0023264284],"category_scores_gemma":[0.009176969,0.0007859403,0.0024166105,0.0028588644,0.0009630307,0.0026700662,0.0023989722,0.0018862357,0.0022896826],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029236535,0.00021402864,0.0031174012,0.00046656444,0.000495221,0.00016124143,0.00049622066,0.5179336,0.012022764,0.04718503,0.014546231,0.40306935],"study_design_scores_gemma":[0.000012254307,0.000027046786,0.0006077466,0.0000162801,0.000019695844,0.00008531978,0.00003710396,0.98132175,0.0030646098,0.009978327,0.00479491,0.00003496758],"about_ca_topic_score_codex":0.0067436546,"about_ca_topic_score_gemma":0.006548903,"teacher_disagreement_score":0.0067436546,"about_ca_system_score_codex":0.0011332721,"about_ca_system_score_gemma":0.002244687,"threshold_uncertainty_score":0.01340884},"labels":[],"label_agreement":null},{"id":"W2923023146","doi":"10.1109/tbdata.2019.2907624","title":"Hierarchical Density-Based Clustering Using MapReduce","year":2019,"lang":"en","type":"article","venue":"IEEE Transactions on Big Data","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Fundação de Amparo à Pesquisa do Estado de Minas Gerais; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Natural Sciences and Engineering Research Council of Canada; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Computer science; Cluster analysis; Scalability; Automatic summarization; Correlation clustering; Hierarchical clustering; CURE data clustering algorithm; Data mining; Canopy clustering algorithm; Constrained clustering; Artificial intelligence; Database","score_opus":0.1308743166587772,"score_gpt":0.3353534078399405,"score_spread":0.2044790911811633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2923023146","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0083112195,0.00022392608,0.97952205,0.00018376303,0.00005726308,0.00018962704,0.0006036907,0.008026351,0.0028821218],"genre_scores_gemma":[0.22214764,0.00043100645,0.76855195,0.0002062581,0.000071874565,0.00052930485,0.0036675017,0.0008128196,0.0035816047],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986765,0.00020312484,0.00006268795,0.00025830374,0.000673063,0.00012616446],"domain_scores_gemma":[0.9989784,0.00020314344,0.000055294597,0.00030946534,0.00037179954,0.00008202947],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008565807,0.0009837473,0.0011493777,0.0013896544,0.0012649423,0.0012851341,0.002279169,0.00062480516,0.0023056823],"category_scores_gemma":[0.002722541,0.0005978104,0.0013282425,0.0024006115,0.000498505,0.0016837929,0.0019854116,0.0011610343,0.0014940738],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035229264,0.00028244813,0.0023695792,0.00053776853,0.00028968882,0.00029899,0.0006579127,0.5395772,0.02003823,0.037400436,0.04763719,0.3505583],"study_design_scores_gemma":[0.00004875009,0.000035702265,0.0007136447,0.000011392667,0.000021921041,0.00011559032,0.00014544571,0.94820964,0.0081555415,0.028044434,0.014454353,0.000043592783],"about_ca_topic_score_codex":0.016656213,"about_ca_topic_score_gemma":0.01581814,"teacher_disagreement_score":0.016656213,"about_ca_system_score_codex":0.0012099238,"about_ca_system_score_gemma":0.0025483766,"threshold_uncertainty_score":0.033118546},"labels":[],"label_agreement":null},{"id":"W2940105449","doi":"10.1002/9780470057339.vac028.pub2","title":"Clustering","year":2012,"lang":"en","type":"other","venue":"Encyclopedia of Environmetrics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Principal component analysis; Cluster analysis; Ordination; Pattern recognition (psychology); Cluster (spacecraft); Artificial intelligence; Basis (linear algebra); Projection (relational algebra); Computer science; Group (periodic table); Subdivision; Object (grammar); Data mining; Projection pursuit; Mathematics; Geography; Machine learning; Algorithm","score_opus":0.014610352294990935,"score_gpt":0.2573610325764483,"score_spread":0.24275068028145738,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2940105449","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005085174,0.0024349499,0.8866833,0.0011147637,0.0010208386,0.0012153788,0.013497591,0.010307774,0.07864029],"genre_scores_gemma":[0.086462,0.0038698146,0.7621284,0.00072339963,0.0005422841,0.0013135101,0.061488967,0.003521895,0.07994975],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9943136,0.0008159557,0.00041421843,0.001669044,0.0023590364,0.00042814185],"domain_scores_gemma":[0.9962572,0.00044845656,0.00017032484,0.0010953336,0.0018898228,0.00013889198],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021769777,0.002424333,0.0019732192,0.0067932517,0.0025279531,0.0067795883,0.0038382933,0.0018336402,0.048046596],"category_scores_gemma":[0.008296391,0.0007035422,0.0023072832,0.00992356,0.0009886322,0.0033716273,0.003980122,0.0015865004,0.04606004],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024758806,0.000103589904,0.0019475312,0.0015402113,0.00029816042,0.00022979554,0.00070385734,0.038312458,0.0052883527,0.08548759,0.23800047,0.62784046],"study_design_scores_gemma":[0.00006814014,0.00010516147,0.0026683467,0.0006533629,0.00013142363,0.0005977471,0.0009334149,0.15306206,0.011595879,0.14321822,0.68680745,0.00015868554],"about_ca_topic_score_codex":0.005005518,"about_ca_topic_score_gemma":0.0046965578,"teacher_disagreement_score":0.048046596,"about_ca_system_score_codex":0.0019107524,"about_ca_system_score_gemma":0.0032256856,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2940228547","doi":"","title":"Formulation relaxée de la séparation équilibrée d’un graphe","year":2017,"lang":"fr","type":"preprint","venue":"HAL (Le Centre pour la Communication Scientifique Directe)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Humanities; Combinatorics; Mathematics; Computer science; Philosophy","score_opus":0.026009490997036817,"score_gpt":0.29127812826218347,"score_spread":0.26526863726514666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2940228547","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013205752,0.00083814125,0.96436733,0.0027861756,0.00042410937,0.0001011925,0.00057053997,0.00013497642,0.017571697],"genre_scores_gemma":[0.35768038,0.0017539468,0.54275733,0.0019413552,0.0014927648,0.0010163614,0.0024892145,0.0012218889,0.089646794],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99835205,0.0006692835,0.000047063553,0.00046538963,0.00030456387,0.00016178336],"domain_scores_gemma":[0.99610484,0.0029347,0.00017306939,0.00016290213,0.00041626132,0.00020831524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020928981,0.0022158616,0.0018760123,0.001377305,0.0009628421,0.0033447077,0.002799554,0.0050067673,0.012228308],"category_scores_gemma":[0.008946846,0.0013026325,0.0016349504,0.0015533512,0.0021495312,0.003319322,0.0034700737,0.0048256894,0.0016264614],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018032473,0.00009425566,0.0004194685,0.0004469921,0.00009125828,0.0003813803,0.00015018336,0.8156021,0.0021671918,0.15187566,0.011032875,0.017558306],"study_design_scores_gemma":[0.000029555818,0.00002647684,0.000104470266,0.000027614891,0.0000125602155,0.000052115287,0.000041653322,0.9666659,0.0003677946,0.02818053,0.004480198,0.000011105779],"about_ca_topic_score_codex":0.016809981,"about_ca_topic_score_gemma":0.010691417,"teacher_disagreement_score":0.016809981,"about_ca_system_score_codex":0.0026944883,"about_ca_system_score_gemma":0.0016239401,"threshold_uncertainty_score":0.0409078},"labels":[],"label_agreement":null},{"id":"W2949559209","doi":"","title":"Efficient unimodality test in clustering by signature testing","year":2014,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Unimodality; Signature (topology); Cluster analysis; Statistic; Hierarchical clustering; Statistical hypothesis testing; Test statistic; Mathematics; Computer science; Algorithm; Pattern recognition (psychology); Statistics; Data mining; Artificial intelligence","score_opus":0.06277861707316333,"score_gpt":0.22055729697849485,"score_spread":0.1577786799053315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949559209","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04427677,0.00024970382,0.9528097,0.00014446815,0.000053537144,0.0000926054,0.00019325745,0.0011285285,0.0010514886],"genre_scores_gemma":[0.60124195,0.00022564779,0.3951637,0.00022727597,0.00016289359,0.00031809876,0.0011368174,0.00032348718,0.0011999913],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99122834,0.0035317168,0.00052084826,0.0016645015,0.0026340964,0.00042048446],"domain_scores_gemma":[0.97258174,0.018414814,0.0018649495,0.0028378253,0.0034309474,0.0008697964],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005757039,0.0009503975,0.0018583968,0.004408671,0.0011173004,0.0018249281,0.0026344527,0.0016042847,0.0029666177],"category_scores_gemma":[0.037327126,0.00044772367,0.000977492,0.0032765965,0.0027037684,0.0030069978,0.0028810222,0.0021108384,0.0014510521],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018457281,0.00045240787,0.022496508,0.00041173154,0.00035374277,0.0005471114,0.0004012475,0.17781532,0.036274005,0.05527888,0.0045362464,0.69958705],"study_design_scores_gemma":[0.000073234136,0.00034115976,0.0039815702,0.000023776824,0.00004998108,0.0005102209,0.00011885565,0.91898793,0.022537224,0.051255886,0.0020122498,0.000107888736],"about_ca_topic_score_codex":0.0012474474,"about_ca_topic_score_gemma":0.00077842973,"teacher_disagreement_score":0.005757039,"about_ca_system_score_codex":0.0011141059,"about_ca_system_score_gemma":0.0022449978,"threshold_uncertainty_score":0.03044647},"labels":[],"label_agreement":null},{"id":"W2950793146","doi":"10.2298/csis180601007l","title":"Density-based clustering with constraints","year":2019,"lang":"en","type":"article","venue":"Computer Science and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"DBSCAN; Cluster analysis; Computer science; Benchmark (surveying); Data mining; Process (computing); Computation; Correlation clustering; Quality (philosophy); Constrained clustering; Algorithm; CURE data clustering algorithm; Machine learning","score_opus":0.01108352661053865,"score_gpt":0.23954492617872325,"score_spread":0.2284613995681846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950793146","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034204437,0.00032192335,0.9939528,0.00010487024,0.000028222805,0.00010565324,0.00020255004,0.00052130525,0.0013422603],"genre_scores_gemma":[0.09173143,0.0005484438,0.90355235,0.00017256218,0.00007186201,0.00031774188,0.0015143092,0.0002983863,0.0017928499],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949393,0.00112419,0.0003161964,0.00091220415,0.002473637,0.0002343828],"domain_scores_gemma":[0.9942403,0.002514816,0.00040050835,0.0010144402,0.0016766301,0.00015339509],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030186037,0.0014594232,0.0022904235,0.0035629354,0.0012829925,0.003077802,0.00430308,0.001749145,0.003127632],"category_scores_gemma":[0.01781444,0.0009369347,0.0014317626,0.005716801,0.0012556739,0.0031890175,0.0031656811,0.002226444,0.0015064542],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019570152,0.00008578829,0.0017291792,0.00043751302,0.00017564885,0.00009612517,0.00023450454,0.65826637,0.0022455829,0.06567695,0.0075505553,0.26330614],"study_design_scores_gemma":[0.000017846049,0.000022333532,0.0002842175,0.000030603387,0.000017599123,0.000090527086,0.00004993934,0.9622611,0.0019343116,0.03014092,0.0051210723,0.000029599934],"about_ca_topic_score_codex":0.017352264,"about_ca_topic_score_gemma":0.011741258,"teacher_disagreement_score":0.017352264,"about_ca_system_score_codex":0.0025448317,"about_ca_system_score_gemma":0.0028691553,"threshold_uncertainty_score":0.034502506},"labels":[],"label_agreement":null},{"id":"W2950811391","doi":"10.3390/info10060208","title":"Latent Feature Group Learning for High-Dimensional Data Clustering","year":2019,"lang":"en","type":"article","venue":"Information","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Shenzhen University; China Postdoctoral Science Foundation","keywords":"Feature (linguistics); Pattern recognition (psychology); Cluster analysis; Artificial intelligence; Weighting; Computer science; Fitness function; Constraint (computer-aided design); Clustering high-dimensional data; Crossover; Data mining; Genetic algorithm; Mathematics; Machine learning","score_opus":0.02428100589534357,"score_gpt":0.2843044617072172,"score_spread":0.2600234558118736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950811391","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043083206,0.000220885,0.99467975,0.00009467053,0.00002119771,0.00003127374,0.00004034136,0.00035784225,0.00024568598],"genre_scores_gemma":[0.2801504,0.00037843964,0.71586883,0.00024044936,0.00013904143,0.0003530794,0.000885537,0.00020771439,0.0017764813],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974474,0.00076886045,0.0001477808,0.00072612934,0.0007056894,0.00020413657],"domain_scores_gemma":[0.9977811,0.0008269248,0.00029520984,0.00043896397,0.00054953265,0.00010828351],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024284709,0.0010895665,0.0018738508,0.002763347,0.0015274079,0.0012113947,0.0023837867,0.0016776187,0.0014654786],"category_scores_gemma":[0.0058657415,0.00044747582,0.0015956023,0.00323175,0.0013208954,0.0025855114,0.001981604,0.0018317706,0.00092321273],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018978877,0.00026285485,0.004668868,0.00030252387,0.00033394797,0.00015394545,0.0005106179,0.40147427,0.0067445645,0.02904566,0.006537904,0.549775],"study_design_scores_gemma":[0.00001479336,0.00004064674,0.00034158223,0.000011607189,0.000018179579,0.000043827098,0.00004923104,0.9790688,0.0011613042,0.017889038,0.0013413158,0.000019653193],"about_ca_topic_score_codex":0.0042124866,"about_ca_topic_score_gemma":0.0041458854,"teacher_disagreement_score":0.0042124866,"about_ca_system_score_codex":0.0012148242,"about_ca_system_score_gemma":0.0016183539,"threshold_uncertainty_score":0.012843132},"labels":[],"label_agreement":null},{"id":"W2951747536","doi":"10.1145/3321386","title":"Hierarchical Clustering","year":2019,"lang":"en","type":"article","venue":"Journal of the ACM","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":207,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Engineering and Physical Sciences Research Council; Alan Turing Institute; National Science Foundation","keywords":"Cluster analysis; Hierarchical clustering; Granularity; Constrained clustering; Computer science; Correlation clustering; Data mining; Similarity (geometry); Hierarchy; Fuzzy clustering; Set (abstract data type); Function (biology); Canopy clustering algorithm; Mathematics; Artificial intelligence","score_opus":0.020996761164785226,"score_gpt":0.29897148618925345,"score_spread":0.27797472502446824,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951747536","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026327735,0.002177528,0.9594924,0.00074731634,0.0002957739,0.000735456,0.008067493,0.006840693,0.019010475],"genre_scores_gemma":[0.06916197,0.0024848047,0.8647962,0.0011005558,0.0003414628,0.0010355965,0.03783154,0.00233954,0.020908352],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99246275,0.0016243635,0.00049319037,0.0024617729,0.0024201444,0.0005377101],"domain_scores_gemma":[0.99383014,0.0012165441,0.0003931941,0.0023481313,0.0019751978,0.00023679837],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003429312,0.0026124462,0.0028753239,0.006708002,0.002982405,0.006115853,0.0060565146,0.0026949304,0.027587803],"category_scores_gemma":[0.012733254,0.0010737558,0.0039728587,0.008202929,0.0014567343,0.0050952816,0.0055894163,0.0023896464,0.023623632],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029923566,0.00016739544,0.0042653447,0.0017765999,0.0008212059,0.00028669613,0.0007755331,0.07483504,0.004662727,0.13407712,0.23175183,0.54628134],"study_design_scores_gemma":[0.00008430688,0.00011906388,0.0028087837,0.0005067384,0.0002821847,0.00058651704,0.00073202566,0.31195626,0.00611871,0.2863572,0.3902747,0.00017354282],"about_ca_topic_score_codex":0.009738219,"about_ca_topic_score_gemma":0.016975138,"teacher_disagreement_score":0.027587803,"about_ca_system_score_codex":0.0033436823,"about_ca_system_score_gemma":0.0044331695,"threshold_uncertainty_score":0.0922904},"labels":[],"label_agreement":null},{"id":"W2952461389","doi":"10.1016/j.ins.2018.01.013","title":"RECOME: A new density-based clustering algorithm using relative KNN kernel density","year":2018,"lang":"en","type":"article","venue":"Information Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":74,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Beijing Municipal Natural Science Foundation; National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Pattern recognition (psychology); Jump; Single-linkage clustering; Merge (version control); Nearest-neighbor chain algorithm; Kernel density estimation; Algorithm; Graph; Kernel (algebra); Discontinuity (linguistics); k-nearest neighbors algorithm; Data mining; Artificial intelligence; Mathematics; Correlation clustering; CURE data clustering algorithm; Canopy clustering algorithm; Theoretical computer science; Combinatorics; Statistics; Physics","score_opus":0.06083948764184269,"score_gpt":0.3449632069298229,"score_spread":0.2841237192879802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952461389","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026248943,0.00023325138,0.9949233,0.00007512803,0.00007461251,0.0000724016,0.00007931401,0.0014078817,0.0005092454],"genre_scores_gemma":[0.042352423,0.00022061824,0.953111,0.00012482954,0.00005980209,0.00020355186,0.0005270432,0.00040075163,0.0029999332],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99753654,0.0004896289,0.00012324862,0.0004897926,0.0012186858,0.00014213483],"domain_scores_gemma":[0.9977344,0.00049965153,0.0001279131,0.00048858946,0.0010416678,0.000107823915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021252332,0.0013164718,0.0025846667,0.003130736,0.001805811,0.0020625433,0.00508545,0.0022704762,0.0039122603],"category_scores_gemma":[0.007451648,0.001144015,0.0016328782,0.0031590417,0.0010115028,0.0035229127,0.003751785,0.0024539814,0.0034974257],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035513658,0.00024312257,0.0014653045,0.0002781906,0.0002404281,0.00010860088,0.00024282178,0.112141654,0.010172768,0.0201129,0.013931118,0.84070784],"study_design_scores_gemma":[0.00003546243,0.00004371945,0.0005465161,0.00001748637,0.00002727034,0.00017812912,0.000051726773,0.97893846,0.00531612,0.006792321,0.007997538,0.00005526114],"about_ca_topic_score_codex":0.00878654,"about_ca_topic_score_gemma":0.010807758,"teacher_disagreement_score":0.00878654,"about_ca_system_score_codex":0.0013638604,"about_ca_system_score_gemma":0.002091153,"threshold_uncertainty_score":0.017470777},"labels":[],"label_agreement":null},{"id":"W2954483433","doi":"","title":"Density-based clustering based on hierarchical density estimates","year":2013,"lang":"en","type":"article","venue":"ResearchOnline at James Cook University (James Cook University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Partition (number theory); Hierarchical clustering; Single-linkage clustering; Hierarchy; Complete-linkage clustering; Measure (data warehouse); Hierarchical clustering of networks; Correlation clustering; Data mining; Cluster (spacecraft); Mathematics; CURE data clustering algorithm; Computer science; Stability (learning theory); Tree (set theory); Algorithm; Artificial intelligence; Machine learning; Combinatorics","score_opus":0.017394209216698518,"score_gpt":0.24577310594874077,"score_spread":0.22837889673204226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2954483433","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003662375,0.00018364725,0.9948184,0.000075705015,0.000029575463,0.0000764099,0.000081398444,0.00039607607,0.0006764502],"genre_scores_gemma":[0.212075,0.0005870975,0.78034896,0.00015616482,0.00013558108,0.00039455725,0.0014420005,0.00047336597,0.0043872544],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99580777,0.001792882,0.00023024276,0.0007887898,0.0011431006,0.00023715169],"domain_scores_gemma":[0.9880503,0.006420529,0.00068390026,0.0018931343,0.0026747473,0.00027744926],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044848127,0.0011351369,0.002258218,0.0046702735,0.002145933,0.0026339141,0.00419477,0.0025102466,0.0035393508],"category_scores_gemma":[0.028309947,0.0016051679,0.002630365,0.0043511135,0.0015407578,0.0030779878,0.0034290452,0.0027580026,0.0027380812],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030367097,0.00016330915,0.0033253469,0.00037938316,0.00036336287,0.00014985989,0.0007494217,0.65545356,0.005268907,0.0944816,0.00873995,0.23062164],"study_design_scores_gemma":[0.000013064502,0.000015680032,0.000564299,0.000025032785,0.00002356055,0.00005531486,0.000035743706,0.9731041,0.0007660554,0.02426723,0.0010991871,0.000030631258],"about_ca_topic_score_codex":0.017388655,"about_ca_topic_score_gemma":0.019292621,"teacher_disagreement_score":0.017388655,"about_ca_system_score_codex":0.0022804432,"about_ca_system_score_gemma":0.0023056485,"threshold_uncertainty_score":0.034574866},"labels":[],"label_agreement":null},{"id":"W2964040819","doi":"10.1609/aaai.v32i1.12221","title":"Clustering - What Both Theoreticians and Practitioners Are Doing Wrong","year":2018,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Conceptual clustering; Machine learning; Selection (genetic algorithm); Correlation clustering; Artificial intelligence; Task (project management); Implementation; Consensus clustering; Constrained clustering; CURE data clustering algorithm; Data mining; Engineering; Software engineering","score_opus":0.016906449382870597,"score_gpt":0.3005511694067407,"score_spread":0.2836447200238701,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2964040819","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025027692,0.064767584,0.16845584,0.7300433,0.011933441,0.000054746215,0.00031353522,0.00088754017,0.021041261],"genre_scores_gemma":[0.18429288,0.15339933,0.36536774,0.23922314,0.033635944,0.00046871364,0.0011802657,0.0020225686,0.020409517],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9676356,0.016005171,0.0014295685,0.004115466,0.009821975,0.0009922732],"domain_scores_gemma":[0.8901276,0.07343911,0.0029150534,0.01722171,0.0136142615,0.00268221],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.03636779,0.0014253462,0.0024960877,0.005398404,0.0071510277,0.014526874,0.006290438,0.014400475,0.005349004],"category_scores_gemma":[0.102613926,0.001078902,0.0016733347,0.008456095,0.027338957,0.031449623,0.0053658253,0.011416504,0.0051644496],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053305826,0.000033607124,0.0013474051,0.0011096367,0.000143315,0.00009635764,0.002051652,0.0027251998,0.00024429263,0.64938164,0.19293624,0.14987741],"study_design_scores_gemma":[0.000014451045,0.000012168163,0.00032508295,0.0007683446,0.000016034764,0.00014662281,0.001598592,0.0020982167,0.0002832613,0.8962738,0.098400585,0.000062849445],"about_ca_topic_score_codex":0.0066716704,"about_ca_topic_score_gemma":0.0052807676,"teacher_disagreement_score":0.9636322,"about_ca_system_score_codex":0.0049022874,"about_ca_system_score_gemma":0.00468672,"threshold_uncertainty_score":0.19233358},"labels":[],"label_agreement":null},{"id":"W2969639627","doi":"10.1007/s10618-019-00651-1","title":"A unified view of density-based methods forsemi-supervised clustering and classification","year":2019,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; Natural Sciences and Engineering Research Council of Canada; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Cluster analysis; Computer science; Artificial intelligence; Semi-supervised learning; Supervised learning; Graph; Machine learning; Correlation clustering; Data stream clustering; Data mining; Pattern recognition (psychology); CURE data clustering algorithm; Artificial neural network; Theoretical computer science","score_opus":0.09721739508829468,"score_gpt":0.386530902237729,"score_spread":0.2893135071494343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2969639627","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00040294402,0.00034780693,0.9983767,0.00021542404,0.00002926632,0.000024739753,0.000022183047,0.00007569274,0.00050524477],"genre_scores_gemma":[0.0800152,0.0022903266,0.913923,0.0004141931,0.0006461876,0.0004792611,0.00034593735,0.00016759719,0.0017182849],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9917488,0.003681258,0.00041906704,0.0013056166,0.0026390848,0.0002061383],"domain_scores_gemma":[0.988817,0.005699157,0.0006955759,0.0019280101,0.0025378114,0.00032249116],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008548399,0.0016479761,0.002592206,0.005399259,0.0017634814,0.0049750106,0.004579043,0.0027045128,0.0016686392],"category_scores_gemma":[0.018012038,0.0011979692,0.0021555761,0.0044396985,0.004903835,0.0067162453,0.0047688065,0.005066001,0.001331289],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006104752,0.00010082028,0.0013056955,0.00053831417,0.00023414181,0.00014088054,0.0005996998,0.14472307,0.0028968533,0.663209,0.0057146586,0.18047583],"study_design_scores_gemma":[0.000013538476,0.000054745247,0.00044203197,0.00008716229,0.00004147612,0.00020253203,0.00011335389,0.64480215,0.0017313062,0.34208295,0.010363308,0.000065412096],"about_ca_topic_score_codex":0.0027986583,"about_ca_topic_score_gemma":0.0025412743,"teacher_disagreement_score":0.008548399,"about_ca_system_score_codex":0.002352287,"about_ca_system_score_gemma":0.0022640678,"threshold_uncertainty_score":0.045208752},"labels":[],"label_agreement":null},{"id":"W2972986690","doi":"10.1016/j.ijar.2019.09.001","title":"A three-way cluster ensemble approach for large-scale data","year":2019,"lang":"en","type":"article","venue":"International Journal of Approximate Reasoning","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":70,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Data mining; Correlation clustering; SPARK (programming language); CURE data clustering algorithm; Cluster (spacecraft); k-medians clustering; Granularity; Fuzzy clustering; Single-linkage clustering; Data stream clustering; Scale (ratio); Constrained clustering; Artificial intelligence","score_opus":0.03398508273410222,"score_gpt":0.3263792715860228,"score_spread":0.2923941888519206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2972986690","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0057312963,0.00019972585,0.99257225,0.000109600645,0.00008060716,0.00005861086,0.00013852654,0.00066536973,0.000443955],"genre_scores_gemma":[0.1429332,0.0003114429,0.85197943,0.0001859902,0.00019025616,0.00022577711,0.001421441,0.00030399862,0.0024484338],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99737763,0.0006553183,0.00017137064,0.00051959854,0.0010260277,0.00025011064],"domain_scores_gemma":[0.9955456,0.0011457647,0.00017823673,0.0009914915,0.0018814801,0.0002572939],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032917466,0.0010913793,0.0025168674,0.00252998,0.0024232825,0.0023009595,0.0037134008,0.0018422717,0.0025089537],"category_scores_gemma":[0.00769264,0.00063834,0.0026549061,0.004153103,0.0005987986,0.0027507374,0.0030057786,0.0024812627,0.001245956],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041401613,0.00033003034,0.005409164,0.00017869931,0.00097261654,0.00018657236,0.00054435205,0.31262147,0.0059536844,0.013854352,0.012616407,0.6469186],"study_design_scores_gemma":[0.000008613836,0.000027633207,0.00045976797,0.000009950687,0.000057228077,0.000039856117,0.00008548503,0.98955417,0.0008846401,0.0073996256,0.0014548831,0.000018145283],"about_ca_topic_score_codex":0.015966928,"about_ca_topic_score_gemma":0.028241936,"teacher_disagreement_score":0.015966928,"about_ca_system_score_codex":0.0008389937,"about_ca_system_score_gemma":0.0019221305,"threshold_uncertainty_score":0.031747997},"labels":[],"label_agreement":null},{"id":"W2979735657","doi":"10.1109/fuzz-ieee.2019.8858972","title":"Evidential clustering for categorical data","year":2019,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Categorical variable; Cluster analysis; Outlier; Data mining; Computer science; Partition (number theory); Fuzzy clustering; Artificial intelligence; Pattern recognition (psychology); Fuzzy logic; Mathematics; Machine learning","score_opus":0.17710035093563814,"score_gpt":0.4150464112990636,"score_spread":0.23794606036342547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979735657","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003678419,0.00030575594,0.99504143,0.00011870078,0.000017785678,0.000021109738,0.000071468916,0.00007453353,0.00067070493],"genre_scores_gemma":[0.21385756,0.00063037465,0.7833758,0.00015777111,0.00012576918,0.00013030654,0.00056616456,0.00005890412,0.0010973257],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99570197,0.0018488058,0.00029536153,0.0008565671,0.0011630098,0.00013437534],"domain_scores_gemma":[0.9924545,0.00432037,0.00074166904,0.0013367956,0.00097156793,0.00017500484],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00434071,0.0006236967,0.0009589868,0.0024692123,0.00096203055,0.0023683985,0.0015644998,0.0013038683,0.0015775822],"category_scores_gemma":[0.015151776,0.0003985192,0.0011937267,0.0023806046,0.0018823382,0.002628477,0.0021048698,0.0025305937,0.00058223784],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000106039006,0.00005610402,0.001470562,0.00044019966,0.00016716773,0.00015836346,0.00073845417,0.16920292,0.004678462,0.6010789,0.0027023614,0.21920048],"study_design_scores_gemma":[0.000011274246,0.000041414867,0.000621051,0.000060257757,0.000023208582,0.00014281583,0.00008703587,0.50883824,0.0020478603,0.48304304,0.0050442424,0.000039539187],"about_ca_topic_score_codex":0.0009209368,"about_ca_topic_score_gemma":0.0010780246,"teacher_disagreement_score":0.00434071,"about_ca_system_score_codex":0.0013482851,"about_ca_system_score_gemma":0.0011517769,"threshold_uncertainty_score":0.022956133},"labels":[],"label_agreement":null},{"id":"W2980402111","doi":"10.1109/tfuzz.2019.2947231","title":"Hyperplane Division in Fuzzy C-Means: Clustering Big Data","year":2019,"lang":"en","type":"article","venue":"IEEE Transactions on Fuzzy Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Alberta Innovates","keywords":"Cluster analysis; Disjoint sets; Hyperplane; Fuzzy clustering; Correlation clustering; Single-linkage clustering; CURE data clustering algorithm; Linear subspace; Computer science; Big data; Data mining; Division (mathematics); Clustering high-dimensional data; Data stream clustering; Mathematics; Canopy clustering algorithm; Data set; Artificial intelligence; Combinatorics","score_opus":0.06256838438048681,"score_gpt":0.29738658859684824,"score_spread":0.23481820421636143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2980402111","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060014473,0.00047332043,0.99247825,0.000114410585,0.00007131133,0.00006855374,0.000028099084,0.00026936678,0.0004952615],"genre_scores_gemma":[0.16267712,0.00062231126,0.83471894,0.00014007013,0.00014436408,0.00022224423,0.00017984366,0.000101984944,0.0011932258],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975432,0.0009258783,0.00011215891,0.00043450965,0.0008389151,0.00014533695],"domain_scores_gemma":[0.99806947,0.00078884047,0.00014978723,0.0003154602,0.0005778225,0.00009858057],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032597496,0.0011761922,0.0013054885,0.0017236081,0.0012831802,0.0014906285,0.0017490459,0.001734492,0.0013236463],"category_scores_gemma":[0.006285013,0.0004772616,0.0007770231,0.0028188962,0.0018405942,0.0020775325,0.0018886126,0.0020322613,0.000697675],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046242587,0.00016905433,0.0022171403,0.0004226256,0.0002217448,0.00015329331,0.0006660635,0.3817755,0.01203549,0.0623715,0.0063884244,0.5331167],"study_design_scores_gemma":[0.000021366144,0.00006443574,0.0003914882,0.000024098528,0.000015343441,0.000047531783,0.00008142202,0.96591616,0.006900441,0.02319935,0.0032995583,0.00003874484],"about_ca_topic_score_codex":0.006054402,"about_ca_topic_score_gemma":0.0032347252,"teacher_disagreement_score":0.006054402,"about_ca_system_score_codex":0.0009332147,"about_ca_system_score_gemma":0.0013842705,"threshold_uncertainty_score":0.017239392},"labels":[],"label_agreement":null},{"id":"W2990431786","doi":"10.1016/j.dib.2019.104899","title":"Data on cut-edge for spatial clustering based on proximity graphs","year":2019,"lang":"en","type":"article","venue":"Data in Brief","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Data mining; Similarity (geometry); Computer science; Cluster (spacecraft); Single-linkage clustering; Enhanced Data Rates for GSM Evolution; Correlation clustering; Spatial analysis; Data point; CURE data clustering algorithm; Pattern recognition (psychology); Artificial intelligence; Mathematics; Image (mathematics); Statistics","score_opus":0.08391157172496429,"score_gpt":0.35313526980121274,"score_spread":0.26922369807624846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990431786","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016300024,0.0021145905,0.9635349,0.0014902831,0.00034372814,0.00030648062,0.0044952515,0.0015208761,0.009893897],"genre_scores_gemma":[0.14569007,0.0023919388,0.8361094,0.0005781502,0.0002582471,0.0007842454,0.009827049,0.00044302066,0.0039178124],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99632746,0.00097445335,0.00031586096,0.0008418895,0.0014121833,0.00012816761],"domain_scores_gemma":[0.9894196,0.005618483,0.00057353463,0.0019634673,0.0022578926,0.00016705252],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024346705,0.00089580985,0.00076955615,0.0038702292,0.0013983792,0.0028512839,0.0020333845,0.002086274,0.0077202064],"category_scores_gemma":[0.02700228,0.0004075941,0.0011480985,0.007980599,0.0014326825,0.0043437304,0.0021692892,0.0022495342,0.003023297],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059481297,0.00014830472,0.005954135,0.0013855045,0.00018046927,0.00037031833,0.0008480455,0.09728356,0.005789117,0.31907523,0.05075156,0.5176189],"study_design_scores_gemma":[0.00006842604,0.0002617635,0.005219475,0.00034336638,0.00015474367,0.0014946581,0.0006492905,0.41810614,0.015630912,0.4004792,0.15742116,0.00017092508],"about_ca_topic_score_codex":0.0027492112,"about_ca_topic_score_gemma":0.0025489258,"teacher_disagreement_score":0.0077202064,"about_ca_system_score_codex":0.0015242659,"about_ca_system_score_gemma":0.0012834668,"threshold_uncertainty_score":0.025826633},"labels":[],"label_agreement":null},{"id":"W2990511412","doi":"10.1109/access.2020.2968290","title":"K-MACE and Kernel K-MACE Clustering","year":2020,"lang":"en","type":"preprint","venue":"IEEE Access","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Mace; Cluster analysis; Estimator; Kernel (algebra); Computer science; Mathematics; Cluster (spacecraft); Algorithm; Artificial intelligence; Statistics; Combinatorics","score_opus":0.09332408950713195,"score_gpt":0.38665922038682365,"score_spread":0.2933351308796917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990511412","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004739114,0.0006808368,0.99267733,0.00019618153,0.000084861545,0.000070485534,0.00010292032,0.00037346096,0.0010747401],"genre_scores_gemma":[0.17513703,0.0006364487,0.8184577,0.0002871046,0.00015954369,0.00033444617,0.0008028583,0.00042161538,0.0037631546],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99312335,0.0022385574,0.00046472895,0.0014637465,0.002378779,0.00033092708],"domain_scores_gemma":[0.98336303,0.0053802296,0.0013058943,0.0044133537,0.0052094534,0.00032808815],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0064306,0.001300963,0.0019817953,0.0032562546,0.0017372345,0.0027374143,0.0034677528,0.0029646337,0.0025159435],"category_scores_gemma":[0.03740092,0.0007247764,0.001769442,0.00333154,0.002315753,0.0042123585,0.0031596813,0.0026802132,0.0017964611],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005466985,0.00016523333,0.0036922956,0.0006641466,0.00045072768,0.00017400789,0.0006318177,0.49523655,0.0070106597,0.14086804,0.011431095,0.33912873],"study_design_scores_gemma":[0.000021166958,0.000051450894,0.0013758531,0.00003739575,0.000033311066,0.0001868971,0.00007607489,0.9419079,0.004734457,0.043609846,0.007899689,0.000065968496],"about_ca_topic_score_codex":0.004520086,"about_ca_topic_score_gemma":0.0042929137,"teacher_disagreement_score":0.0064306,"about_ca_system_score_codex":0.0016115246,"about_ca_system_score_gemma":0.0022373185,"threshold_uncertainty_score":0.034008622},"labels":[],"label_agreement":null},{"id":"W2991006005","doi":"10.1007/978-981-15-0994-0_7","title":"A Hybridized Data Clustering for Breast Cancer Prognosis and Risk Exposure Using Fuzzy C-means and Cohort Intelligence","year":2019,"lang":"en","type":"book-chapter","venue":"Algorithms for intelligent systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Cluster analysis; Breast cancer; Computer science; Artificial intelligence; Fuzzy logic; Fuzzy clustering; Machine learning; Metaheuristic; Data mining; Cancer; Medicine; Internal medicine","score_opus":0.09180944651642126,"score_gpt":0.34054516961173614,"score_spread":0.2487357230953149,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2991006005","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004393582,0.0002074537,0.99242425,0.00011080015,0.0000830363,0.000058321682,0.00014999883,0.000755928,0.0018165263],"genre_scores_gemma":[0.035671163,0.00020889961,0.95784813,0.00007284144,0.000052643543,0.000114220325,0.0004660645,0.00014060627,0.0054253777],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99928206,0.00012917406,0.000042717464,0.00019556371,0.0003128621,0.000037672005],"domain_scores_gemma":[0.9993204,0.0002128462,0.000019110228,0.000104275,0.00032221677,0.000021250353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015225948,0.0005380433,0.0008458199,0.0015491741,0.00084640837,0.0010954092,0.0015233807,0.00083792914,0.003406248],"category_scores_gemma":[0.0022533562,0.00045248854,0.0017370367,0.0021460413,0.00042707942,0.00076823914,0.00092137215,0.0009109522,0.0014999048],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011682445,0.00010080333,0.0027768123,0.00015935666,0.00027490547,0.00012447228,0.00025538402,0.09872806,0.018424233,0.035886142,0.01270578,0.8304472],"study_design_scores_gemma":[0.000011289453,0.00009664659,0.003470461,0.000032647367,0.00010332892,0.00029754033,0.00009325577,0.9402844,0.0110619515,0.028038431,0.016442116,0.00006797117],"about_ca_topic_score_codex":0.011992669,"about_ca_topic_score_gemma":0.011717043,"teacher_disagreement_score":0.011992669,"about_ca_system_score_codex":0.0010579439,"about_ca_system_score_gemma":0.0012117807,"threshold_uncertainty_score":0.023845732},"labels":[],"label_agreement":null},{"id":"W2995714961","doi":"10.1109/cwit.2019.8929909","title":"Correct Number of Clusters (CNC) Description Length in Arbitrary Shape Clustering","year":2019,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Cluster analysis; Computer science; Cluster (spacecraft); k-medians clustering; Determining the number of clusters in a data set; Compact space; Separable space; Context (archaeology); Data mining; Point (geometry); Algorithm; Correlation clustering; Mathematics; Artificial intelligence; CURE data clustering algorithm","score_opus":0.023176129352283057,"score_gpt":0.2908982372374981,"score_spread":0.267722107885215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2995714961","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021931464,0.00066288543,0.9759043,0.0002656808,0.000044570348,0.000044830384,0.00009109926,0.00018685279,0.00086829014],"genre_scores_gemma":[0.54776144,0.00079135085,0.44793093,0.00021138907,0.000104948595,0.00024462675,0.00063694344,0.0002758619,0.002042499],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99311066,0.0026203168,0.00034828924,0.0012895431,0.0023105852,0.00032062278],"domain_scores_gemma":[0.97330314,0.017447384,0.0021703308,0.003021015,0.003495775,0.0005624285],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007105469,0.0008141936,0.0012393933,0.0018137435,0.0014938768,0.0020078945,0.0030653658,0.0017161195,0.00072496117],"category_scores_gemma":[0.043712594,0.00058179797,0.000735241,0.002440044,0.0029462578,0.0035532857,0.0028174866,0.0020818708,0.00033060313],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034300523,0.00004990004,0.0049106176,0.0002567587,0.00008678992,0.00020147245,0.0003140837,0.7941903,0.0037751852,0.0801602,0.0022594691,0.11345217],"study_design_scores_gemma":[0.000008913488,0.000037254154,0.0007111555,0.00002444047,0.000010316207,0.00009956398,0.000042859356,0.96784973,0.002879986,0.02755423,0.00074544165,0.000036107962],"about_ca_topic_score_codex":0.007743191,"about_ca_topic_score_gemma":0.0055104992,"teacher_disagreement_score":0.007743191,"about_ca_system_score_codex":0.0031855663,"about_ca_system_score_gemma":0.0034598242,"threshold_uncertainty_score":0.03757775},"labels":[],"label_agreement":null},{"id":"W3011819927","doi":"10.1142/s0218488520500087","title":"Estimating and Controlling Overlap in Gaussian Mixtures for Clustering Methods Evaluation","year":2020,"lang":"en","type":"article","venue":"International Journal of Uncertainty Fuzziness and Knowledge-Based Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thales (Canada)","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Mixture model; Set (abstract data type); Artificial intelligence; Data set; Machine learning; Pattern recognition (psychology)","score_opus":0.05904577131457585,"score_gpt":0.40376231642250937,"score_spread":0.34471654510793354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011819927","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01377921,0.0002548092,0.98480195,0.00008610837,0.00002079689,0.00010920802,0.00003908531,0.0002610676,0.00064781134],"genre_scores_gemma":[0.3085519,0.00019597808,0.68985975,0.000068274385,0.00004081409,0.0004520684,0.00020515265,0.0002670277,0.00035904977],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.952856,0.024836378,0.0032547694,0.003219112,0.014641847,0.0011918551],"domain_scores_gemma":[0.82103264,0.14234826,0.009332345,0.012097856,0.013938371,0.001250576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.060623657,0.002529714,0.0022197033,0.0070136418,0.002207185,0.0051715495,0.0025479756,0.0029714748,0.0014353893],"category_scores_gemma":[0.23614253,0.0010078283,0.0016818229,0.003172832,0.0049207476,0.007144297,0.006533527,0.0030192682,0.00031871584],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005999128,0.00032378166,0.0129453195,0.00058931875,0.0004156394,0.00019485745,0.0014338068,0.6141857,0.014216016,0.17418118,0.0011442779,0.17977028],"study_design_scores_gemma":[0.000023484512,0.00023276401,0.0024853237,0.00013030606,0.00005594429,0.0001207725,0.00021321932,0.92977357,0.016217308,0.04922977,0.0014282381,0.00008929757],"about_ca_topic_score_codex":0.0036851258,"about_ca_topic_score_gemma":0.0025008055,"teacher_disagreement_score":0.060623657,"about_ca_system_score_codex":0.0039850683,"about_ca_system_score_gemma":0.0031727355,"threshold_uncertainty_score":0.32061237},"labels":[],"label_agreement":null},{"id":"W3016108274","doi":"10.1093/bioinformatics/btaa243","title":"Pooled variable scaling for cluster analysis","year":2020,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; KU Leuven","keywords":"Cluster (spacecraft); Scaling; Variable (mathematics); Computer science; Statistics; Mathematics; Programming language","score_opus":0.03153041412441573,"score_gpt":0.292185891279953,"score_spread":0.2606554771555373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016108274","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00082110113,0.00029715206,0.99720097,0.00014054833,0.00010433774,0.000115397655,0.00024434764,0.0005445188,0.00053174276],"genre_scores_gemma":[0.05454522,0.00066311983,0.9387827,0.00032140993,0.000537876,0.0015669253,0.0015916264,0.0010391827,0.0009519353],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9795725,0.010763131,0.0012022494,0.0043320763,0.0037094415,0.0004205557],"domain_scores_gemma":[0.9654409,0.017145945,0.002493835,0.008475514,0.005898864,0.00054489454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018092316,0.0023935998,0.0025712654,0.004422941,0.0018002661,0.0035941745,0.0034239858,0.0018848402,0.0073255175],"category_scores_gemma":[0.08534949,0.0009827511,0.0037943348,0.006996623,0.0036762937,0.003470336,0.005187569,0.004698406,0.003973896],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038775703,0.00014358453,0.0047033713,0.0023348841,0.001325622,0.00029162833,0.0012567294,0.103095524,0.00791371,0.3416163,0.034617737,0.50231326],"study_design_scores_gemma":[0.000097886325,0.00026307275,0.0034440928,0.00062351464,0.00028385976,0.00029222635,0.00024936008,0.4082718,0.0069558285,0.5313033,0.048036937,0.00017811071],"about_ca_topic_score_codex":0.0018947598,"about_ca_topic_score_gemma":0.0012754344,"teacher_disagreement_score":0.018092316,"about_ca_system_score_codex":0.001750095,"about_ca_system_score_gemma":0.0033158609,"threshold_uncertainty_score":0.0956825},"labels":[],"label_agreement":null},{"id":"W3022527078","doi":"10.1142/s0129065702001060","title":"ON COMPUTING THE FUZZIFIER IN ↓FLVQ: A DATA DRIVEN APPROACH","year":2002,"lang":"en","type":"article","venue":"International Journal of Neural Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Cluster analysis; Fuzzy clustering; Data mining; Computer science; Heuristic; Fuzzy logic; Correlation clustering; Set (abstract data type); Artificial intelligence; Algorithm","score_opus":0.11912871943748654,"score_gpt":0.3499688145090679,"score_spread":0.23084009507158137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3022527078","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033626324,0.00012142221,0.9955948,0.00010646664,0.000014346771,0.000051282062,0.00004828492,0.00020991964,0.00049076247],"genre_scores_gemma":[0.116020896,0.00026196326,0.88138556,0.0001712896,0.000053182455,0.00024301065,0.00034918712,0.00014497053,0.0013700314],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960836,0.0010016185,0.00025604933,0.00061756483,0.0017947662,0.0002463351],"domain_scores_gemma":[0.99112636,0.005318084,0.00037983857,0.0009164343,0.002116755,0.00014243173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005865227,0.0009930823,0.0016233526,0.0031227549,0.00136825,0.0034940685,0.0035375885,0.0017383306,0.0029723817],"category_scores_gemma":[0.018966224,0.00086612994,0.001483656,0.0025570225,0.0018845892,0.0045012124,0.002736376,0.002099533,0.00081900327],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002339139,0.00011611604,0.0023719554,0.00026950054,0.000090434965,0.00015096914,0.0004942475,0.40069407,0.004733815,0.13003162,0.0026785338,0.45813483],"study_design_scores_gemma":[0.000013422393,0.00003651975,0.0003340564,0.000034483477,0.000012758583,0.000059899212,0.000047223304,0.9301896,0.001737594,0.0655654,0.0019449564,0.000024172268],"about_ca_topic_score_codex":0.010220003,"about_ca_topic_score_gemma":0.008790004,"teacher_disagreement_score":0.010220003,"about_ca_system_score_codex":0.0023274356,"about_ca_system_score_gemma":0.0020597873,"threshold_uncertainty_score":0.031018674},"labels":[],"label_agreement":null},{"id":"W3037218554","doi":"","title":"Statistical guarantees for local graph clustering","year":2020,"lang":"en","type":"article","venue":"Journal of Machine Learning Research","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"PageRank; Cluster analysis; Computer science; Graph; Theoretical computer science; Cluster (spacecraft); Clustering coefficient; Path (computing); Mathematics; Algorithm; Artificial intelligence","score_opus":0.07959760419103978,"score_gpt":0.4021193523576084,"score_spread":0.3225217481665686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037218554","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021576412,0.0004912241,0.96994007,0.0009031841,0.000048935497,0.00006773396,0.00049413653,0.0029347248,0.0035435306],"genre_scores_gemma":[0.7298564,0.0009105541,0.25802267,0.0008749522,0.00044170942,0.0004970606,0.002374311,0.0013528971,0.0056694653],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9929463,0.002320331,0.00025038767,0.0017539718,0.0021382272,0.0005908761],"domain_scores_gemma":[0.943094,0.03330391,0.004735878,0.012493761,0.004877054,0.0014953522],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057116533,0.001690186,0.0022759172,0.0023371074,0.0013007363,0.0025823442,0.0043925177,0.0025784222,0.0052499287],"category_scores_gemma":[0.05561022,0.0010812161,0.0011757341,0.0031155623,0.0027530456,0.0059633916,0.003997232,0.0034595975,0.0028240962],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047030157,0.0001261815,0.0022532966,0.00039410577,0.00011110423,0.00021551082,0.0002498368,0.78001076,0.0043674856,0.13556764,0.010935086,0.06529866],"study_design_scores_gemma":[0.00003277632,0.000068392685,0.0003210048,0.000016730659,0.000009598919,0.00006197731,0.000029229977,0.9230369,0.0010678818,0.07426909,0.0010711696,0.00001531713],"about_ca_topic_score_codex":0.002202871,"about_ca_topic_score_gemma":0.0029231107,"teacher_disagreement_score":0.0057116533,"about_ca_system_score_codex":0.0022948594,"about_ca_system_score_gemma":0.0027037004,"threshold_uncertainty_score":0.030206501},"labels":[],"label_agreement":null},{"id":"W3038456932","doi":"10.3390/a13070158","title":"Fuzzy C-Means Clustering Algorithm with Multiple Fuzzification Coefficients","year":2020,"lang":"en","type":"article","venue":"Algorithms","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Foundation for Science and Technology Development","keywords":"Cluster analysis; Fuzzy clustering; Correlation clustering; CURE data clustering algorithm; Canopy clustering algorithm; Data mining; Computer science; FLAME clustering; Fuzzy set; Artificial intelligence; Algorithm; Data stream clustering; Fuzzy logic; Mathematics; Pattern recognition (psychology)","score_opus":0.02957732675180462,"score_gpt":0.27321826288198814,"score_spread":0.2436409361301835,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3038456932","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0050686365,0.00049542874,0.9918777,0.00009037764,0.00007652021,0.00012790586,0.000060021557,0.00046868192,0.0017346989],"genre_scores_gemma":[0.08378393,0.0005432808,0.9127035,0.00006887659,0.000050268314,0.00034504736,0.00024640711,0.00006693579,0.0021916563],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99816847,0.00022153315,0.000120544195,0.000481336,0.000882906,0.00012524826],"domain_scores_gemma":[0.99888855,0.00027041094,0.000082745624,0.00009994124,0.0006292209,0.000029120029],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015382678,0.0013671011,0.0017047387,0.0024453935,0.0019119477,0.0015354626,0.002120067,0.0020316646,0.0015031492],"category_scores_gemma":[0.0037677365,0.0005725714,0.0013712196,0.0030950753,0.0007166403,0.0014289764,0.000981176,0.0016400589,0.00079098187],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019069044,0.000100876925,0.0014697852,0.0004360496,0.00025553777,0.00016972124,0.00033969455,0.36321247,0.015072742,0.020710418,0.0078352485,0.59020674],"study_design_scores_gemma":[0.000027125949,0.000040872026,0.00061646476,0.000040660958,0.000040462746,0.0001589967,0.000040892413,0.9765768,0.007879425,0.0068184,0.0077008116,0.000059061113],"about_ca_topic_score_codex":0.019680986,"about_ca_topic_score_gemma":0.012349476,"teacher_disagreement_score":0.019680986,"about_ca_system_score_codex":0.0017808125,"about_ca_system_score_gemma":0.002948074,"threshold_uncertainty_score":0.039132833},"labels":[],"label_agreement":null},{"id":"W3039221356","doi":"10.1109/tcss.2020.3003538","title":"Discovering Density-Based Clustering Structures Using Neighborhood Distance Entropy Consistency","year":2020,"lang":"en","type":"article","venue":"IEEE Transactions on Computational Social Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; DBSCAN; Correlation clustering; CURE data clustering algorithm; Pattern recognition (psychology); Single-linkage clustering; Artificial intelligence; Outlier; Fuzzy clustering; Entropy (arrow of time); k-medians clustering; Mathematics; Computer science; Determining the number of clusters in a data set; Algorithm; Data mining; Physics","score_opus":0.042809567367067224,"score_gpt":0.29744679765202353,"score_spread":0.25463723028495633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3039221356","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046373602,0.00027940515,0.95155543,0.00011764782,0.000013001003,0.00009174663,0.00013680068,0.00035951825,0.0010728462],"genre_scores_gemma":[0.58091444,0.00027465375,0.41677338,0.000078074685,0.000048231374,0.000185446,0.0007560282,0.00018697501,0.0007827997],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980332,0.00053444837,0.00013858432,0.00047772366,0.0007036522,0.00011240658],"domain_scores_gemma":[0.99434316,0.0027248925,0.00081911735,0.00059447426,0.0013720854,0.0001463559],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033208143,0.00080982264,0.0013622527,0.0049264403,0.0010962277,0.0021515982,0.0015095364,0.00097247213,0.0006052984],"category_scores_gemma":[0.012097354,0.00059187674,0.0010293277,0.0025165277,0.0012394453,0.002193548,0.001941644,0.00096489786,0.0002885118],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002881301,0.000137014,0.022382855,0.0002802532,0.00038910715,0.00025444187,0.0007957143,0.69311076,0.010405046,0.049307764,0.0031548338,0.21949404],"study_design_scores_gemma":[0.000013339228,0.000025046904,0.0018774964,0.000015288351,0.000022080178,0.00007129703,0.00008091712,0.9786904,0.0022735663,0.016231705,0.0006740133,0.000024843388],"about_ca_topic_score_codex":0.0050881077,"about_ca_topic_score_gemma":0.0045450865,"teacher_disagreement_score":0.0050881077,"about_ca_system_score_codex":0.001576413,"about_ca_system_score_gemma":0.001612526,"threshold_uncertainty_score":0.01756233},"labels":[],"label_agreement":null},{"id":"W3042760544","doi":"10.4018/978-1-7998-3476-2.ch005","title":"Sequence Clustering Techniques in Educational Data Mining","year":2020,"lang":"en","type":"book-chapter","venue":"IGI Global eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Medical Council of Canada","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Sequence (biology); CURE data clustering algorithm; Correlation clustering; Clustering high-dimensional data; Data stream clustering; Artificial intelligence","score_opus":0.09898537637041742,"score_gpt":0.35664151093204677,"score_spread":0.25765613456162934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3042760544","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016846932,0.015755022,0.9746739,0.00071423815,0.00034780664,0.00016483084,0.00052264374,0.0011021232,0.0050348006],"genre_scores_gemma":[0.025010653,0.016329145,0.94948965,0.0003361398,0.00044078723,0.00036692465,0.0016944852,0.00023102196,0.0061011966],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99726,0.0009526557,0.00022043879,0.00053201115,0.00094476267,0.0000902263],"domain_scores_gemma":[0.99677235,0.0020386209,0.00015035903,0.00032643342,0.00064732484,0.00006491547],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029927748,0.0015415172,0.0016850614,0.006925508,0.001198919,0.0022710273,0.002250425,0.0016489287,0.0037650166],"category_scores_gemma":[0.007689663,0.00077789585,0.0015954503,0.016540963,0.0009565012,0.0031852939,0.0013171918,0.0025161505,0.004345201],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005621546,0.00011307207,0.001415136,0.0013438856,0.00025068005,0.00018784382,0.00045617367,0.05021978,0.0016318902,0.059618644,0.03133537,0.85337126],"study_design_scores_gemma":[0.000033053442,0.00012748956,0.002998163,0.00093663385,0.00013920359,0.0009827495,0.0005387798,0.49315482,0.006316855,0.30705816,0.18756019,0.00015385175],"about_ca_topic_score_codex":0.0041079978,"about_ca_topic_score_gemma":0.004319683,"teacher_disagreement_score":0.006925508,"about_ca_system_score_codex":0.001270332,"about_ca_system_score_gemma":0.001566337,"threshold_uncertainty_score":0.015827537},"labels":[],"label_agreement":null},{"id":"W3089543401","doi":"10.1007/978-3-030-58942-4_9","title":"An Ising Framework for Constrained Clustering on Special Purpose Hardware","year":2020,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Cluster analysis; Heuristic; Benchmark (surveying); Theoretical computer science; Constraint programming; Partition (number theory); Constrained clustering; Constraint (computer-aided design); Machine learning; Artificial intelligence; Correlation clustering; Mathematical optimization; CURE data clustering algorithm","score_opus":0.036714655109650836,"score_gpt":0.3189121647864331,"score_spread":0.28219750967678225,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3089543401","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027813632,0.0018306517,0.9002105,0.0009755182,0.00034002136,0.00006786628,0.00023055209,0.000799775,0.06773154],"genre_scores_gemma":[0.6980526,0.0024972784,0.2500113,0.0005771716,0.00041349506,0.00025049207,0.0003309975,0.0006751026,0.047191586],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996977,0.00007002408,0.000011149534,0.00004332938,0.00011731176,0.000060520153],"domain_scores_gemma":[0.9995591,0.00016502995,0.00002833407,0.00014334155,0.00006800361,0.00003616422],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041102737,0.0005689936,0.0009034313,0.00073433353,0.0010603605,0.0014882479,0.00185808,0.0013649195,0.0087272525],"category_scores_gemma":[0.0013342542,0.00040167783,0.0006478973,0.001203615,0.0012164685,0.0019944478,0.0010740688,0.0017140019,0.0010622567],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002026264,0.000017057613,0.00004270897,0.000055535227,0.00001093566,0.000031400872,0.000029192392,0.047536083,0.0011094046,0.9394959,0.0032336235,0.008417892],"study_design_scores_gemma":[0.000008031994,0.000011433415,0.000059599377,0.0000138295,0.0000070478745,0.000030355446,0.000013153671,0.2859761,0.00039034264,0.70947075,0.004007997,0.000011460915],"about_ca_topic_score_codex":0.0021820124,"about_ca_topic_score_gemma":0.0041891746,"teacher_disagreement_score":0.0087272525,"about_ca_system_score_codex":0.0012099238,"about_ca_system_score_gemma":0.00096161815,"threshold_uncertainty_score":0.029195547},"labels":[],"label_agreement":null},{"id":"W3090078726","doi":"10.20382/jocg.v11i1a11","title":"An improved cost function for hierarchical cluster trees","year":2019,"lang":"en","type":"article","venue":"Journal of Computational Geometry (Carleton University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Ohio State University; National Science Foundation","keywords":"Hierarchical clustering; Cluster analysis; Granularity; Tree (set theory); Function (biology); Mathematics; Graph; Similarity (geometry); Tree structure; Computer science; Set (abstract data type); Hierarchical clustering of networks; Data mining; Algorithm; Correlation clustering; Combinatorics; CURE data clustering algorithm; Artificial intelligence; Statistics","score_opus":0.012001304747632493,"score_gpt":0.26031672254219884,"score_spread":0.24831541779456634,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3090078726","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011429044,0.0005326101,0.98368734,0.00052462687,0.00009368736,0.0001477529,0.00024932457,0.00038500974,0.0029506392],"genre_scores_gemma":[0.25271446,0.0007181219,0.734918,0.0004585945,0.00020470629,0.00062747265,0.001367736,0.0006242662,0.008366722],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99703705,0.000917497,0.00015496789,0.0003522382,0.0011908896,0.00034728524],"domain_scores_gemma":[0.99499476,0.0021459726,0.00030797473,0.000537781,0.0017193169,0.0002942886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036450154,0.0014091992,0.001864888,0.0024410426,0.0011098007,0.001913223,0.0036006456,0.003036758,0.0056217476],"category_scores_gemma":[0.011810967,0.0005466718,0.0015149782,0.0033943276,0.0009888894,0.0051978584,0.002322358,0.002611193,0.0011465187],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023335218,0.00016653979,0.0014301876,0.00025976988,0.00006737715,0.0001610842,0.00013900224,0.703162,0.0043088556,0.11484453,0.014141619,0.16108571],"study_design_scores_gemma":[0.0000122640895,0.000046970123,0.00025340382,0.000016695663,0.00001077916,0.00006745964,0.000018812281,0.97824734,0.0007039233,0.017823065,0.0027854622,0.000013695365],"about_ca_topic_score_codex":0.005589978,"about_ca_topic_score_gemma":0.0046326965,"teacher_disagreement_score":0.0056217476,"about_ca_system_score_codex":0.0033516975,"about_ca_system_score_gemma":0.0025431484,"threshold_uncertainty_score":0.024318397},"labels":[],"label_agreement":null},{"id":"W3092332002","doi":"10.1109/tcyb.2020.3023373","title":"A <i>Group</i>-Based Distance Learning Method for Semisupervised Fuzzy Clustering","year":2020,"lang":"en","type":"article","venue":"IEEE Transactions on Cybernetics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"China Electronics Technology Group Corporation; Higher Education Discipline Innovation Project; Academy of Finland; National Natural Science Foundation of China","keywords":"Cluster analysis; Pairwise comparison; Artificial intelligence; Constrained clustering; Fuzzy clustering; Computer science; Machine learning; Fuzzy logic; Artificial neural network; Distance measures; Data mining; Mathematics; CURE data clustering algorithm","score_opus":0.03472825080664537,"score_gpt":0.306341575228065,"score_spread":0.2716133244214196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092332002","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018054097,0.00012820977,0.99717903,0.0000960169,0.000021890744,0.000028461309,0.000023887917,0.00011597046,0.0006011376],"genre_scores_gemma":[0.1420267,0.00030616103,0.85313505,0.00023070718,0.00009184278,0.00033233655,0.00031021904,0.00015488727,0.0034121033],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981365,0.0005454545,0.0001356201,0.00060830585,0.00049472577,0.000079247555],"domain_scores_gemma":[0.9986059,0.00048005674,0.00017417787,0.00020939729,0.00046169557,0.00006881937],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019329599,0.0012065949,0.0014244065,0.001501993,0.0009373369,0.0012444017,0.0023059468,0.0021281121,0.0024488473],"category_scores_gemma":[0.004196513,0.00044264275,0.0014071816,0.0015834265,0.0013082873,0.0024079552,0.0016841913,0.0023129454,0.0009570409],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016115585,0.00015038649,0.0010595698,0.00048191648,0.00025119094,0.0001858745,0.00053909735,0.46000218,0.017943893,0.10560589,0.008060825,0.40555802],"study_design_scores_gemma":[0.000010959947,0.000047316706,0.00013170419,0.000014683832,0.000012675432,0.000075371776,0.000023649456,0.9744863,0.003176377,0.01915475,0.0028385816,0.000027637281],"about_ca_topic_score_codex":0.002821507,"about_ca_topic_score_gemma":0.0028194431,"teacher_disagreement_score":0.002821507,"about_ca_system_score_codex":0.0013690008,"about_ca_system_score_gemma":0.001358762,"threshold_uncertainty_score":0.010222614},"labels":[],"label_agreement":null},{"id":"W3094325709","doi":"10.18280/ria.340408","title":"An Improved Deep Learning Algorithm for Risk Prediction of Corporate Internet Reporting","year":2020,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Artificial intelligence; Machine learning; Autoregressive model; Autoregressive–moving-average model; Time series; The Internet; Index (typography); Data mining; Algorithm; Deep learning; Econometrics; Mathematics","score_opus":0.0884579502347997,"score_gpt":0.3171150574558491,"score_spread":0.22865710722104943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3094325709","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05898128,0.0009950576,0.93625176,0.00059357623,0.000122199,0.00005339614,0.00020820517,0.0011253844,0.0016692225],"genre_scores_gemma":[0.7530002,0.00076571235,0.23786116,0.0003615462,0.00015709201,0.00017754945,0.0009847692,0.00009809019,0.0065938537],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995123,0.00011175743,0.00005277608,0.00013213731,0.00011261075,0.00007836168],"domain_scores_gemma":[0.9992036,0.0003483484,0.00007886534,0.000061859624,0.0002736769,0.00003360332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014773612,0.00082826224,0.0009925568,0.0012590518,0.00042165568,0.0009217519,0.001423157,0.0011881097,0.0015903838],"category_scores_gemma":[0.0032888395,0.00046632657,0.00090031716,0.0011914439,0.00036567662,0.0015051815,0.00096544536,0.0016220832,0.00048544703],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012257212,0.0001522506,0.003501836,0.00006521083,0.00010390811,0.00008762499,0.000063176965,0.6116617,0.002447497,0.0046579554,0.003621016,0.37351522],"study_design_scores_gemma":[0.0000026439013,0.000007788264,0.00009843659,0.000002247405,0.0000037647146,0.000004838602,0.0000015320184,0.9987412,0.00023299642,0.000798179,0.00010448266,0.0000019492006],"about_ca_topic_score_codex":0.011152089,"about_ca_topic_score_gemma":0.008232021,"teacher_disagreement_score":0.011152089,"about_ca_system_score_codex":0.0010307634,"about_ca_system_score_gemma":0.0014123122,"threshold_uncertainty_score":0.022174358},"labels":[],"label_agreement":null},{"id":"W3097249106","doi":"10.1007/s10618-020-00720-w","title":"An exemplar-based clustering using efficient variational message passing","year":2020,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec en Outaouais","funders":"","keywords":"Cluster analysis; Computer science; Correlation clustering; Message passing; Affinity propagation; Constrained clustering; Graph partition; CURE data clustering algorithm; Canopy clustering algorithm; Data mining; Graph; Theoretical computer science; Artificial intelligence","score_opus":0.1118581280584065,"score_gpt":0.3593757391597357,"score_spread":0.2475176111013292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3097249106","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003207884,0.0000544327,0.9959266,0.000079922815,0.000031146872,0.000033347365,0.00002178038,0.00024533653,0.00039962397],"genre_scores_gemma":[0.12414245,0.0001354405,0.87103724,0.0001414105,0.00007982337,0.00019110575,0.00030125282,0.00022236051,0.0037489664],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984395,0.00045574832,0.00008758149,0.00033775766,0.00052023074,0.00015914835],"domain_scores_gemma":[0.99782526,0.0009211592,0.00012869295,0.0003438941,0.00062335114,0.00015769037],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002456837,0.0010115693,0.0027719198,0.0016696809,0.0016264725,0.0019439452,0.0051489137,0.0029759647,0.0033931711],"category_scores_gemma":[0.0057723215,0.0012358003,0.0018201581,0.0022361905,0.0016322386,0.002377099,0.0039469865,0.0023200943,0.0013890556],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021721917,0.00017541088,0.00076049316,0.00015764857,0.00018497088,0.00009706475,0.00025023718,0.71175194,0.010414481,0.059464842,0.0038758772,0.21264975],"study_design_scores_gemma":[0.000006555282,0.000014628227,0.000036553432,0.0000026696187,0.000007160662,0.00001692642,0.000007862978,0.9942445,0.00081754714,0.004456814,0.0003811786,0.000007607154],"about_ca_topic_score_codex":0.012127617,"about_ca_topic_score_gemma":0.010142283,"teacher_disagreement_score":0.012127617,"about_ca_system_score_codex":0.0016930373,"about_ca_system_score_gemma":0.0024052775,"threshold_uncertainty_score":0.024114013},"labels":[],"label_agreement":null},{"id":"W3097529996","doi":"10.1016/j.patcog.2020.107748","title":"Projected fuzzy C-means clustering with locality preservation","year":2020,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":83,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Locality; Cluster analysis; Fuzzy clustering; Computer science; Fuzzy logic; Data mining; Curse of dimensionality; Clustering high-dimensional data; Partition (number theory); Benchmark (surveying); Pattern recognition (psychology); Artificial intelligence; Projection (relational algebra); Mathematics; Algorithm","score_opus":0.07571112933585794,"score_gpt":0.28262663107455643,"score_spread":0.2069155017386985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3097529996","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014294303,0.00012978082,0.9836881,0.00007366709,0.000032946027,0.000058622678,0.00010369463,0.000702869,0.0009160375],"genre_scores_gemma":[0.21886557,0.00013406586,0.77763546,0.00007197234,0.000052321306,0.00017529062,0.0005319187,0.00021332185,0.0023200486],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980478,0.00039155735,0.00011265592,0.00043420863,0.0008618634,0.00015188252],"domain_scores_gemma":[0.99786466,0.0003031564,0.00011539146,0.0005530377,0.0010938075,0.00006993508],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014688462,0.0007130155,0.0012890943,0.0017447294,0.0018409657,0.0017014424,0.0019627307,0.0012741855,0.0024838168],"category_scores_gemma":[0.0046381284,0.0006481008,0.0015345432,0.0026772502,0.0010519255,0.0015970302,0.0021547854,0.0010697228,0.0012650864],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00069085666,0.000185759,0.0020361429,0.00029087166,0.00025597913,0.00012337288,0.0003593336,0.24500805,0.03039898,0.018651612,0.005811721,0.6961873],"study_design_scores_gemma":[0.000022105693,0.000076567034,0.00075978326,0.0000116874835,0.000036185716,0.00013293188,0.00006193377,0.9755491,0.013033743,0.008571078,0.001709431,0.000035463963],"about_ca_topic_score_codex":0.014006089,"about_ca_topic_score_gemma":0.012241996,"teacher_disagreement_score":0.014006089,"about_ca_system_score_codex":0.0010399563,"about_ca_system_score_gemma":0.0028412044,"threshold_uncertainty_score":0.027849078},"labels":[],"label_agreement":null},{"id":"W3099831365","doi":"10.1007/s10618-015-0426-x","title":"Generalization of clustering agreements and distances for overlapping clusters and network communities","year":2015,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Generalization; Measure (data warehouse); Context (archaeology); Single-linkage clustering; Correlation clustering; Rand index; Contingency table; Cluster (spacecraft)","score_opus":0.11934352232843241,"score_gpt":0.3518487569469944,"score_spread":0.23250523461856198,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3099831365","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.067917116,0.00080797845,0.92451406,0.00071998156,0.00008862452,0.00014934414,0.00052093726,0.00032314836,0.004958825],"genre_scores_gemma":[0.61918485,0.0014276869,0.37161818,0.0003080779,0.00039056534,0.000517234,0.0017550678,0.00030885334,0.0044894787],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9907832,0.0029701374,0.0006410959,0.0028553223,0.002322295,0.00042790643],"domain_scores_gemma":[0.9243514,0.046606857,0.007830225,0.010204546,0.008800325,0.0022067174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0154444,0.0013222119,0.0021929876,0.0077506215,0.0030832775,0.004336335,0.005461771,0.002765425,0.0035253793],"category_scores_gemma":[0.11074733,0.0011080271,0.0020739539,0.006505052,0.0046343985,0.012689646,0.0053320765,0.0049826265,0.00087042095],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022085573,0.00017772354,0.010066546,0.0003943261,0.00021127627,0.00022656737,0.0022202113,0.11215893,0.0014645912,0.76793283,0.00482158,0.10010457],"study_design_scores_gemma":[0.000020949767,0.000046067715,0.002209862,0.00006790943,0.000052176725,0.00024625383,0.00042439668,0.29192472,0.0005645145,0.70205647,0.002345444,0.00004130535],"about_ca_topic_score_codex":0.003681614,"about_ca_topic_score_gemma":0.0029549778,"teacher_disagreement_score":0.0154444,"about_ca_system_score_codex":0.0032262604,"about_ca_system_score_gemma":0.0017886839,"threshold_uncertainty_score":0.08167875},"labels":[],"label_agreement":null},{"id":"W3105265316","doi":"10.1007/s10955-020-02651-8","title":"Correction To: Clustering in the Three and Four Color Cyclic Particle Systems in One Dimension","year":2020,"lang":"en","type":"article","venue":"Journal of Statistical Physics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Dimension (graph theory); Cluster analysis; Mathematics; Statistical physics; Particle system; Particle (ecology); Computer science; Combinatorics; Physics; Statistics; Computer graphics (images); Geology","score_opus":0.0678446815159414,"score_gpt":0.31649388882122165,"score_spread":0.24864920730528023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3105265316","genre_codex":"editorial","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00017962331,0.0005362411,0.0013341244,0.016010592,0.9794886,0.000021191165,0.0009601455,0.0006433776,0.00082607067],"genre_scores_gemma":[0.04637764,0.0057050185,0.020410195,0.071686246,0.6627311,0.00049611175,0.0075362287,0.0043353797,0.18072197],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9911124,0.0012675582,0.0013497353,0.0017981939,0.003362196,0.0011099323],"domain_scores_gemma":[0.9517942,0.008422486,0.002063334,0.00534272,0.028555108,0.0038221097],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004673952,0.0038628737,0.0063177496,0.006864659,0.005269711,0.006095467,0.009387796,0.011880754,0.14581476],"category_scores_gemma":[0.06761463,0.0024227663,0.0038436302,0.0056704967,0.0038523523,0.0050589107,0.0038813355,0.014240541,0.052673843],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000074165975,0.000013698813,0.00008065041,0.00021223137,0.000036962432,0.0001605345,0.000029072302,0.00018751192,0.00013243764,0.0016474578,0.9933303,0.0040948796],"study_design_scores_gemma":[0.00031469658,0.00009525211,0.0028795772,0.00032371868,0.00013882449,0.0016202366,0.00020547454,0.0045978753,0.0017524416,0.015043123,0.9727653,0.00026337546],"about_ca_topic_score_codex":0.013922941,"about_ca_topic_score_gemma":0.019319776,"teacher_disagreement_score":0.14581476,"about_ca_system_score_codex":0.0061565065,"about_ca_system_score_gemma":0.005932231,"threshold_uncertainty_score":0.487799},"labels":[],"label_agreement":null},{"id":"W3107330311","doi":"10.1007/s10489-020-02108-1","title":"Improving cluster recovery with feature rescaling factors","year":2021,"lang":"en","type":"article","venue":"Applied Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Cluster analysis; Normalization (sociology); Computer science; Data mining; Preprocessor; Feature (linguistics); Artificial intelligence; Pattern recognition (psychology); Data pre-processing; Machine learning","score_opus":0.018261338284883202,"score_gpt":0.2635628413197893,"score_spread":0.2453015030349061,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3107330311","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020383256,0.00033214162,0.9760702,0.000180857,0.00010908555,0.00006216835,0.00010538003,0.00211069,0.000646129],"genre_scores_gemma":[0.24149638,0.00030243272,0.75401175,0.00017779136,0.00013195383,0.00010994958,0.00071758754,0.0006663715,0.0023858517],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99801195,0.00043286916,0.00012113892,0.0007104526,0.0005449552,0.00017853794],"domain_scores_gemma":[0.9952786,0.0015626085,0.00028424073,0.0016784644,0.0010678761,0.00012820313],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024511453,0.0014110423,0.0016086302,0.0016386393,0.00098528,0.0012361406,0.0017574615,0.0014784096,0.003977603],"category_scores_gemma":[0.016934684,0.00048775296,0.0013022206,0.002491073,0.0010540993,0.0025498292,0.001948583,0.0016882208,0.0021665967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00075697224,0.00020647845,0.0025756303,0.00025164435,0.00019431367,0.00013008158,0.00028996228,0.092273474,0.04108703,0.012186617,0.01098143,0.83906627],"study_design_scores_gemma":[0.00006349643,0.00014589875,0.0016929582,0.00002420067,0.00008757086,0.00019296269,0.0001318024,0.9515361,0.023551516,0.017023241,0.0054966444,0.000053626747],"about_ca_topic_score_codex":0.005292862,"about_ca_topic_score_gemma":0.0056810216,"teacher_disagreement_score":0.005292862,"about_ca_system_score_codex":0.0005563699,"about_ca_system_score_gemma":0.0013062952,"threshold_uncertainty_score":0.013306379},"labels":[],"label_agreement":null},{"id":"W3111477678","doi":"10.17605/osf.io/pejaq","title":"FOCAL3D Simulated Data","year":2020,"lang":"en","type":"article","venue":"OSF Preprints (OSF Preprints)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science","score_opus":0.059571032367751214,"score_gpt":0.32486404666885876,"score_spread":0.2652930143011075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111477678","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.83785236,0.00049896096,0.09139692,0.0011703904,0.00028978093,0.00059869775,0.05279231,0.0026534337,0.0127471695],"genre_scores_gemma":[0.9176625,0.00011545133,0.02927647,0.00019929872,0.000028296507,0.0005512297,0.049444344,0.00031665602,0.0024057205],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976248,0.0011602879,0.00011134635,0.00048394638,0.0004107096,0.00020903142],"domain_scores_gemma":[0.9857853,0.0075690863,0.0004898733,0.0028676235,0.0028383953,0.00044967962],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042759054,0.0007850728,0.0010803369,0.0018263584,0.0011171965,0.0016134415,0.002699195,0.0028208809,0.008977602],"category_scores_gemma":[0.018811692,0.0007178943,0.0016069004,0.0029666896,0.0015034619,0.0009645862,0.0011836793,0.0018820766,0.0020596108],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013072138,0.00038568085,0.012616939,0.00030849877,0.00023940846,0.0002201279,0.00022184952,0.945507,0.0018818709,0.011756675,0.013639704,0.011915048],"study_design_scores_gemma":[0.0005870556,0.00031996414,0.012774159,0.00007379492,0.0000721239,0.00028449026,0.0002924547,0.9596024,0.005823908,0.010923233,0.00913847,0.000108029555],"about_ca_topic_score_codex":0.031912208,"about_ca_topic_score_gemma":0.02271065,"teacher_disagreement_score":0.031912208,"about_ca_system_score_codex":0.0027992227,"about_ca_system_score_gemma":0.0016885994,"threshold_uncertainty_score":0.0634529},"labels":[],"label_agreement":null},{"id":"W3119850208","doi":"10.1007/978-3-030-73050-5_43","title":"An Efficient K-Means Clustering Algorithm for Analysing COVID-19","year":2021,"lang":"en","type":"preprint","venue":"Advances in intelligent systems and computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Cluster analysis; Centroid; Computer science; Coronavirus disease 2019 (COVID-19); Pandemic; Similarity (geometry); Quality (philosophy); Data mining; Health care; k-means clustering; Algorithm; Artificial intelligence; Medicine; Economics; Disease; Economic growth","score_opus":0.04157513408384758,"score_gpt":0.38017399012812736,"score_spread":0.33859885604427975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3119850208","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035519255,0.0001524985,0.9926951,0.000045814515,0.00006957623,0.00011179172,0.00027618944,0.0021371634,0.0009599253],"genre_scores_gemma":[0.019524582,0.00009109213,0.97630364,0.00003822537,0.00002885408,0.00013392154,0.0011087641,0.00042642694,0.002344466],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967619,0.0004990919,0.00024491063,0.0007962592,0.0013902524,0.00030762271],"domain_scores_gemma":[0.9969902,0.0005545178,0.00015288936,0.00055382994,0.0016537675,0.0000947734],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016394273,0.0019133305,0.0025167896,0.0051941974,0.0028486156,0.0031906248,0.0036799405,0.0023769306,0.0064670458],"category_scores_gemma":[0.0070419023,0.0011382331,0.0022193866,0.0066719837,0.00092792633,0.002125044,0.0026215843,0.0020251463,0.008084379],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004051588,0.00025995084,0.002265853,0.000376791,0.0003289065,0.00021967075,0.00044087248,0.05908864,0.034857903,0.018263135,0.017277267,0.8662158],"study_design_scores_gemma":[0.00006645169,0.0000943404,0.0028150785,0.000051275456,0.00011074886,0.00055068877,0.00032503423,0.9196232,0.020845145,0.029502984,0.025856568,0.00015846199],"about_ca_topic_score_codex":0.014589507,"about_ca_topic_score_gemma":0.020370496,"teacher_disagreement_score":0.014589507,"about_ca_system_score_codex":0.001337138,"about_ca_system_score_gemma":0.0040128785,"threshold_uncertainty_score":0.029009223},"labels":[],"label_agreement":null},{"id":"W3120082291","doi":"10.1109/ieem45057.2020.9309941","title":"Business Applications for Current Developments in Big Data Clustering: An Overview","year":2020,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; Western University","funders":"","keywords":"Big data; Cluster analysis; Headline; Computer science; Data science; Raw data; Currency; Data mining; Business; Artificial intelligence","score_opus":0.4828641149166046,"score_gpt":0.44616679371825857,"score_spread":0.03669732119834601,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3120082291","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00097389077,0.9271355,0.030264756,0.0140715,0.002928377,0.00011881402,0.00022900864,0.00060340576,0.023674736],"genre_scores_gemma":[0.008107512,0.9358054,0.037516654,0.004682836,0.0061237197,0.00013018998,0.00064904813,0.0001933573,0.0067913453],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.997561,0.00056421314,0.00023803867,0.00035269294,0.0011537158,0.00013030066],"domain_scores_gemma":[0.98843783,0.006736337,0.00055791764,0.00044351222,0.003169216,0.0006552555],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004226752,0.0014319337,0.0010474623,0.010505615,0.0011703828,0.0059941416,0.0017297156,0.0036886483,0.009666632],"category_scores_gemma":[0.008849991,0.0010713913,0.0012931328,0.015914433,0.0017044985,0.009603128,0.0021245314,0.004133053,0.007934731],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007127689,0.00013546748,0.0012500176,0.0056779548,0.000050018585,0.00012492982,0.0002411365,0.0015187496,0.00087006437,0.0540571,0.13228813,0.8037152],"study_design_scores_gemma":[0.000008212331,0.00008704475,0.0010944054,0.0039690416,0.000040911793,0.00058891746,0.00033601234,0.0036428745,0.00056795444,0.03467449,0.95491624,0.00007383792],"about_ca_topic_score_codex":0.0017637393,"about_ca_topic_score_gemma":0.0019144362,"teacher_disagreement_score":0.010505615,"about_ca_system_score_codex":0.0019591956,"about_ca_system_score_gemma":0.0022807156,"threshold_uncertainty_score":0.032338083},"labels":[],"label_agreement":null},{"id":"W3122289361","doi":"10.18280/ria.340604","title":"Big Data Clustering Using Improvised Fuzzy C-Means Clustering","year":2020,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"MNIST database; Computer science; Cluster analysis; Artificial intelligence; Fuzzy clustering; Data mining; Convolutional neural network; Big data; Rand index; Fuzzy logic; Encoder; Pattern recognition (psychology); Deep learning; Machine learning","score_opus":0.21354878917626946,"score_gpt":0.34659951147689494,"score_spread":0.13305072230062548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122289361","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017905973,0.00032724513,0.97868437,0.0002082109,0.000062682986,0.00014535477,0.00016838437,0.0010327274,0.0014650982],"genre_scores_gemma":[0.27212566,0.00031749718,0.7243342,0.00012805004,0.000048030495,0.00021189946,0.0005865307,0.000103307306,0.0021448664],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99823064,0.00022606959,0.00014770486,0.00046712178,0.00078984036,0.00013877546],"domain_scores_gemma":[0.99783486,0.00049620715,0.00022964679,0.00026456395,0.0010988615,0.000075764656],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017563351,0.0009524018,0.0010902487,0.0027937747,0.0014443857,0.0017686144,0.002672495,0.0017722861,0.0011652957],"category_scores_gemma":[0.005080232,0.000488697,0.0015567814,0.0023007218,0.000883854,0.0017015364,0.0010712923,0.0014157459,0.00046803913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030180172,0.00014632284,0.0039772284,0.00027581793,0.00027895765,0.00018017339,0.00038302803,0.59900075,0.00739567,0.015305761,0.004102789,0.36865172],"study_design_scores_gemma":[0.000007289652,0.000026247615,0.00048689396,0.000016198625,0.000012171403,0.000049590344,0.0000401015,0.9886063,0.004153416,0.0052456926,0.0013313937,0.00002465606],"about_ca_topic_score_codex":0.027713846,"about_ca_topic_score_gemma":0.019636791,"teacher_disagreement_score":0.027713846,"about_ca_system_score_codex":0.0024310309,"about_ca_system_score_gemma":0.0028865202,"threshold_uncertainty_score":0.05510503},"labels":[],"label_agreement":null},{"id":"W3126183644","doi":"10.1007/s11634-022-00514-6","title":"Determinantal consensus clustering","year":2022,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données","keywords":"Cluster analysis; Point process; Single-linkage clustering; Mathematics; Computer science; Kernel (algebra); Similarity (geometry); Consensus clustering; Data mining; Artificial intelligence; Correlation clustering; CURE data clustering algorithm; Combinatorics; Statistics","score_opus":0.053565366313241526,"score_gpt":0.36025569701971255,"score_spread":0.306690330706471,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126183644","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011461968,0.00042988203,0.98090255,0.00033036558,0.0001447537,0.00004847888,0.0000787851,0.00049599924,0.00610717],"genre_scores_gemma":[0.6241547,0.0007937188,0.3486748,0.000512854,0.00040312958,0.00025641953,0.0010361517,0.0007658947,0.023402372],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99615127,0.001358851,0.0001644325,0.0010721693,0.0010209696,0.00023214368],"domain_scores_gemma":[0.99267346,0.0020045144,0.00041274878,0.002033559,0.0024334418,0.00044221996],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033155147,0.00080613245,0.0019959745,0.0026508651,0.0020482887,0.0027067184,0.0025358049,0.002153284,0.005756711],"category_scores_gemma":[0.012573305,0.0008660825,0.0012368024,0.0022266484,0.0025710575,0.002565794,0.003932743,0.002266012,0.0026528311],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026408737,0.00010062675,0.0018017656,0.00027558728,0.00023035471,0.00017283585,0.00048682268,0.22310592,0.008048375,0.5463808,0.017672999,0.20145981],"study_design_scores_gemma":[0.000018818955,0.000039779363,0.0006082726,0.000019429484,0.000020690333,0.00011064238,0.00008450648,0.81945497,0.00255164,0.17005306,0.006997537,0.0000407363],"about_ca_topic_score_codex":0.0021903692,"about_ca_topic_score_gemma":0.0028838783,"teacher_disagreement_score":0.005756711,"about_ca_system_score_codex":0.0016606364,"about_ca_system_score_gemma":0.0018443248,"threshold_uncertainty_score":0.019258082},"labels":[],"label_agreement":null},{"id":"W3130186424","doi":"10.1109/icjece.2020.3018433","title":"Evaluation of Dimensionality Reduction Techniques for Load Profiling Application in Smart Grid Environment","year":2021,"lang":"en","type":"article","venue":"Canadian Journal of Electrical and Computer Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"University of Regina","keywords":"Profiling (computer programming); Dimensionality reduction; Computer science; Smart grid; Grid; Reduction (mathematics); Artificial intelligence; Engineering; Mathematics; Operating system; Electrical engineering","score_opus":0.01787113559275199,"score_gpt":0.2570653424765557,"score_spread":0.23919420688380372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3130186424","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.62026,0.0035383399,0.35398048,0.0013291051,0.00042485332,0.0008127141,0.002233596,0.007269801,0.010151131],"genre_scores_gemma":[0.6226167,0.0015334153,0.37026954,0.00009912127,0.0000703103,0.00034264708,0.0027164877,0.00024685392,0.0021049604],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99805886,0.0006527809,0.0001839282,0.00017813031,0.00078200916,0.00014419683],"domain_scores_gemma":[0.994666,0.0023495737,0.00035944942,0.00061715784,0.001852829,0.000155038],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028650027,0.0010686981,0.0008589894,0.0020992714,0.00063849706,0.001226191,0.0006037511,0.0005784026,0.0014605228],"category_scores_gemma":[0.010983591,0.00017950521,0.0009304256,0.002532895,0.00035433887,0.0016070419,0.0007573406,0.00073006214,0.0006528203],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013277575,0.0009252831,0.014967389,0.0011977673,0.00046534764,0.00032074784,0.0008593044,0.20898645,0.02282314,0.0057216366,0.01596304,0.7264422],"study_design_scores_gemma":[0.00009551969,0.00072273327,0.019183515,0.00006785916,0.00009394754,0.00024861627,0.0006898724,0.9494934,0.020732371,0.0027000248,0.005892938,0.00007912564],"about_ca_topic_score_codex":0.005815988,"about_ca_topic_score_gemma":0.00461735,"teacher_disagreement_score":0.005815988,"about_ca_system_score_codex":0.0005141848,"about_ca_system_score_gemma":0.0009174362,"threshold_uncertainty_score":0.015151739},"labels":[],"label_agreement":null},{"id":"W3131740359","doi":"10.1016/j.datak.2021.101876","title":"Generation of Gaussian sets for clustering methods assessment","year":2021,"lang":"en","type":"article","venue":"Data & Knowledge Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thales (Canada)","funders":"","keywords":"Cluster analysis; Computer science; Generator (circuit theory); Fuzzy clustering; Curse of dimensionality; Rand index; Mixture model; Data mining; Homogeneity (statistics); Artificial intelligence; Gaussian; Pattern recognition (psychology); Set (abstract data type); Hierarchical clustering; Maximization; Sensitivity (control systems); Algorithm; Machine learning; Mathematics; Mathematical optimization; Power (physics); Engineering","score_opus":0.1936835263463012,"score_gpt":0.47320929907024467,"score_spread":0.27952577272394347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3131740359","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016625797,0.0001336606,0.9788477,0.00014305922,0.000080497215,0.0004186926,0.00037643992,0.0015848668,0.0017893169],"genre_scores_gemma":[0.17255045,0.00014815663,0.82220834,0.00010005046,0.000035749392,0.0007899234,0.0017777812,0.000709928,0.0016794789],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9928108,0.0027005489,0.00040443847,0.00087575923,0.0027693128,0.00043914656],"domain_scores_gemma":[0.9839677,0.0058430005,0.00062133174,0.0023440474,0.0068748957,0.00034909305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009924402,0.0015130701,0.0012830726,0.0052485643,0.0019547737,0.0029163668,0.002567967,0.0021444652,0.005694978],"category_scores_gemma":[0.040438406,0.0007541541,0.0024856238,0.0029356705,0.0010445595,0.0019335966,0.0039442056,0.002505636,0.0020064723],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000797116,0.00047841354,0.008731701,0.0006865494,0.00037454863,0.00026348693,0.0013956546,0.22840476,0.01591539,0.1112095,0.018864147,0.61287886],"study_design_scores_gemma":[0.00007129453,0.00018278747,0.0022630072,0.00012661524,0.00007467827,0.00013980227,0.00022280274,0.9240623,0.015146071,0.04927241,0.008371458,0.0000666709],"about_ca_topic_score_codex":0.005646474,"about_ca_topic_score_gemma":0.0058634263,"teacher_disagreement_score":0.009924402,"about_ca_system_score_codex":0.0021200378,"about_ca_system_score_gemma":0.0037608177,"threshold_uncertainty_score":0.052485883},"labels":[],"label_agreement":null},{"id":"W3139417388","doi":"10.1016/j.neucom.2021.02.062","title":"Automatic determining optimal parameters in multi-kernel collaborative fuzzy clustering based on dimension constraint","year":2021,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Kernel (algebra); Computer science; Kernel embedding of distributions; Kernel method; Kernel principal component analysis; Variable kernel density estimation; Fuzzy clustering; Dimensionality reduction; Constrained clustering; Constraint (computer-aided design); Dimension (graph theory); Artificial intelligence; Data mining; Mathematics; Pattern recognition (psychology); CURE data clustering algorithm; Support vector machine","score_opus":0.03790629680017404,"score_gpt":0.31660966168509636,"score_spread":0.2787033648849223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3139417388","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02697274,0.0001900979,0.97191656,0.00004913905,0.000016132713,0.000026384212,0.000030476347,0.00023425653,0.0005642963],"genre_scores_gemma":[0.6483606,0.00017643975,0.35014227,0.000039955226,0.000021843041,0.000123888,0.00018776288,0.00012445025,0.0008228126],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99840444,0.00043691928,0.00013038497,0.00044809558,0.00041464402,0.00016552457],"domain_scores_gemma":[0.997715,0.0009275871,0.0002034329,0.00030737257,0.0007605267,0.00008613718],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016183847,0.0008035536,0.0017033111,0.0010090037,0.001052363,0.0016500022,0.0018883413,0.0015317322,0.0009894685],"category_scores_gemma":[0.0068815774,0.00060058216,0.0010119693,0.0011458592,0.00084181526,0.0021852707,0.0015003027,0.0011024883,0.00040956686],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006322052,0.00016795956,0.0027604823,0.0003340499,0.0001786795,0.00015422863,0.00047220386,0.6158856,0.025999162,0.016975116,0.002904117,0.33353618],"study_design_scores_gemma":[0.000009332332,0.000021500862,0.00037039738,0.000009560916,0.000017849883,0.000034819124,0.000047717178,0.992413,0.0034148141,0.003372881,0.00026938372,0.000018744215],"about_ca_topic_score_codex":0.005112281,"about_ca_topic_score_gemma":0.0041091996,"teacher_disagreement_score":0.005112281,"about_ca_system_score_codex":0.0009001829,"about_ca_system_score_gemma":0.0015767583,"threshold_uncertainty_score":0.010165095},"labels":[],"label_agreement":null},{"id":"W3146117809","doi":"10.18280/isi.260106","title":"Parallelization of the K-Means++ Clustering Algorithm","year":2021,"lang":"en","type":"article","venue":"Ingénierie des systèmes d information","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Cluster analysis; Parallel computing; Initialization; SIMD; CUDA; Parallel algorithm; Graphics processing unit; Graphics; Algorithm; Throughput; Parallel programming model; Programming paradigm; Operating system","score_opus":0.01641005298293848,"score_gpt":0.2551106498400835,"score_spread":0.23870059685714504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3146117809","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032605965,0.00026196954,0.94016486,0.00026985985,0.00020140319,0.00018553884,0.00036151314,0.014822172,0.011126727],"genre_scores_gemma":[0.17969601,0.00020494431,0.81110543,0.00009099347,0.00005480759,0.0002486827,0.001157277,0.0011223175,0.006319446],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992205,0.00009302607,0.000059551905,0.00020601021,0.0003197938,0.00010106382],"domain_scores_gemma":[0.99939334,0.00009433484,0.000032055832,0.00015337662,0.00029122343,0.00003569058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048461088,0.0008431753,0.00075775816,0.00054932054,0.0007972992,0.0010134358,0.001307543,0.0004594373,0.004409324],"category_scores_gemma":[0.0017421742,0.00037911048,0.0006718144,0.0008654741,0.00039001153,0.0007385795,0.00097586546,0.0009307106,0.0023936443],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006601055,0.00023339437,0.0034524784,0.00035104758,0.00013972195,0.00034000972,0.0003713554,0.31396332,0.061778076,0.02313001,0.02972249,0.56585807],"study_design_scores_gemma":[0.000079191785,0.000061435814,0.00092415995,0.000011824205,0.000023356553,0.00016760026,0.00006304442,0.93839943,0.028678795,0.008014729,0.02354668,0.000029797291],"about_ca_topic_score_codex":0.010390407,"about_ca_topic_score_gemma":0.009890884,"teacher_disagreement_score":0.010390407,"about_ca_system_score_codex":0.00080094516,"about_ca_system_score_gemma":0.0018495875,"threshold_uncertainty_score":0.020659864},"labels":[],"label_agreement":null},{"id":"W3174452951","doi":"10.1109/access.2021.3091397","title":"Automatic Data Clustering Framework Using Nature-Inspired Binary Optimization Algorithms","year":2021,"lang":"en","type":"article","venue":"IEEE Access","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Computer science; Distortion (music); Correlation clustering; CURE data clustering algorithm; Algorithm; Binary number; Canopy clustering algorithm; Data mining; Artificial intelligence; Mathematics","score_opus":0.10296852308607309,"score_gpt":0.4151459321661608,"score_spread":0.31217740908008773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174452951","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023340248,0.00008256997,0.9968213,0.000060396345,0.0000122992815,0.00002564383,0.000018416666,0.00017802215,0.00046739494],"genre_scores_gemma":[0.1446775,0.00021830991,0.85277903,0.00009667491,0.00004500571,0.00034446857,0.00022638527,0.00012160955,0.0014910746],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99847144,0.00043099068,0.00008755956,0.00030986007,0.0005917795,0.00010833942],"domain_scores_gemma":[0.9989869,0.0003756009,0.00015873979,0.0001157398,0.000320119,0.00004284436],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002187018,0.0008883939,0.0013713541,0.0022195545,0.0008896976,0.0018353027,0.0023991216,0.001376939,0.0012084778],"category_scores_gemma":[0.0033135153,0.00067796756,0.0014760182,0.0022746443,0.0009941113,0.0015611702,0.0018162403,0.0014869948,0.0003885931],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006940129,0.00006131457,0.00063149066,0.00009971348,0.00007227537,0.000043578875,0.00011726182,0.84657955,0.003329738,0.037496384,0.0014719737,0.11002737],"study_design_scores_gemma":[0.000006275824,0.000008489886,0.00006454907,0.000004165644,0.0000034827472,0.000010479991,0.000007012367,0.9944734,0.00053939555,0.0042890585,0.0005884461,0.000005207944],"about_ca_topic_score_codex":0.0048285625,"about_ca_topic_score_gemma":0.0030803164,"teacher_disagreement_score":0.0048285625,"about_ca_system_score_codex":0.0016170864,"about_ca_system_score_gemma":0.0018211624,"threshold_uncertainty_score":0.011732876},"labels":[],"label_agreement":null},{"id":"W3182227316","doi":"10.18280/ria.350302","title":"Minimization of the Number of Iterations in K-Medoids Clustering with Purity Algorithm","year":2021,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; k-medoids; Minification; Algorithm; Medoid; k-means clustering; Mathematics; Computer science; Mathematical optimization; Correlation clustering; CURE data clustering algorithm; Statistics","score_opus":0.028706898763195963,"score_gpt":0.3027406127607996,"score_spread":0.2740337139976036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3182227316","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008167668,0.00024891185,0.99039966,0.00009979179,0.0000311583,0.000073839205,0.000027365619,0.0003276068,0.0006239087],"genre_scores_gemma":[0.1420259,0.00031689618,0.85550034,0.00011998373,0.00004499012,0.00037275525,0.0002482924,0.0002636066,0.0011071248],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99255013,0.0035371059,0.00071516295,0.0010842392,0.0017966056,0.000316746],"domain_scores_gemma":[0.98702776,0.007907729,0.0009540938,0.0008903527,0.00291445,0.00030560608],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008226012,0.0016896743,0.0020172978,0.0019233392,0.001951497,0.002073567,0.0023043172,0.0022152448,0.0013810562],"category_scores_gemma":[0.026851017,0.0011760069,0.0015469452,0.0018066319,0.0019026073,0.0028765232,0.00258168,0.0022656259,0.0010399895],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008277597,0.0001813432,0.004188348,0.0007399458,0.0003071984,0.00016798908,0.0010473212,0.75756985,0.010546244,0.023051037,0.0032946863,0.1980783],"study_design_scores_gemma":[0.000041709092,0.00014867916,0.00054473075,0.00004647026,0.000045377892,0.000147075,0.00016132998,0.9789725,0.008707969,0.008875032,0.002265808,0.000043273587],"about_ca_topic_score_codex":0.0036954468,"about_ca_topic_score_gemma":0.0031749923,"teacher_disagreement_score":0.008226012,"about_ca_system_score_codex":0.0012725367,"about_ca_system_score_gemma":0.00304634,"threshold_uncertainty_score":0.04350382},"labels":[],"label_agreement":null},{"id":"W3184306616","doi":"10.3390/math9141680","title":"Self-Expressive Kernel Subspace Clustering Algorithm for Categorical Data with Embedded Feature Selection","year":2021,"lang":"en","type":"article","venue":"Mathematics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Kernel (algebra); Categorical variable; Computer science; Pattern recognition (psychology); Artificial intelligence; Variable kernel density estimation; Mathematics; Kernel method; Data mining; Algorithm; Machine learning; Support vector machine","score_opus":0.03362400087073737,"score_gpt":0.31652423611581404,"score_spread":0.2829002352450767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3184306616","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006027661,0.00010436562,0.9932007,0.000046210935,0.000013865899,0.000028251787,0.000030418709,0.0002908727,0.000257637],"genre_scores_gemma":[0.256027,0.00024588857,0.7405128,0.000104457315,0.0000489225,0.00024316694,0.0006732559,0.00013376356,0.0020107338],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981945,0.0005687556,0.00013106722,0.00037736364,0.0006117653,0.000116596166],"domain_scores_gemma":[0.99816114,0.00045752275,0.00016277515,0.00036494745,0.00076849037,0.00008514793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015839541,0.0006627,0.0012445144,0.0015478086,0.0008214246,0.00093034527,0.0016942477,0.00086288684,0.0012470079],"category_scores_gemma":[0.0046740607,0.0003117111,0.0010877355,0.0021819212,0.0006535131,0.0017928581,0.0017379805,0.0011685792,0.0007839018],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002219503,0.00018448979,0.0031600373,0.00017019514,0.000184424,0.0000970432,0.0003712398,0.29455245,0.013539375,0.034798782,0.005165292,0.64755476],"study_design_scores_gemma":[0.000008417149,0.000030496163,0.00037441816,0.0000051879647,0.000008471758,0.000050683735,0.000036028585,0.9884604,0.001993972,0.007771896,0.0012419098,0.000018163051],"about_ca_topic_score_codex":0.003278187,"about_ca_topic_score_gemma":0.0030811015,"teacher_disagreement_score":0.003278187,"about_ca_system_score_codex":0.00065528904,"about_ca_system_score_gemma":0.0013763044,"threshold_uncertainty_score":0.008376837},"labels":[],"label_agreement":null},{"id":"W3185688577","doi":"","title":"CANUCS: The CAnadian NIRISS Unbiased Cluster Survey","year":2017,"lang":"en","type":"article","venue":"JWST Proposal. Cycle 1","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cluster (spacecraft); Computer science","score_opus":0.03733052967814782,"score_gpt":0.32608347648642094,"score_spread":0.2887529468082731,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3185688577","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021055752,0.0052384227,0.17428835,0.024439977,0.0044550165,0.0036353036,0.5353786,0.018506613,0.21300201],"genre_scores_gemma":[0.10130343,0.0039372547,0.18238291,0.008333412,0.0015741818,0.004423329,0.5187561,0.009624153,0.16966517],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.98692757,0.0025028863,0.0002642223,0.0013118716,0.0067681097,0.0022252814],"domain_scores_gemma":[0.9597756,0.0030198845,0.00085305644,0.0062057003,0.026386825,0.0037587928],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0219465,0.0017428567,0.0017719225,0.008221217,0.0075099804,0.0062074075,0.006018228,0.0017051088,0.032885678],"category_scores_gemma":[0.052704345,0.0010233008,0.0010949585,0.013613003,0.0018402376,0.0020573451,0.006003507,0.002671659,0.016930446],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018857923,0.000041479914,0.0073009413,0.00017019094,0.00008955911,0.000025618036,0.00019909389,0.0019806037,0.00033204805,0.014930562,0.9140272,0.060714208],"study_design_scores_gemma":[0.00017711314,0.000048735663,0.03150373,0.00030297466,0.00009465945,0.000058487145,0.00063813623,0.008853509,0.0014732603,0.01340516,0.9433063,0.00013800296],"about_ca_topic_score_codex":0.9255908,"about_ca_topic_score_gemma":0.95180625,"teacher_disagreement_score":0.07440919,"about_ca_system_score_codex":0.026815373,"about_ca_system_score_gemma":0.11590603,"threshold_uncertainty_score":0.19456005},"labels":[],"label_agreement":null},{"id":"W3189005758","doi":"10.6084/m9.figshare.c.5127031.v1","title":"Supplementary material from \"Multivariate phenotypic divergence along an urbanization gradient\"","year":2020,"lang":"en","type":"article","venue":"Figshare","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multivariate statistics; Divergence (linguistics); Multivariate analysis; Urbanization; Geography; Mathematics; Biology; Statistics; Ecology","score_opus":0.05498864000589632,"score_gpt":0.28979764688820386,"score_spread":0.23480900688230755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3189005758","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008346478,0.00006238034,0.0036446617,0.00021602598,0.0003934949,0.000080517035,0.9850856,0.0024855796,0.007196991],"genre_scores_gemma":[0.008689896,0.00024392863,0.011666648,0.00043928454,0.00018067572,0.0006000538,0.9535219,0.0033911904,0.021266356],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99900985,0.00014691593,0.00009481264,0.0002486062,0.00032378593,0.00017601317],"domain_scores_gemma":[0.9947083,0.0025097365,0.00024691943,0.00079031166,0.001312308,0.00043249645],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0013709129,0.0016273886,0.0013702433,0.0030405023,0.0014442435,0.0021939843,0.0024500217,0.0012617236,0.81229067],"category_scores_gemma":[0.012322136,0.00092627277,0.001561846,0.004637281,0.00036810065,0.001408643,0.0024496396,0.00094628416,0.31543955],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001933694,0.000050454633,0.0019313626,0.0007581516,0.00007407076,0.000089393194,0.00008721321,0.00047586655,0.0007093035,0.0012224044,0.977368,0.017040273],"study_design_scores_gemma":[0.00058536604,0.00010848349,0.026854174,0.00064922555,0.0001013591,0.0004104676,0.00017311178,0.002268343,0.001187079,0.008391034,0.959153,0.00011834923],"about_ca_topic_score_codex":0.027423106,"about_ca_topic_score_gemma":0.053031236,"teacher_disagreement_score":0.81229067,"about_ca_system_score_codex":0.0012097959,"about_ca_system_score_gemma":0.0024795742,"threshold_uncertainty_score":0.26774448},"labels":[],"label_agreement":null},{"id":"W3198815918","doi":"10.3390/a14090258","title":"A Novel Semi-Supervised Fuzzy C-Means Clustering Algorithm Using Multiple Fuzzification Coefficients","year":2021,"lang":"en","type":"article","venue":"Algorithms","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Foundation for Science and Technology Development","keywords":"Cluster analysis; Fuzzy clustering; Computer science; Artificial intelligence; Data mining; Pattern recognition (psychology); Fuzzy logic; Fuzzy set; Correlation clustering; Mathematics; Algorithm; Machine learning","score_opus":0.05065401274004585,"score_gpt":0.3084842384201754,"score_spread":0.2578302256801296,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3198815918","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062949345,0.00018261593,0.99192786,0.00007375233,0.000037341517,0.000051440114,0.000026742855,0.00033522383,0.0010701442],"genre_scores_gemma":[0.15322904,0.0002225595,0.84351873,0.00010228968,0.000043142412,0.00021340887,0.00017616924,0.00006539769,0.0024293126],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991172,0.00011336778,0.00004735382,0.00023791238,0.00042502276,0.000059153343],"domain_scores_gemma":[0.99925715,0.00016431013,0.00006872227,0.000058861442,0.00042071092,0.0000302115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008635774,0.00083879783,0.0013168331,0.0011803753,0.0010735799,0.0010647334,0.0019906682,0.0015676774,0.0012279167],"category_scores_gemma":[0.0017767857,0.00044273198,0.0010392616,0.0012920826,0.00066661683,0.0011082973,0.0007889182,0.0011778595,0.00055724365],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016932075,0.00010733796,0.0015480809,0.00023057,0.00015911645,0.00013726039,0.0003189382,0.44078618,0.023646874,0.017895967,0.005736105,0.50926423],"study_design_scores_gemma":[0.000010570287,0.000018373741,0.00020257477,0.000008732175,0.0000092193395,0.000057478726,0.00001155154,0.9940295,0.0024283521,0.0015457412,0.001662349,0.000015615853],"about_ca_topic_score_codex":0.012278357,"about_ca_topic_score_gemma":0.011756367,"teacher_disagreement_score":0.012278357,"about_ca_system_score_codex":0.0011155572,"about_ca_system_score_gemma":0.0020479283,"threshold_uncertainty_score":0.024413764},"labels":[],"label_agreement":null},{"id":"W3199460270","doi":"10.1007/s10618-021-00794-0","title":"A Lagrangian-based score for assessing the quality of pairwise constraints in semi-supervised clustering","year":2021,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal; Group for Research in Decision Analysis","funders":"","keywords":"Cluster analysis; Pairwise comparison; Constrained clustering; Constraint (computer-aided design); Computer science; Data mining; Machine learning; Key (lock); Artificial intelligence; Correlation clustering; Mathematics; CURE data clustering algorithm","score_opus":0.1432772631699324,"score_gpt":0.40712493858682247,"score_spread":0.2638476754168901,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3199460270","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044995077,0.00038777618,0.9511561,0.0004508108,0.00008828169,0.00016918727,0.0004561699,0.0005238377,0.0017727531],"genre_scores_gemma":[0.5308143,0.00022643458,0.46474674,0.0002531692,0.00017880024,0.000344751,0.0018473099,0.00023144318,0.0013569709],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.986406,0.0051641795,0.0011296603,0.001355385,0.005344044,0.0006007172],"domain_scores_gemma":[0.9651881,0.019711105,0.0038190107,0.002310764,0.0074116746,0.0015593158],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01602347,0.001441052,0.0027428286,0.0061821057,0.0017957407,0.0037484465,0.003964273,0.0037040205,0.0026334017],"category_scores_gemma":[0.052933082,0.0006531999,0.0015522134,0.0045232135,0.003716595,0.004442091,0.003937851,0.0025324335,0.00071301416],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014164159,0.0006294053,0.031325255,0.00108296,0.001120738,0.00045871004,0.0005473424,0.60686344,0.011698918,0.06800734,0.012573553,0.2642759],"study_design_scores_gemma":[0.0000809103,0.0002673114,0.002241407,0.00005343985,0.00006120275,0.00016391127,0.00009352922,0.97109884,0.0019517784,0.023320125,0.00060281454,0.00006478135],"about_ca_topic_score_codex":0.0035882178,"about_ca_topic_score_gemma":0.0043167556,"teacher_disagreement_score":0.01602347,"about_ca_system_score_codex":0.002280107,"about_ca_system_score_gemma":0.0036066724,"threshold_uncertainty_score":0.084741235},"labels":[],"label_agreement":null},{"id":"W3205343175","doi":"10.1093/nsr/nwab183","title":"Massive data clustering by multi-scale psychological observations","year":2021,"lang":"en","type":"article","venue":"National Science Review","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"National Key Research and Development Program of China; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; National Science Foundation","keywords":"Cluster analysis; Scale (ratio); Psychology; Computer science; Artificial intelligence; Geography; Cartography","score_opus":0.30351777484378367,"score_gpt":0.4854582473962416,"score_spread":0.18194047255245793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3205343175","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.095854096,0.00031656388,0.90038145,0.0006269286,0.00007752391,0.00015960576,0.00023457683,0.0007416424,0.0016077305],"genre_scores_gemma":[0.66257316,0.00027367304,0.33500025,0.00021472218,0.00006924977,0.00023879475,0.00049758825,0.00008477252,0.0010477721],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99889475,0.0004330026,0.00004461849,0.00032491845,0.00023148418,0.000071242364],"domain_scores_gemma":[0.99593395,0.002299239,0.00043696852,0.00074709003,0.00042122457,0.00016150746],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002332397,0.00056049356,0.000754931,0.0011835097,0.0009801096,0.0013412634,0.0021511877,0.0010078609,0.0009005711],"category_scores_gemma":[0.008981425,0.00044315614,0.0014256625,0.0012828202,0.0016111587,0.002026927,0.001881549,0.0015380137,0.00027762173],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014312536,0.000097285476,0.011139415,0.00019212403,0.00018538145,0.00015739452,0.00068390125,0.8513283,0.0031984765,0.05130517,0.0030066126,0.07856276],"study_design_scores_gemma":[0.000011162123,0.000013673028,0.00094769575,0.0000074914865,0.000009200987,0.000022786415,0.000051770046,0.97780186,0.00044632764,0.020059703,0.0006171416,0.00001118738],"about_ca_topic_score_codex":0.010082479,"about_ca_topic_score_gemma":0.010263108,"teacher_disagreement_score":0.010082479,"about_ca_system_score_codex":0.001594511,"about_ca_system_score_gemma":0.0013393334,"threshold_uncertainty_score":0.020047605},"labels":[],"label_agreement":null},{"id":"W3211395167","doi":"10.1109/access.2021.3126854","title":"KNN-SC: Novel Spectral Clustering Algorithm Using k-Nearest Neighbors","year":2021,"lang":"en","type":"article","venue":"IEEE Access","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"Institute for Information and Communications Technology Promotion; National Research Foundation of Korea","keywords":"Cluster analysis; Spectral clustering; CURE data clustering algorithm; Computer science; Correlation clustering; k-nearest neighbors algorithm; Pattern recognition (psychology); Nearest-neighbor chain algorithm; Canopy clustering algorithm; Graph; Algorithm; Noise (video); Artificial intelligence; Data mining; Theoretical computer science","score_opus":0.08371092394036551,"score_gpt":0.3714442137006123,"score_spread":0.2877332897602468,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3211395167","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007027723,0.00035317647,0.989153,0.00012555529,0.000074715885,0.00007081639,0.00013801352,0.0017244413,0.0013325614],"genre_scores_gemma":[0.13297388,0.00033961225,0.8616398,0.00023199199,0.00008288252,0.00021253302,0.0011984908,0.00039912682,0.0029216106],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99781406,0.00041585136,0.000111242865,0.0005606893,0.000963101,0.00013504694],"domain_scores_gemma":[0.9983321,0.00036259482,0.00019066413,0.00031238288,0.000714374,0.000087910106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012154435,0.0012941073,0.001670026,0.003941369,0.0019146148,0.0011714079,0.0029889643,0.0017436551,0.0017489034],"category_scores_gemma":[0.004392835,0.0005532122,0.0010913886,0.0051638763,0.0012242147,0.0026569094,0.0019125383,0.0012768764,0.0019120568],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003403889,0.0002443297,0.0026642624,0.00031112076,0.00022226562,0.0001873057,0.000440023,0.28198647,0.010993428,0.017035875,0.021990666,0.6635839],"study_design_scores_gemma":[0.000025259387,0.00003589631,0.00033519958,0.0000140246975,0.000014842743,0.00016099887,0.00007794847,0.97986424,0.0034141175,0.0113367,0.004684584,0.000036209738],"about_ca_topic_score_codex":0.012142925,"about_ca_topic_score_gemma":0.014132043,"teacher_disagreement_score":0.012142925,"about_ca_system_score_codex":0.0011115712,"about_ca_system_score_gemma":0.0021859517,"threshold_uncertainty_score":0.02414447},"labels":[],"label_agreement":null},{"id":"W4205263674","doi":"10.1007/978-0-387-39940-9_3579","title":"Semi-Supervised Clustering","year":2009,"lang":"en","type":"book-chapter","venue":"Encyclopedia of Database Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.024568357284607774,"score_gpt":0.2700766950451616,"score_spread":0.2455083377605538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205263674","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016541487,0.00076270977,0.98461366,0.00015411018,0.00015722617,0.00011165158,0.0006595578,0.0036302155,0.008256731],"genre_scores_gemma":[0.03557984,0.0009000828,0.9255518,0.00022575325,0.00019297254,0.0002828746,0.006681727,0.0013646602,0.029220348],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9978143,0.00043773424,0.00014526157,0.0007406848,0.0007791239,0.00008295158],"domain_scores_gemma":[0.997855,0.00046481294,0.000083238585,0.00081275677,0.0007313608,0.000052758256],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015456253,0.0014078103,0.0018296968,0.0021498206,0.0011155349,0.00216099,0.002666356,0.0014134881,0.013909565],"category_scores_gemma":[0.0032791998,0.00090410945,0.0017078285,0.002650038,0.00078438374,0.0019814256,0.0020543262,0.0015255676,0.021744404],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000089359506,0.000081362676,0.0004043366,0.00032964913,0.00015796712,0.000066772,0.00013682197,0.028389072,0.008721656,0.014835896,0.069534115,0.87725306],"study_design_scores_gemma":[0.000028231527,0.00010866619,0.0020811225,0.00014658026,0.000102162194,0.0009444792,0.00018929463,0.7450705,0.028054483,0.0920121,0.13114987,0.00011253521],"about_ca_topic_score_codex":0.0016143873,"about_ca_topic_score_gemma":0.004150398,"teacher_disagreement_score":0.013909565,"about_ca_system_score_codex":0.000559187,"about_ca_system_score_gemma":0.001308139,"threshold_uncertainty_score":0.046532154},"labels":[],"label_agreement":null},{"id":"W4205661532","doi":"10.1109/ijcnn.2006.1716671","title":"Refining Spherical K-Means for Clustering Documents","year":2006,"lang":"en","type":"article","venue":"The 2006 IEEE International Joint Conference on Neural Network Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Convexity; Cluster analysis; Mathematical optimization; Constraint (computer-aided design); Set (abstract data type); Maximization; Mathematics; Algorithm; Computer science; Function (biology); Linear programming; Optimization problem; Artificial intelligence","score_opus":0.06047978020347221,"score_gpt":0.318466311927053,"score_spread":0.25798653172358077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205661532","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019819592,0.00010432902,0.9973972,0.00004324934,0.000011076778,0.000017760602,0.00002320547,0.00013370623,0.00028756456],"genre_scores_gemma":[0.094304666,0.00052316196,0.9025238,0.00010416812,0.000058639685,0.00015125616,0.0004487855,0.00030817522,0.0015773283],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969574,0.0011924895,0.00020683277,0.00054873974,0.000927604,0.00016700382],"domain_scores_gemma":[0.99748516,0.0008899088,0.00021510896,0.0005642634,0.0007701526,0.00007539025],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002851189,0.0012456955,0.0014947727,0.0016745955,0.0008606969,0.0016585862,0.002027016,0.0015871335,0.0015007606],"category_scores_gemma":[0.0102959415,0.0007628833,0.0017837188,0.0029604083,0.0013882648,0.0026056725,0.0019917144,0.0019572973,0.0018351497],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015909108,0.000051445742,0.0007253377,0.0002628831,0.00010419677,0.00006571626,0.00025445313,0.7637171,0.012631699,0.06372601,0.003975192,0.1543268],"study_design_scores_gemma":[0.000009548727,0.00002588807,0.00011000232,0.000007828073,0.000009864763,0.00003468385,0.000027451371,0.9792184,0.0031465485,0.015086467,0.0023016022,0.000021682257],"about_ca_topic_score_codex":0.009750403,"about_ca_topic_score_gemma":0.010162916,"teacher_disagreement_score":0.009750403,"about_ca_system_score_codex":0.0013803225,"about_ca_system_score_gemma":0.001872683,"threshold_uncertainty_score":0.019387305},"labels":[],"label_agreement":null},{"id":"W4206148502","doi":"10.36227/techrxiv.16926754","title":"Unified Embedding and Clustering","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Cluster analysis; Embedding; Computer science; Correlation clustering; Manifold (fluid mechanics); Clustering high-dimensional data; CURE data clustering algorithm; Space (punctuation); Artificial intelligence; Mathematics; Data mining; Algorithm; Pattern recognition (psychology); Engineering","score_opus":0.0380694894070643,"score_gpt":0.3374502951707269,"score_spread":0.2993808057636626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206148502","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027812144,0.00075488596,0.9941062,0.00018555208,0.00007387223,0.00003916147,0.00012053246,0.00044589932,0.001492677],"genre_scores_gemma":[0.12627898,0.0015714831,0.86349905,0.00026853633,0.00029142393,0.00023103558,0.0017837875,0.00061905914,0.005456596],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99636036,0.0009765285,0.0001817171,0.001237589,0.0010320661,0.00021171299],"domain_scores_gemma":[0.9975535,0.0006343505,0.00023131429,0.0008711887,0.00060335867,0.00010622881],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024882606,0.0019035917,0.0021316297,0.0036477195,0.0013502693,0.0031598343,0.0028922241,0.002518556,0.003467336],"category_scores_gemma":[0.00853218,0.0008090702,0.0015161262,0.0047294777,0.002149117,0.0056089405,0.004454848,0.002673395,0.00252731],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000163343,0.00009389226,0.0014334389,0.00048217905,0.00025984706,0.00012906406,0.00042711364,0.23697573,0.006040652,0.28097063,0.01176025,0.4612639],"study_design_scores_gemma":[0.000017759863,0.00007363506,0.00063680176,0.00007055852,0.00004251007,0.00021377986,0.00012801342,0.72778934,0.003389499,0.24880269,0.018781722,0.000053567885],"about_ca_topic_score_codex":0.0022645127,"about_ca_topic_score_gemma":0.002266,"teacher_disagreement_score":0.0036477195,"about_ca_system_score_codex":0.001439967,"about_ca_system_score_gemma":0.0014147689,"threshold_uncertainty_score":0.013159335},"labels":[],"label_agreement":null},{"id":"W4210265465","doi":"10.36227/techrxiv.16926754.v1","title":"Unified Embedding and Clustering","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Cluster analysis; Embedding; Computer science; Correlation clustering; Clustering high-dimensional data; CURE data clustering algorithm; Manifold (fluid mechanics); Artificial intelligence; Space (punctuation); Data mining; Mathematics; Algorithm; Engineering","score_opus":0.0380694894070643,"score_gpt":0.3374502951707269,"score_spread":0.2993808057636626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210265465","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027812144,0.00075488596,0.9941062,0.00018555208,0.00007387223,0.00003916147,0.00012053246,0.00044589932,0.001492677],"genre_scores_gemma":[0.12627898,0.0015714831,0.86349905,0.00026853633,0.00029142393,0.00023103558,0.0017837875,0.00061905914,0.005456596],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99636036,0.0009765285,0.0001817171,0.001237589,0.0010320661,0.00021171299],"domain_scores_gemma":[0.9975535,0.0006343505,0.00023131429,0.0008711887,0.00060335867,0.00010622881],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024882606,0.0019035917,0.0021316297,0.0036477195,0.0013502693,0.0031598343,0.0028922241,0.002518556,0.003467336],"category_scores_gemma":[0.00853218,0.0008090702,0.0015161262,0.0047294777,0.002149117,0.0056089405,0.004454848,0.002673395,0.00252731],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000163343,0.00009389226,0.0014334389,0.00048217905,0.00025984706,0.00012906406,0.00042711364,0.23697573,0.006040652,0.28097063,0.01176025,0.4612639],"study_design_scores_gemma":[0.000017759863,0.00007363506,0.00063680176,0.00007055852,0.00004251007,0.00021377986,0.00012801342,0.72778934,0.003389499,0.24880269,0.018781722,0.000053567885],"about_ca_topic_score_codex":0.0022645127,"about_ca_topic_score_gemma":0.002266,"teacher_disagreement_score":0.0036477195,"about_ca_system_score_codex":0.001439967,"about_ca_system_score_gemma":0.0014147689,"threshold_uncertainty_score":0.013159335},"labels":[],"label_agreement":null},{"id":"W4210362242","doi":"10.1287/ijoc.2021.1137","title":"Efficient Algebraic Multigrid Methods for Multilevel Overlapping Coclustering of User-Item Relationships","year":2022,"lang":"en","type":"article","venue":"INFORMS journal on computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Toronto Metropolitan University","funders":"","keywords":"Bipartite graph; Computer science; Theoretical computer science; Scalability; Feature (linguistics); Graph; Algorithm","score_opus":0.07169727211548858,"score_gpt":0.38226519267080344,"score_spread":0.31056792055531485,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210362242","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063323565,0.000083983905,0.99249953,0.00008878373,0.000018764147,0.00003325928,0.00006663897,0.00026273256,0.0006139272],"genre_scores_gemma":[0.1568821,0.00017908387,0.84013426,0.000118556345,0.00003229847,0.00030413948,0.00054816285,0.0002618639,0.0015395938],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995227,0.00015863388,0.000030487201,0.000088156,0.00015204313,0.000048025806],"domain_scores_gemma":[0.9970835,0.0019576177,0.00018574165,0.00028013618,0.00040032557,0.00009260589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009786768,0.00076839764,0.0009484039,0.001065051,0.0005385962,0.0010343539,0.0013951136,0.0009245507,0.0023024005],"category_scores_gemma":[0.0063206353,0.00046989138,0.0011660991,0.0012330073,0.00086394366,0.0010089119,0.001506736,0.0015689767,0.0006740206],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008965075,0.0000594617,0.0014910022,0.00019166007,0.000076034965,0.00010838404,0.0002809239,0.8574923,0.0042047603,0.04461925,0.0031977089,0.08818886],"study_design_scores_gemma":[0.000004910946,0.0000048082134,0.00006918504,0.00000295593,0.0000018546932,0.0000064816822,0.000019257172,0.9917515,0.00021710772,0.00748272,0.00043588397,0.0000033565295],"about_ca_topic_score_codex":0.008329648,"about_ca_topic_score_gemma":0.012023103,"teacher_disagreement_score":0.008329648,"about_ca_system_score_codex":0.00087682804,"about_ca_system_score_gemma":0.0011580064,"threshold_uncertainty_score":0.016562343},"labels":[],"label_agreement":null},{"id":"W4211047464","doi":"10.1007/978-1-4471-7452-3_9","title":"Clustering I: Basic Clustering Models and Algorithms","year":2019,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Correlation clustering; Fuzzy clustering; Learning vector quantization; Artificial intelligence; Computer science; CURE data clustering algorithm; Pattern recognition (psychology); Canopy clustering algorithm; Conceptual clustering; Competitive learning; Unsupervised learning; Machine learning; Determining the number of clusters in a data set; Single-linkage clustering; Consensus clustering; Artificial neural network; Data mining","score_opus":0.05182116940461246,"score_gpt":0.2868308259429649,"score_spread":0.23500965653835243,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211047464","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011727418,0.07966339,0.78209156,0.0032442657,0.0050900932,0.00023284696,0.002571676,0.0040559825,0.121877424],"genre_scores_gemma":[0.01650767,0.08029868,0.65365773,0.0025868753,0.0058535296,0.00069682533,0.0076541686,0.0034844424,0.22926007],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99875486,0.00016281559,0.00006431102,0.00033202217,0.0006264646,0.00005951632],"domain_scores_gemma":[0.9989605,0.00034437416,0.000043140557,0.00022298418,0.00037943624,0.000049578295],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010913813,0.0019377862,0.0018539345,0.0029679392,0.001144795,0.004455166,0.00287886,0.0020205162,0.026326291],"category_scores_gemma":[0.0030284857,0.001161441,0.0013451063,0.008694341,0.0013579743,0.0048325276,0.0016519293,0.002845743,0.032836806],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026946176,0.000043146767,0.00016477688,0.0009483512,0.000050384402,0.00003365219,0.00015403959,0.0088521335,0.0015445036,0.15667306,0.37033498,0.46117404],"study_design_scores_gemma":[0.000007941361,0.000031717827,0.0004770134,0.0003746111,0.000029154477,0.00029885516,0.00010151941,0.021004662,0.0014652887,0.23855495,0.7375968,0.00005752884],"about_ca_topic_score_codex":0.0024301521,"about_ca_topic_score_gemma":0.0031143983,"teacher_disagreement_score":0.026326291,"about_ca_system_score_codex":0.002060346,"about_ca_system_score_gemma":0.0017016619,"threshold_uncertainty_score":0.088070214},"labels":[],"label_agreement":null},{"id":"W4211218449","doi":"10.1007/978-1-4471-7452-3_10","title":"Clustering II: Topics in Clustering","year":2019,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Computer science; Correlation clustering; CURE data clustering algorithm; Single-linkage clustering; Fuzzy clustering; Consensus clustering; Hierarchical clustering; Data mining; Brown clustering; Artificial intelligence","score_opus":0.03726779996790174,"score_gpt":0.2929327815631323,"score_spread":0.25566498159523054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211218449","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014649512,0.18646586,0.4111576,0.0061466857,0.015152672,0.00026893977,0.0021215512,0.003975772,0.37324592],"genre_scores_gemma":[0.014556891,0.12162307,0.21997271,0.0032952458,0.013495104,0.000477262,0.004963449,0.004214975,0.6174013],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99898976,0.00012866435,0.00004468201,0.00028794518,0.00049158535,0.00005735784],"domain_scores_gemma":[0.9990483,0.00033856073,0.000038730665,0.00015906992,0.00032142934,0.00009397314],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007560321,0.0017289551,0.0019189187,0.0036858455,0.0012961976,0.0053037587,0.0018956601,0.0016206561,0.050078906],"category_scores_gemma":[0.0022072596,0.0009884137,0.0012169401,0.010058782,0.0012775523,0.005751476,0.0018337265,0.0031971436,0.044864655],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025023404,0.000040808798,0.00010726533,0.0008621992,0.00002820692,0.000030330517,0.00017851107,0.0029954738,0.0011115585,0.09956348,0.47888294,0.41617417],"study_design_scores_gemma":[0.0000062025924,0.000024237695,0.00044052614,0.00038515308,0.000017376633,0.0002370516,0.00009699795,0.004608167,0.00084630353,0.10846532,0.8848433,0.000029367691],"about_ca_topic_score_codex":0.0017678872,"about_ca_topic_score_gemma":0.0025603618,"teacher_disagreement_score":0.050078906,"about_ca_system_score_codex":0.0021692766,"about_ca_system_score_gemma":0.001709626,"threshold_uncertainty_score":0.1675306},"labels":[],"label_agreement":null},{"id":"W4212923685","doi":"10.2139/ssrn.4038110","title":"A Process Convolution Model for Crash Count Data on a Network","year":2022,"lang":"en","type":"article","venue":"SSRN Electronic Journal","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; HEC Montréal","funders":"","keywords":"Count data; Crash; Computer science; Process (computing); Statistics; Convolution (computer science); Econometrics; Mathematics; Artificial intelligence; Programming language; Artificial neural network","score_opus":0.04158118083929049,"score_gpt":0.3369409990319969,"score_spread":0.29535981819270646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4212923685","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08071691,0.0005079246,0.91411495,0.0010374,0.00013701592,0.00008867867,0.0009231431,0.000877072,0.001596947],"genre_scores_gemma":[0.90863603,0.0010295524,0.07368839,0.00031717564,0.00035450645,0.0003286469,0.0018893825,0.00019842305,0.013557816],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998494,0.00038784175,0.00009364653,0.000446068,0.00029015334,0.0002881321],"domain_scores_gemma":[0.99316275,0.0040829806,0.0005799492,0.00085417787,0.0010344915,0.0002856474],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004950778,0.0010790203,0.001685464,0.0018377326,0.00082051917,0.0019599434,0.0035170107,0.0030845774,0.0030659065],"category_scores_gemma":[0.013014119,0.0010204031,0.0016283744,0.002632569,0.0012853465,0.0039321133,0.0017193469,0.0025901587,0.0011685553],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029141162,0.0001679193,0.007980663,0.000115252515,0.000117850366,0.00024864718,0.00022790702,0.8634437,0.0027865998,0.08037441,0.0033139286,0.040931795],"study_design_scores_gemma":[0.0000035629846,0.000013275023,0.00035811384,0.000004725435,0.00001099725,0.000025503396,0.0000064699084,0.9928772,0.00017915899,0.00629751,0.00021566146,0.000007765193],"about_ca_topic_score_codex":0.02197016,"about_ca_topic_score_gemma":0.0150879035,"teacher_disagreement_score":0.02197016,"about_ca_system_score_codex":0.0022428788,"about_ca_system_score_gemma":0.0018020638,"threshold_uncertainty_score":0.043684542},"labels":[],"label_agreement":null},{"id":"W4220790596","doi":"10.1109/bigcomp54360.2022.00017","title":"The Border K-Means Clustering Algorithm for One Dimensional Data","year":2022,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Cluster analysis; Computer science; Algorithm; Canopy clustering algorithm; CURE data clustering algorithm; Focus (optics); Data mining; Flexibility (engineering); Data stream clustering; Centroid; Cluster (spacecraft); Algorithm design; Correlation clustering; Artificial intelligence; Mathematics","score_opus":0.0641923449340145,"score_gpt":0.3551603780778739,"score_spread":0.2909680331438594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220790596","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015524966,0.0006281033,0.9950596,0.00011034006,0.000106839434,0.00012295526,0.0001625629,0.0012507705,0.0010064376],"genre_scores_gemma":[0.020868996,0.000742285,0.97457856,0.000110619396,0.00007645899,0.0004005313,0.00082693214,0.00030764964,0.0020879835],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99719256,0.0005449187,0.00021062138,0.00080438575,0.0010915598,0.00015597662],"domain_scores_gemma":[0.9983454,0.00042876418,0.00014875043,0.00030937663,0.0007042081,0.00006346764],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019298227,0.0013873727,0.002025308,0.0025846998,0.0021842541,0.0023697906,0.0026799901,0.0021619855,0.0031951505],"category_scores_gemma":[0.0055539976,0.00066255213,0.002031739,0.004418699,0.0011715551,0.0023276296,0.0021789675,0.0026540111,0.0038224694],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031297826,0.00013315803,0.0023270806,0.0008972589,0.00040563743,0.00020624824,0.0006186389,0.16427141,0.014079349,0.05482293,0.041099615,0.7208256],"study_design_scores_gemma":[0.000054751305,0.000104990104,0.0016344404,0.00013555524,0.00008999869,0.00048233723,0.00020097413,0.8622171,0.012587361,0.049120594,0.073189594,0.00018223842],"about_ca_topic_score_codex":0.009129803,"about_ca_topic_score_gemma":0.0084731765,"teacher_disagreement_score":0.009129803,"about_ca_system_score_codex":0.0012637741,"about_ca_system_score_gemma":0.0035426838,"threshold_uncertainty_score":0.01815331},"labels":[],"label_agreement":null},{"id":"W4230355264","doi":"10.1007/978-1-4939-7131-2_100724","title":"Network Dataset","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Computer science","score_opus":0.04994951798517892,"score_gpt":0.309039389115118,"score_spread":0.2590898711299391,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4230355264","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003468463,0.00046382606,0.0012226566,0.00035528964,0.00024337792,0.00011804497,0.9826469,0.0027119056,0.008769463],"genre_scores_gemma":[0.0040311827,0.00037038757,0.0021917277,0.00013845599,0.000046699388,0.00016883174,0.9896903,0.00015932435,0.0032032065],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992411,0.00012310274,0.000054667376,0.00023560663,0.00023862203,0.00010697046],"domain_scores_gemma":[0.99847037,0.00040085905,0.00011214952,0.00051746255,0.00033909592,0.0001600571],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061796204,0.0022288784,0.000782388,0.0045106104,0.00087104726,0.001479394,0.0018334166,0.0015576072,0.039178923],"category_scores_gemma":[0.0044343797,0.00038429812,0.000932788,0.0046586,0.00029068583,0.0016179172,0.0010429686,0.001396653,0.039211243],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009827559,0.00009614966,0.0011752362,0.0004894888,0.00004940467,0.000049778617,0.000025850853,0.0029533866,0.00058352854,0.0014422175,0.9793785,0.013658223],"study_design_scores_gemma":[0.00022432904,0.00010680448,0.0049851686,0.00022428177,0.00009651217,0.00026601905,0.00018491012,0.009371858,0.0021404976,0.006290513,0.9760562,0.0000529007],"about_ca_topic_score_codex":0.012210317,"about_ca_topic_score_gemma":0.02918608,"teacher_disagreement_score":0.039178923,"about_ca_system_score_codex":0.0010544359,"about_ca_system_score_gemma":0.0013750867,"threshold_uncertainty_score":0.13106662},"labels":[],"label_agreement":null},{"id":"W4231076572","doi":"10.1007/978-0-387-39940-9_2216","title":"Clustering Index","year":2009,"lang":"en","type":"book-chapter","venue":"Encyclopedia of Database Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Index (typography); Computer science; Artificial intelligence; World Wide Web","score_opus":0.021481385931594668,"score_gpt":0.27206912690312873,"score_spread":0.2505877409715341,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4231076572","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0065749874,0.004316302,0.12865293,0.0011390246,0.0025723884,0.0011019314,0.3069569,0.046822976,0.5018625],"genre_scores_gemma":[0.026213415,0.0061875605,0.1865715,0.0007604068,0.00074538524,0.0009539033,0.45036015,0.009401563,0.3188061],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99937385,0.000029272034,0.000044072887,0.00013470429,0.00036454783,0.000053504144],"domain_scores_gemma":[0.99899095,0.000115867275,0.00004555508,0.00024913502,0.0005150961,0.0000833983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003163054,0.0014579764,0.0013055606,0.008912007,0.0014604137,0.0030461112,0.0017644941,0.000831489,0.26872957],"category_scores_gemma":[0.002343585,0.00047441226,0.0008351209,0.011120398,0.00023099668,0.0027542058,0.0015884186,0.00090650015,0.2269764],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011225178,0.00005453312,0.00037994396,0.00046887965,0.000027136837,0.000073928684,0.00004245877,0.0014306951,0.0038210694,0.007773741,0.72268504,0.2631303],"study_design_scores_gemma":[0.000028616694,0.000040110597,0.0014038767,0.00012264283,0.000035176043,0.00042397142,0.000095523064,0.0053737345,0.005895356,0.014350082,0.9721872,0.000043742908],"about_ca_topic_score_codex":0.0029376321,"about_ca_topic_score_gemma":0.005345129,"teacher_disagreement_score":0.26872957,"about_ca_system_score_codex":0.0009821174,"about_ca_system_score_gemma":0.0012250328,"threshold_uncertainty_score":0.89899},"labels":[],"label_agreement":null},{"id":"W4231833091","doi":"10.1109/ijcnn.2006.1716663","title":"Fuzzy Clustering of Open-Source Software Quality Data: A Case Study of Mozilla","year":2006,"lang":"en","type":"article","venue":"The 2006 IEEE International Joint Conference on Neural Network Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Mozilla Foundation","keywords":"Computer science; Data mining; Software quality; Fuzzy logic; Software; Cluster analysis; Software metric; Ranking (information retrieval); Artificial intelligence; Software development; Operating system","score_opus":0.16967533549687067,"score_gpt":0.38844686424841907,"score_spread":0.2187715287515484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4231833091","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9847973,0.0004584202,0.011087398,0.0006189488,0.000014858998,0.00011717148,0.001261382,0.00037412223,0.001270319],"genre_scores_gemma":[0.95769733,0.0002373223,0.035824843,0.000077384386,0.00001711176,0.00011958662,0.004405203,0.00014583973,0.001475388],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972511,0.00069357204,0.00018488755,0.00052901986,0.001125,0.00021648512],"domain_scores_gemma":[0.98879075,0.005713514,0.0013763333,0.0012758035,0.002325251,0.0005183121],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034845585,0.00046314122,0.0007137178,0.005709455,0.0016760197,0.0015059162,0.0013246357,0.0015462819,0.0003577653],"category_scores_gemma":[0.014227715,0.00026476238,0.00094013644,0.006710973,0.0009354807,0.001024858,0.0010775165,0.00078825804,0.00019537844],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021553824,0.0015295731,0.4975295,0.0013998362,0.0010280639,0.011716991,0.015933253,0.112537876,0.035583336,0.008520849,0.021114718,0.2909506],"study_design_scores_gemma":[0.00024845861,0.0004668675,0.6304354,0.00023603397,0.00025029367,0.0035713527,0.0064708786,0.3019041,0.025457427,0.005368754,0.025349295,0.00024120255],"about_ca_topic_score_codex":0.053709056,"about_ca_topic_score_gemma":0.064244494,"teacher_disagreement_score":0.053709056,"about_ca_system_score_codex":0.0027847295,"about_ca_system_score_gemma":0.0010087937,"threshold_uncertainty_score":0.10679287},"labels":[],"label_agreement":null},{"id":"W4232476000","doi":"10.1007/978-1-4614-6170-8_100950","title":"Clustering Objective Function","year":2014,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.02291046649971108,"score_gpt":0.2636089981528809,"score_spread":0.24069853165316984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232476000","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019086415,0.0017022669,0.9115087,0.00068426994,0.0005318301,0.00012670935,0.0010022913,0.0012653456,0.08127004],"genre_scores_gemma":[0.057536658,0.004059053,0.6728678,0.0008717513,0.0006142205,0.0006474272,0.005872058,0.0026116436,0.25491947],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9990351,0.0001636697,0.00004000458,0.00022473569,0.0004786725,0.000057831945],"domain_scores_gemma":[0.9994522,0.000123385,0.000029941211,0.00009269735,0.00026470306,0.00003714787],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011514198,0.0016410609,0.0012585866,0.0017463273,0.00077581895,0.0022752557,0.0015860625,0.001534424,0.026308073],"category_scores_gemma":[0.001950746,0.00041806576,0.0009203469,0.001998518,0.00049818197,0.001900665,0.0016275884,0.001639297,0.017766139],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059544866,0.0000964333,0.00025734707,0.0004753322,0.000074248885,0.000046855926,0.00005439915,0.094045,0.009583459,0.12191098,0.15310822,0.6202882],"study_design_scores_gemma":[0.000020025178,0.00010998419,0.0009818021,0.00023622919,0.0000654476,0.0005323144,0.00010432537,0.5180019,0.0155510595,0.13419686,0.33010963,0.000090508154],"about_ca_topic_score_codex":0.0013185455,"about_ca_topic_score_gemma":0.002185541,"teacher_disagreement_score":0.026308073,"about_ca_system_score_codex":0.0013622377,"about_ca_system_score_gemma":0.0010994928,"threshold_uncertainty_score":0.08800924},"labels":[],"label_agreement":null},{"id":"W4234033320","doi":"10.1145/564410.564412","title":"Document clustering with committees","year":2002,"lang":"en","type":"article","venue":"Proceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '02","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Document clustering; Similarity (geometry); Set (abstract data type); Information retrieval; Task (project management); Data mining; Fuzzy clustering; Space (punctuation); Group (periodic table); Cluster (spacecraft); Artificial intelligence; Image (mathematics)","score_opus":0.06517323618068653,"score_gpt":0.3218864163618488,"score_spread":0.25671318018116224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4234033320","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0046055,0.00059556164,0.987379,0.0001459978,0.000117400945,0.00038801413,0.00045268633,0.0025745884,0.0037412914],"genre_scores_gemma":[0.06968011,0.00047160216,0.9204945,0.00015461212,0.00024911342,0.0007264272,0.0024111157,0.0007596722,0.0050527803],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98330647,0.0084580695,0.0014472804,0.0026135084,0.0037149782,0.0004596161],"domain_scores_gemma":[0.97884756,0.008628575,0.0012581812,0.0057687764,0.005040655,0.00045635406],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011341143,0.0015087554,0.0018639188,0.005260872,0.0018990266,0.004119373,0.0023810226,0.0020597912,0.006072144],"category_scores_gemma":[0.042945478,0.0008841383,0.0022590766,0.0074524065,0.0012039635,0.004288554,0.0031027896,0.0020261556,0.0049627135],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060008484,0.00016136438,0.0023814735,0.0006352631,0.00042249315,0.0001233209,0.00056946656,0.123865046,0.0056516533,0.058540486,0.030264338,0.776785],"study_design_scores_gemma":[0.00017230188,0.00037517835,0.0017890924,0.00013637717,0.00021519136,0.0005560393,0.00024900457,0.7769753,0.01947154,0.1285168,0.07142216,0.00012108093],"about_ca_topic_score_codex":0.0023460623,"about_ca_topic_score_gemma":0.0022915397,"teacher_disagreement_score":0.011341143,"about_ca_system_score_codex":0.0012367618,"about_ca_system_score_gemma":0.0016869194,"threshold_uncertainty_score":0.059978426},"labels":[],"label_agreement":null},{"id":"W4235761050","doi":"10.1007/978-1-4614-6170-8_100658","title":"Cluster Analysis","year":2014,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster (spacecraft); Computer science; Operating system","score_opus":0.022553855722679404,"score_gpt":0.2845981466991351,"score_spread":0.26204429097645565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4235761050","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017017083,0.0012745584,0.9118728,0.00054939673,0.00070302526,0.00030567322,0.0017155343,0.0071697766,0.074707545],"genre_scores_gemma":[0.025547748,0.0020559747,0.8139235,0.00049567985,0.0005667251,0.0007139111,0.0066754636,0.004071288,0.1459497],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99822384,0.00023971662,0.00008257905,0.00055743544,0.0007901594,0.000106234416],"domain_scores_gemma":[0.9984623,0.00032890338,0.00005621102,0.0004235812,0.0006772819,0.000051772662],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013403045,0.001788208,0.0015114328,0.0040438734,0.0020280443,0.003282479,0.0026428401,0.0009355458,0.05538906],"category_scores_gemma":[0.0035707958,0.0009031756,0.0018054611,0.0046936916,0.0008847522,0.0021019862,0.0022711935,0.0018269913,0.044701207],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007590009,0.00005606998,0.0004539817,0.0005189937,0.00012607002,0.000073683696,0.0003359311,0.008014171,0.00839289,0.107239164,0.2063953,0.6683179],"study_design_scores_gemma":[0.000024837016,0.00007345398,0.0020301624,0.00019343467,0.00012654763,0.0005991136,0.0004989648,0.08338171,0.02232304,0.17021312,0.7204124,0.00012317709],"about_ca_topic_score_codex":0.0023375344,"about_ca_topic_score_gemma":0.0038094765,"teacher_disagreement_score":0.05538906,"about_ca_system_score_codex":0.001324198,"about_ca_system_score_gemma":0.0017022224,"threshold_uncertainty_score":0.18529487},"labels":[],"label_agreement":null},{"id":"W4236921545","doi":"10.1109/ijcnn.2006.1716710","title":"PARTCAT: A Subspace Clustering Algorithm for High Dimensional Categorical Data","year":2006,"lang":"en","type":"article","venue":"The 2006 IEEE International Joint Conference on Neural Network Proceedings","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Subspace topology; Dimension (graph theory); Data mining; Similarity (geometry); Algorithm; Clustering high-dimensional data; Artificial neural network; Cluster (spacecraft); Artificial intelligence; Pattern recognition (psychology); Mathematics; Machine learning","score_opus":0.08700548172302576,"score_gpt":0.3200420799469654,"score_spread":0.23303659822393963,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4236921545","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026203846,0.00010034203,0.9952122,0.00006898117,0.00006902368,0.000083337174,0.00022366637,0.0011507196,0.0004713302],"genre_scores_gemma":[0.05266235,0.00020225308,0.9410489,0.00016958643,0.00008145356,0.00039973768,0.0022608647,0.0003696042,0.0028051422],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99800104,0.00053419516,0.00013288854,0.00042737165,0.00076884177,0.00013568926],"domain_scores_gemma":[0.9972947,0.0004900109,0.00015216006,0.0006953429,0.0012142422,0.00015351208],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001711393,0.0014244093,0.0020038644,0.003046576,0.0017898551,0.0021024554,0.0025626644,0.0016448951,0.0045768395],"category_scores_gemma":[0.0058653443,0.0006693224,0.0019400932,0.005330152,0.0009726528,0.0032827272,0.0026178716,0.0022429428,0.0032051941],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031809145,0.00018058879,0.0022057446,0.00023115706,0.00035453934,0.00011849335,0.00026855583,0.13049671,0.011467281,0.028406842,0.025144732,0.80080724],"study_design_scores_gemma":[0.00003381914,0.00011290272,0.0010237669,0.00002282877,0.000036681457,0.00020561791,0.00008129395,0.947122,0.004865456,0.035883687,0.010527416,0.000084492516],"about_ca_topic_score_codex":0.0065875375,"about_ca_topic_score_gemma":0.008067107,"teacher_disagreement_score":0.0065875375,"about_ca_system_score_codex":0.000864511,"about_ca_system_score_gemma":0.002666503,"threshold_uncertainty_score":0.015311003},"labels":[],"label_agreement":null},{"id":"W4237407686","doi":"10.1007/3-540-36175-8_8","title":"Automatic Extraction of Clusters from Hierarchical Clustering Representations","year":2003,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":95,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Hierarchical clustering; Hierarchical clustering of networks; Cluster analysis; Single-linkage clustering; Reachability; Brown clustering; Data mining; Dendrogram; Complete-linkage clustering; Correlation clustering; Set (abstract data type); Representation (politics); Cluster (spacecraft); Canopy clustering algorithm; Artificial intelligence; Theoretical computer science","score_opus":0.026161983350778222,"score_gpt":0.31537102485687596,"score_spread":0.2892090415060977,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4237407686","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017416434,0.0006932954,0.9639372,0.0002440693,0.00016051136,0.00036427067,0.0022551357,0.010726307,0.004202748],"genre_scores_gemma":[0.079262614,0.0004991587,0.90659076,0.00008752153,0.00008003434,0.0002818222,0.008277352,0.0013816415,0.003539172],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9985739,0.0002100516,0.00009631181,0.00039234495,0.00052333623,0.00020410474],"domain_scores_gemma":[0.99747604,0.000709982,0.0001892826,0.0004916275,0.0010276019,0.000105402476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010937096,0.0017766461,0.0019747484,0.006144172,0.0015361096,0.0035880748,0.0027144717,0.0014980515,0.0056109834],"category_scores_gemma":[0.0062914877,0.0011976487,0.0019458929,0.0064450023,0.00077313435,0.0021658095,0.0025306935,0.0021045343,0.007067174],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046364832,0.00016288685,0.002185713,0.00068686414,0.00018685599,0.00034474663,0.0007283212,0.017072508,0.036081966,0.015695563,0.04748762,0.87890327],"study_design_scores_gemma":[0.00018062063,0.00019087075,0.0069797053,0.0002989761,0.00049862056,0.0011402409,0.0012770147,0.77150625,0.04717517,0.10824301,0.062317193,0.00019242833],"about_ca_topic_score_codex":0.0055195615,"about_ca_topic_score_gemma":0.009911517,"teacher_disagreement_score":0.006144172,"about_ca_system_score_codex":0.0010109706,"about_ca_system_score_gemma":0.0020497767,"threshold_uncertainty_score":0.018770576},"labels":[],"label_agreement":null},{"id":"W4238474289","doi":"10.7287/peerj.preprints.1693","title":"A comparison of clustering methods for biogeography with fossil datasets","year":2016,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Cluster analysis; Hierarchical clustering; Euclidean distance; Set (abstract data type); Complete-linkage clustering; Computer science; Similarity (geometry); Data mining; Cluster (spacecraft); Range (aeronautics); Single-linkage clustering; Complete linkage; Data set; Mathematics; Statistics; Fuzzy clustering; Artificial intelligence; CURE data clustering algorithm; Biology; Engineering","score_opus":0.08915802525528156,"score_gpt":0.4734502304649979,"score_spread":0.38429220520971635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4238474289","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07849738,0.008925287,0.8868128,0.0016518435,0.0010831924,0.0009628538,0.0045976094,0.0065301848,0.010938828],"genre_scores_gemma":[0.12925525,0.003862489,0.8529043,0.0002044655,0.00016597344,0.0012144198,0.007352188,0.002860107,0.0021808904],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9775384,0.013359389,0.0011298573,0.0020012115,0.0055805426,0.0003904837],"domain_scores_gemma":[0.93893313,0.044392627,0.0015671554,0.006810218,0.007678773,0.0006180323],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03176718,0.0016455255,0.0015338273,0.008306606,0.0016446679,0.003079927,0.003377951,0.0021559556,0.0043501197],"category_scores_gemma":[0.08460713,0.00074174086,0.002503138,0.008795785,0.0010265925,0.0042367703,0.002692914,0.0020221253,0.0022487307],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00281572,0.00041165442,0.020072795,0.0042831074,0.0039948337,0.0002500569,0.002310885,0.24441193,0.00475327,0.05436805,0.035869304,0.6264584],"study_design_scores_gemma":[0.00035868105,0.000577908,0.021610422,0.0010866247,0.0003441058,0.0006112686,0.0011824167,0.8575244,0.005649282,0.065419644,0.045263987,0.00037128286],"about_ca_topic_score_codex":0.007960473,"about_ca_topic_score_gemma":0.008072137,"teacher_disagreement_score":0.03176718,"about_ca_system_score_codex":0.0024301254,"about_ca_system_score_gemma":0.002371561,"threshold_uncertainty_score":0.1680029},"labels":[],"label_agreement":null},{"id":"W4239679132","doi":"10.4018/9781591405573.ch031","title":"Clustering Analysis and Algorithms","year":2011,"lang":"en","type":"book-chapter","venue":"IGI Global eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Computer science; Single-linkage clustering; Cluster (spacecraft); Data mining; Similarity (geometry); Set (abstract data type); Pattern recognition (psychology); Artificial intelligence; Consensus clustering; Data set; Clustering high-dimensional data; Correlation clustering; CURE data clustering algorithm; Image (mathematics)","score_opus":0.03173086598057796,"score_gpt":0.2837229048107496,"score_spread":0.2519920388301716,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4239679132","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016402311,0.051973723,0.8600232,0.0045442744,0.0022174437,0.00096439663,0.0040923017,0.005973724,0.06857063],"genre_scores_gemma":[0.025056662,0.047479518,0.88550663,0.001788821,0.0017837767,0.0013605997,0.008062825,0.001439108,0.027522052],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9908264,0.002509777,0.0007660905,0.001902596,0.0037418255,0.00025335365],"domain_scores_gemma":[0.994941,0.0023077098,0.00028556646,0.00088902755,0.001462801,0.00011393614],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049320846,0.0029960675,0.0030639514,0.0077772243,0.0018030205,0.00808095,0.004381883,0.0031554159,0.027882421],"category_scores_gemma":[0.012246049,0.00096532033,0.0025926817,0.011527579,0.0021613357,0.004855532,0.0034378653,0.002757621,0.027578376],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000064024935,0.0000716318,0.0011218208,0.0036234097,0.0004557676,0.00021241196,0.0005398857,0.020275412,0.0013427102,0.21526434,0.17460065,0.5824281],"study_design_scores_gemma":[0.00002969038,0.000046521785,0.0010895443,0.0010224003,0.000093856215,0.00055416086,0.00034742904,0.037061006,0.0014437934,0.32773468,0.6304672,0.000109641995],"about_ca_topic_score_codex":0.0029845012,"about_ca_topic_score_gemma":0.0021245354,"teacher_disagreement_score":0.027882421,"about_ca_system_score_codex":0.0032165744,"about_ca_system_score_gemma":0.0029421428,"threshold_uncertainty_score":0.093276024},"labels":[],"label_agreement":null},{"id":"W4239884948","doi":"10.1007/978-1-4939-7131-2_100253","title":"Data Reduction Method","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Reduction (mathematics); Computer science; Mathematics; Geometry","score_opus":0.14797980495840338,"score_gpt":0.40624248348071534,"score_spread":0.258262678522312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4239884948","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003373302,0.0014312654,0.9131043,0.0008814802,0.001879446,0.0006893369,0.003016559,0.006783317,0.068840906],"genre_scores_gemma":[0.028183814,0.0017972456,0.8228709,0.0008955655,0.00078553066,0.0011357494,0.009659981,0.0018772961,0.13279395],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9984017,0.00012700546,0.000075574106,0.00045005194,0.00084698375,0.000098727716],"domain_scores_gemma":[0.9989126,0.00012551868,0.00002960028,0.0003836281,0.00050686183,0.000041840456],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007568384,0.0013752609,0.001299068,0.0033605017,0.0016293118,0.0025177703,0.0025922514,0.00084010046,0.052610897],"category_scores_gemma":[0.0018815983,0.00060523115,0.00168165,0.0029937683,0.00082688575,0.0015912139,0.0025882365,0.002335996,0.04662361],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100754034,0.00011693937,0.000399189,0.00051400997,0.00005310129,0.000081081605,0.00014987988,0.0035857304,0.020242438,0.055173445,0.108044565,0.8115388],"study_design_scores_gemma":[0.000053718875,0.000121904704,0.0011312756,0.00013764367,0.0001055118,0.0007034007,0.00024756585,0.046033517,0.057748202,0.06320749,0.830429,0.000080646045],"about_ca_topic_score_codex":0.0020030406,"about_ca_topic_score_gemma":0.0019302784,"teacher_disagreement_score":0.052610897,"about_ca_system_score_codex":0.0007334247,"about_ca_system_score_gemma":0.0019511796,"threshold_uncertainty_score":0.17600101},"labels":[],"label_agreement":null},{"id":"W4240667868","doi":"10.7287/peerj.preprints.1693v1","title":"A comparison of clustering methods for biogeography with fossil datasets","year":2016,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Cluster analysis; Hierarchical clustering; Euclidean distance; Complete-linkage clustering; Set (abstract data type); Computer science; Similarity (geometry); Data mining; Cluster (spacecraft); Single-linkage clustering; Complete linkage; Range (aeronautics); Data set; Mathematics; Fuzzy clustering; Artificial intelligence; CURE data clustering algorithm; Biology; Engineering","score_opus":0.08915802525528156,"score_gpt":0.4734502304649979,"score_spread":0.38429220520971635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4240667868","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07849738,0.008925287,0.8868128,0.0016518435,0.0010831924,0.0009628538,0.0045976094,0.0065301848,0.010938828],"genre_scores_gemma":[0.12925525,0.003862489,0.8529043,0.0002044655,0.00016597344,0.0012144198,0.007352188,0.002860107,0.0021808904],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9775384,0.013359389,0.0011298573,0.0020012115,0.0055805426,0.0003904837],"domain_scores_gemma":[0.93893313,0.044392627,0.0015671554,0.006810218,0.007678773,0.0006180323],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03176718,0.0016455255,0.0015338273,0.008306606,0.0016446679,0.003079927,0.003377951,0.0021559556,0.0043501197],"category_scores_gemma":[0.08460713,0.00074174086,0.002503138,0.008795785,0.0010265925,0.0042367703,0.002692914,0.0020221253,0.0022487307],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00281572,0.00041165442,0.020072795,0.0042831074,0.0039948337,0.0002500569,0.002310885,0.24441193,0.00475327,0.05436805,0.035869304,0.6264584],"study_design_scores_gemma":[0.00035868105,0.000577908,0.021610422,0.0010866247,0.0003441058,0.0006112686,0.0011824167,0.8575244,0.005649282,0.065419644,0.045263987,0.00037128286],"about_ca_topic_score_codex":0.007960473,"about_ca_topic_score_gemma":0.008072137,"teacher_disagreement_score":0.03176718,"about_ca_system_score_codex":0.0024301254,"about_ca_system_score_gemma":0.002371561,"threshold_uncertainty_score":0.1680029},"labels":[],"label_agreement":null},{"id":"W4241307588","doi":"10.1007/978-1-4939-7131-2_100106","title":"Cluster Description","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster (spacecraft); Computer science; Programming language","score_opus":0.04605066862676165,"score_gpt":0.2856856847795797,"score_spread":0.23963501615281801,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241307588","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036637404,0.0018801988,0.5529019,0.0012814923,0.0018722166,0.0017188694,0.08453129,0.058158614,0.29399163],"genre_scores_gemma":[0.029533794,0.0025954014,0.470249,0.0012265958,0.00043363502,0.0026906247,0.14160748,0.019700374,0.33196318],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99935454,0.000054382577,0.00003607498,0.00016626269,0.0003088281,0.00007990161],"domain_scores_gemma":[0.999049,0.000106889944,0.000024688134,0.00028703499,0.0004524281,0.00007991651],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007135655,0.0014223461,0.0010551688,0.0025821812,0.00199015,0.002699173,0.0031770621,0.0010480039,0.21085554],"category_scores_gemma":[0.0022557245,0.00075446855,0.0010956159,0.0033996322,0.0003758223,0.0025131388,0.0022218905,0.0016321423,0.13566366],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020568498,0.000051472613,0.00026601308,0.0006889134,0.000030717943,0.0000995814,0.000117990654,0.0066463174,0.0092555955,0.043390624,0.7251608,0.21408625],"study_design_scores_gemma":[0.000026013722,0.000031562937,0.00032017037,0.000069307505,0.000019749943,0.00022672409,0.00009093173,0.014077674,0.009219061,0.023276854,0.9525895,0.000052457384],"about_ca_topic_score_codex":0.0047345553,"about_ca_topic_score_gemma":0.006408173,"teacher_disagreement_score":0.21085554,"about_ca_system_score_codex":0.0012393254,"about_ca_system_score_gemma":0.0018932465,"threshold_uncertainty_score":0.7053821},"labels":[],"label_agreement":null},{"id":"W4241938281","doi":"10.1145/376284.375672","title":"Data bubbles","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Cluster analysis; Computer science; Data mining; CURE data clustering algorithm; Data stream clustering; Canopy clustering algorithm; Hierarchical clustering; Correlation clustering; Set (abstract data type); Data set; Data compression; Key (lock); Constrained clustering; Determining the number of clusters in a data set; Fuzzy clustering; Algorithm; Artificial intelligence","score_opus":0.123803096386827,"score_gpt":0.366511782884858,"score_spread":0.24270868649803096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241938281","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016694076,0.00478333,0.9249112,0.008200641,0.002105885,0.0007143421,0.003767193,0.0040272474,0.034796122],"genre_scores_gemma":[0.18941064,0.0041444814,0.75151587,0.00576521,0.001150831,0.0013564934,0.009064232,0.0019103991,0.035681877],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9950486,0.0011151726,0.0003746202,0.0010955689,0.0021119881,0.0002540601],"domain_scores_gemma":[0.9850007,0.0068379696,0.0007109692,0.004603016,0.002271746,0.0005755757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004386181,0.0007071546,0.0009842973,0.0022737775,0.0017261559,0.0038416295,0.0030857143,0.001957915,0.024653232],"category_scores_gemma":[0.02955338,0.00058529276,0.0010680081,0.0029977176,0.002020341,0.011695137,0.006867439,0.0025594598,0.007902516],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052630843,0.00012455783,0.0030848607,0.00084606075,0.000083555926,0.0006828432,0.0013627026,0.008804914,0.0053592813,0.5522953,0.07293121,0.35389844],"study_design_scores_gemma":[0.00007453088,0.00017459349,0.0008361073,0.00022629075,0.0000469375,0.001445506,0.00079367036,0.03794724,0.007958464,0.3173856,0.6330255,0.000085509586],"about_ca_topic_score_codex":0.001573575,"about_ca_topic_score_gemma":0.0011476026,"teacher_disagreement_score":0.024653232,"about_ca_system_score_codex":0.0013069317,"about_ca_system_score_gemma":0.0012555909,"threshold_uncertainty_score":0.08247328},"labels":[],"label_agreement":null},{"id":"W4243066799","doi":"10.1007/978-1-4939-7131-2_100110","title":"Clustering Evaluation","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.0745023061247611,"score_gpt":0.3441617868366472,"score_spread":0.2696594807118861,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4243066799","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.105067484,0.0080131525,0.38824314,0.0029534835,0.0052716252,0.004210868,0.075456195,0.04397701,0.36680704],"genre_scores_gemma":[0.31197065,0.0020243216,0.335022,0.0011340155,0.00060062774,0.0016165724,0.17279597,0.008423414,0.16641255],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99073255,0.0016357901,0.00048155562,0.0016096287,0.00487834,0.00066213653],"domain_scores_gemma":[0.98875016,0.0015843085,0.00024287446,0.0022468504,0.0068141525,0.00036154935],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058887373,0.0021751646,0.001696752,0.00602812,0.0021291361,0.003964693,0.0029171782,0.0016463444,0.04795622],"category_scores_gemma":[0.019443639,0.00042872617,0.0014919852,0.0052790274,0.000640985,0.0026660212,0.001963464,0.0012536747,0.0234748],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011912265,0.000509586,0.004932078,0.0008842116,0.00038439862,0.00010532492,0.00013324416,0.028277995,0.007311829,0.011799244,0.3303754,0.61409545],"study_design_scores_gemma":[0.00045394135,0.0012139363,0.025710033,0.00066324667,0.00071431755,0.0010785141,0.0012746454,0.476029,0.08873157,0.036274325,0.36751875,0.00033766034],"about_ca_topic_score_codex":0.012010886,"about_ca_topic_score_gemma":0.019348329,"teacher_disagreement_score":0.04795622,"about_ca_system_score_codex":0.002897274,"about_ca_system_score_gemma":0.0025495163,"threshold_uncertainty_score":0.16042954},"labels":[],"label_agreement":null},{"id":"W4244787383","doi":"10.23952/jano.3.2021.1.09","title":"A survey of clustering methods via optimization methodology","year":2021,"lang":"en","type":"article","venue":"Journal of Applied and Numerical Optimization","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Department of Education of Guangdong Province; Shenzhen University; Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Data mining; Artificial intelligence","score_opus":0.0701899445410024,"score_gpt":0.37791770589676704,"score_spread":0.30772776135576463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4244787383","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011161086,0.063038684,0.9260242,0.0011662309,0.00044642188,0.000082098864,0.00023817774,0.00035717004,0.007530788],"genre_scores_gemma":[0.043301404,0.15248853,0.7912197,0.001061166,0.0026599008,0.0005705455,0.0010693419,0.000607265,0.0070221215],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9964179,0.0013412397,0.00027132206,0.00063587446,0.0012263296,0.00010736058],"domain_scores_gemma":[0.9981014,0.000920018,0.00011187972,0.00023454719,0.0005838551,0.000048343194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003288022,0.001510822,0.002508498,0.0038218594,0.001048764,0.0029699802,0.0024230094,0.0021230783,0.0040795114],"category_scores_gemma":[0.00584874,0.0009915356,0.0023411056,0.00750701,0.0014198817,0.0028703217,0.0017168713,0.0021090163,0.0033785466],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007122028,0.00007205142,0.0010768911,0.004267282,0.0003274003,0.00013250984,0.00027640688,0.07554751,0.002739934,0.32388523,0.023678618,0.567925],"study_design_scores_gemma":[0.000040919644,0.00012955612,0.0018978548,0.0011298993,0.00012978009,0.00065941905,0.0001767572,0.33319268,0.0037116013,0.3743321,0.284441,0.00015845349],"about_ca_topic_score_codex":0.002264878,"about_ca_topic_score_gemma":0.0013715834,"teacher_disagreement_score":0.0040795114,"about_ca_system_score_codex":0.0016764252,"about_ca_system_score_gemma":0.0019783087,"threshold_uncertainty_score":0.01738888},"labels":[],"label_agreement":null},{"id":"W4246424543","doi":"10.1007/978-1-4939-7131-2_100111","title":"Clustering Objective Function","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.028964105373926644,"score_gpt":0.2778500299480699,"score_spread":0.24888592457414327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4246424543","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020846971,0.0016938304,0.9064335,0.0008020795,0.00061586255,0.00014749785,0.0012473857,0.0014541713,0.08552093],"genre_scores_gemma":[0.062368494,0.003988379,0.66209644,0.0010113759,0.0006934043,0.0007472689,0.0071599004,0.0029800714,0.2589547],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9989748,0.00018057843,0.00004316027,0.00024907704,0.0004887,0.000063716216],"domain_scores_gemma":[0.9993735,0.00013985213,0.000033847573,0.000110329755,0.00029792814,0.000044610766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012600482,0.0017128126,0.0012559479,0.0017708619,0.00079264713,0.0024455297,0.0016315356,0.001620218,0.029412368],"category_scores_gemma":[0.0022814872,0.00041850097,0.00095217215,0.0019454865,0.0005028059,0.0019853087,0.0017483741,0.0017578715,0.02003735],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006814197,0.00011077424,0.00029059435,0.0005295408,0.00008368227,0.000047821937,0.000055337037,0.09609148,0.009817222,0.13060255,0.16474503,0.5975578],"study_design_scores_gemma":[0.00002230332,0.00011740013,0.000982024,0.00024776126,0.00006904211,0.00050704886,0.000103619575,0.5205324,0.015300066,0.13500302,0.327027,0.000088262845],"about_ca_topic_score_codex":0.0012995055,"about_ca_topic_score_gemma":0.0020995862,"teacher_disagreement_score":0.029412368,"about_ca_system_score_codex":0.0014224275,"about_ca_system_score_gemma":0.0011869763,"threshold_uncertainty_score":0.098394156},"labels":[],"label_agreement":null},{"id":"W4246517075","doi":"10.17771/pucrio.acad.54595","title":"A MODEL-BASED FRAMEWORK FOR SEMI-SUPERVISED CLUSTERING AND COMMUNITY DETECTION","year":2021,"lang":"en","type":"dissertation","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Pontifícia Universidade Católica do Rio de Janeiro; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Centre interuniversitaire de recherche sur les reseaux d'entreprise, la logistique et le transport; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Cluster analysis; Computer science; Focus (optics); Data mining; Perspective (graphical); Machine learning; Artificial intelligence; Stochastic block model","score_opus":0.05005488187165928,"score_gpt":0.34649199057940594,"score_spread":0.29643710870774664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4246517075","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00033012082,0.00008527141,0.99907684,0.000101707796,0.000013774174,0.000028860304,0.00003355186,0.00007928001,0.0002505002],"genre_scores_gemma":[0.056170713,0.00048801932,0.9394203,0.0002531963,0.00018319812,0.0006037354,0.0005413431,0.00024494148,0.002094566],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9925761,0.003879336,0.0003113303,0.0014742734,0.0015269903,0.00023197125],"domain_scores_gemma":[0.9900829,0.0053998027,0.0009914188,0.0016926879,0.0014736672,0.00035954337],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0073135705,0.0019811625,0.0022942624,0.003167436,0.0015856525,0.003312813,0.0045679193,0.002807283,0.002227956],"category_scores_gemma":[0.017034655,0.0014528857,0.0031497844,0.0036783987,0.003148506,0.0041315327,0.0044632307,0.0048150485,0.00192974],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008174395,0.00015542752,0.0011381514,0.00046384297,0.0003365646,0.00021679552,0.0007235731,0.51930994,0.0046888944,0.3342183,0.0074094865,0.13125733],"study_design_scores_gemma":[0.00000957151,0.000031756772,0.00014987602,0.000036930203,0.000019745044,0.00008550958,0.00003832158,0.8628231,0.0006907173,0.13179511,0.0042907763,0.000028611028],"about_ca_topic_score_codex":0.0038458235,"about_ca_topic_score_gemma":0.0050377073,"teacher_disagreement_score":0.0073135705,"about_ca_system_score_codex":0.0027657514,"about_ca_system_score_gemma":0.0031005589,"threshold_uncertainty_score":0.03867829},"labels":[],"label_agreement":null},{"id":"W4247582225","doi":"10.32920/ryerson.14648265.v1","title":"k-MACE Clustering for Gaussian Clusters","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Mace; Cluster analysis; Preprocessor; Computer science; Mathematics; Gaussian; Cluster (spacecraft); Algorithm; Covariance; Statistics; Artificial intelligence; Physics; Medicine","score_opus":0.04531905812185061,"score_gpt":0.34262640442645365,"score_spread":0.29730734630460304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4247582225","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035217574,0.0003411149,0.9938881,0.00019677085,0.000052578795,0.000076073615,0.00010778472,0.0006301055,0.0011858039],"genre_scores_gemma":[0.11542304,0.00048074787,0.87677044,0.0003045269,0.000088325345,0.00031053153,0.00084544555,0.00035514834,0.0054217787],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99619484,0.0011792545,0.00023972931,0.00092655676,0.001207492,0.0002520304],"domain_scores_gemma":[0.9929604,0.002690332,0.00040871368,0.0014966216,0.0022399647,0.00020399845],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048472797,0.0014576189,0.0020085787,0.0030243737,0.0019688723,0.0024611016,0.003555312,0.0030421433,0.0036277215],"category_scores_gemma":[0.018137336,0.0007844645,0.0020155972,0.0031049794,0.001605232,0.002946234,0.0030892175,0.0025779821,0.0024572322],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047447675,0.00014056773,0.0016383997,0.00037001568,0.00025301962,0.00014202966,0.0004656378,0.66775,0.0042683305,0.07304437,0.0182694,0.23318379],"study_design_scores_gemma":[0.000011642706,0.000021502106,0.00022126963,0.000015289883,0.000010289056,0.00006193531,0.000047597216,0.97694004,0.0014290984,0.018505698,0.0027158798,0.000019736712],"about_ca_topic_score_codex":0.011280503,"about_ca_topic_score_gemma":0.01283352,"teacher_disagreement_score":0.011280503,"about_ca_system_score_codex":0.0019393113,"about_ca_system_score_gemma":0.0030489406,"threshold_uncertainty_score":0.025635123},"labels":[],"label_agreement":null},{"id":"W4247791306","doi":"10.1137/1.9781611976335.ch11","title":"Chapter 11: Graph-Based Clustering Algorithms","year":2020,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Hypergraph; Computer science; Clustering coefficient; Graph; Null graph; Voltage graph; Butterfly graph; Strength of a graph; Theoretical computer science; Algorithm; Mathematics; Line graph; Combinatorics; Artificial intelligence","score_opus":0.09129902419943707,"score_gpt":0.27567703356965967,"score_spread":0.18437800937022258,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4247791306","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00089316705,0.020863567,0.9172193,0.0016084139,0.0017289292,0.00030243027,0.0019734476,0.004375284,0.05103538],"genre_scores_gemma":[0.017698206,0.035121933,0.85810345,0.0011320198,0.0014307547,0.00055536383,0.008482509,0.002600442,0.07487536],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9981483,0.00033836672,0.00009158456,0.00042833347,0.000916213,0.00007712348],"domain_scores_gemma":[0.99871683,0.00051738526,0.00004938718,0.00021366957,0.00045916962,0.000043527423],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015878637,0.002595412,0.0017287263,0.0033472565,0.0012045021,0.004290479,0.0030093947,0.0020943368,0.02605288],"category_scores_gemma":[0.004386666,0.0010583416,0.0015761887,0.007555267,0.0008813769,0.0034958276,0.0018180505,0.0029630542,0.0278763],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005807536,0.00008206029,0.00045909092,0.0012001874,0.00017350884,0.00011603375,0.00020120067,0.046653774,0.0023749268,0.10418124,0.25540006,0.5890998],"study_design_scores_gemma":[0.000025801224,0.000047543956,0.0006936178,0.0004928981,0.000074217336,0.00055530656,0.00013204702,0.111226305,0.0031751262,0.19250935,0.69097996,0.0000877339],"about_ca_topic_score_codex":0.0041756243,"about_ca_topic_score_gemma":0.0044071833,"teacher_disagreement_score":0.02605288,"about_ca_system_score_codex":0.001603167,"about_ca_system_score_gemma":0.0014609443,"threshold_uncertainty_score":0.08715558},"labels":[],"label_agreement":null},{"id":"W4248125270","doi":"10.1137/1.9781611976335.ch18","title":"Chapter 18: Evaluation of Clustering Algorithms","year":2020,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Cluster analysis; Computer science; Correlation clustering; Data mining; CURE data clustering algorithm; Determining the number of clusters in a data set; Single-linkage clustering; Set (abstract data type); Fuzzy clustering; Cluster (spacecraft); Canopy clustering algorithm; Clustering high-dimensional data; Algorithm; Artificial intelligence","score_opus":0.19723447155785484,"score_gpt":0.32563627083338675,"score_spread":0.1284017992755319,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4248125270","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028092923,0.06281843,0.6439031,0.0037168337,0.0046512936,0.0018999478,0.009122576,0.008020353,0.23777461],"genre_scores_gemma":[0.1310135,0.028748127,0.7582388,0.00093927834,0.0013020234,0.0017734979,0.017340224,0.004586922,0.056057625],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9825549,0.0046792505,0.0007849935,0.0017069047,0.009924473,0.0003494116],"domain_scores_gemma":[0.9793768,0.010543645,0.0006264197,0.0016658049,0.0075347186,0.00025263597],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010196491,0.0024562387,0.0019840903,0.005391401,0.0016298116,0.0054694694,0.002504528,0.0019487891,0.026275473],"category_scores_gemma":[0.04860757,0.0005892891,0.0012212144,0.0068726996,0.0008826232,0.004499089,0.0019267465,0.0015174587,0.011557893],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050085183,0.00022692607,0.0033687586,0.0032248579,0.0003509746,0.00012131335,0.0004793115,0.040608156,0.0031284436,0.06756152,0.1335512,0.74687773],"study_design_scores_gemma":[0.00017711554,0.0007174411,0.009505638,0.0035524545,0.0004059759,0.0008878546,0.0012860661,0.32205704,0.036603205,0.12406818,0.5004987,0.00024043117],"about_ca_topic_score_codex":0.00285836,"about_ca_topic_score_gemma":0.0018586772,"teacher_disagreement_score":0.026275473,"about_ca_system_score_codex":0.002477971,"about_ca_system_score_gemma":0.0019290423,"threshold_uncertainty_score":0.08790028},"labels":[],"label_agreement":null},{"id":"W4249726331","doi":"10.1137/1.9781611976335.ch7","title":"Chapter 7: Hierarchical Clustering Techniques","year":2020,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Hierarchical clustering; Hierarchical clustering of networks; Computer science; Partition (number theory); Cluster analysis; Hierarchical database model; Single-linkage clustering; Algorithm; Set (abstract data type); Data mining; Hierarchical control system; Pattern recognition (psychology); Canopy clustering algorithm; Artificial intelligence; Mathematics; Correlation clustering; Combinatorics","score_opus":0.08103529639172104,"score_gpt":0.2813290510848872,"score_spread":0.20029375469316618,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4249726331","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00095724757,0.022323294,0.87635607,0.0015352133,0.0012841544,0.0004344729,0.0015179628,0.005421274,0.0901704],"genre_scores_gemma":[0.015029805,0.027991585,0.830358,0.0013334749,0.0016921258,0.0005452139,0.0054657995,0.0026703766,0.114913575],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9978034,0.00030471047,0.00010096593,0.00038513762,0.0013066204,0.000099115954],"domain_scores_gemma":[0.99883765,0.00034128636,0.00005005032,0.0002684754,0.00045434444,0.000048233876],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016888622,0.00166046,0.0012899775,0.0030342485,0.0016097457,0.004210707,0.0028516108,0.0017124066,0.05120304],"category_scores_gemma":[0.003185156,0.0010434173,0.0017319798,0.005042569,0.00083395874,0.0035034553,0.0022265636,0.0027206657,0.04754225],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004203328,0.00005397478,0.0002878452,0.0012516686,0.00011369536,0.0001284528,0.0004252168,0.010979255,0.00539603,0.12439824,0.26403025,0.5928933],"study_design_scores_gemma":[0.000015311402,0.000032854434,0.0005296771,0.0004554546,0.000051844658,0.0004932554,0.0001411681,0.022678936,0.004821176,0.10512681,0.8655975,0.000056139565],"about_ca_topic_score_codex":0.0027034522,"about_ca_topic_score_gemma":0.0035695285,"teacher_disagreement_score":0.05120304,"about_ca_system_score_codex":0.0014934536,"about_ca_system_score_gemma":0.0016040945,"threshold_uncertainty_score":0.17129123},"labels":[],"label_agreement":null},{"id":"W4250669299","doi":"10.1007/978-1-4939-7131-2_100425","title":"Graph Clustering","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.03630852760864116,"score_gpt":0.29069633507219084,"score_spread":0.25438780746354966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4250669299","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022510174,0.0107165165,0.71068275,0.0019666106,0.0025877333,0.0002627063,0.0019751936,0.008270892,0.2612866],"genre_scores_gemma":[0.031137872,0.0137916785,0.5552347,0.001353682,0.0015997954,0.0003280303,0.008782156,0.0047257855,0.3830463],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99947804,0.0000635665,0.000015635522,0.00018815263,0.00022297549,0.000031747426],"domain_scores_gemma":[0.9994118,0.00011200525,0.000019500698,0.00022049621,0.00019462378,0.00004150329],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039559044,0.0015411583,0.00096771494,0.0027676723,0.0011474928,0.0018769698,0.0019025094,0.0010973874,0.055439603],"category_scores_gemma":[0.0013598081,0.0007411381,0.0010922186,0.0040405616,0.0007101843,0.0025685215,0.0016360374,0.0018618734,0.04873253],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027248228,0.000051870942,0.00011051773,0.0004403958,0.000051825602,0.000056222794,0.00008949112,0.010545888,0.003535448,0.09948694,0.31928033,0.5663238],"study_design_scores_gemma":[0.0000103387265,0.00003173286,0.00044540328,0.00015725971,0.00003969052,0.00040583755,0.000088314955,0.030860998,0.0060074017,0.19251849,0.76939094,0.00004350276],"about_ca_topic_score_codex":0.0017611982,"about_ca_topic_score_gemma":0.003454021,"teacher_disagreement_score":0.055439603,"about_ca_system_score_codex":0.0011019143,"about_ca_system_score_gemma":0.0008964558,"threshold_uncertainty_score":0.18546396},"labels":[],"label_agreement":null},{"id":"W4251878303","doi":"10.1137/1.9781611976335.ch10","title":"Chapter 10: Search-Based Clustering Algorithms","year":2020,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Computer science; Cluster analysis; Algorithm; Artificial intelligence","score_opus":0.11434634602538142,"score_gpt":0.286528705407464,"score_spread":0.1721823593820826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4251878303","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001549032,0.021564592,0.8885827,0.0012894059,0.0012819803,0.00019885953,0.00096907944,0.0024057177,0.08215856],"genre_scores_gemma":[0.04022491,0.026153814,0.7901886,0.00128179,0.0016267255,0.00047044482,0.004376455,0.0018440039,0.1338333],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991285,0.00014638482,0.000042753443,0.00021199085,0.00042647222,0.000043891836],"domain_scores_gemma":[0.9994855,0.00022924681,0.000019335923,0.00011039285,0.00013360567,0.000022013015],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007774275,0.0013859529,0.001352083,0.0014582763,0.0007947686,0.0027390474,0.00205782,0.0015048848,0.028622612],"category_scores_gemma":[0.0023629868,0.00055680436,0.0010157005,0.0034687186,0.0006870889,0.0020507982,0.0014341698,0.0021214453,0.020619554],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007274893,0.000107044725,0.00029531537,0.0006879469,0.0001249345,0.000064216205,0.00010309205,0.06262927,0.0027312376,0.11060296,0.16225816,0.660323],"study_design_scores_gemma":[0.000030506151,0.000063130035,0.0006116348,0.00031954306,0.000053977485,0.00043303063,0.00006312148,0.2350184,0.0038802996,0.24550699,0.5139685,0.000050793013],"about_ca_topic_score_codex":0.0021705395,"about_ca_topic_score_gemma":0.0024098183,"teacher_disagreement_score":0.028622612,"about_ca_system_score_codex":0.0010547427,"about_ca_system_score_gemma":0.0009322743,"threshold_uncertainty_score":0.09575212},"labels":[],"label_agreement":null},{"id":"W4252408959","doi":"10.1007/978-1-4939-7131-2_100104","title":"Cluster Analysis","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster (spacecraft); Computer science; Operating system","score_opus":0.028849627619853005,"score_gpt":0.3001337533546835,"score_spread":0.2712841257348305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4252408959","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019024425,0.0013073339,0.899739,0.0006240312,0.00081724656,0.00034943703,0.002144014,0.008349237,0.08476728],"genre_scores_gemma":[0.027968,0.0020754107,0.80319077,0.0005433092,0.0006202579,0.00078987033,0.008160671,0.004603205,0.15204847],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99811673,0.00025409626,0.0000882131,0.00059791317,0.00082883204,0.000114236165],"domain_scores_gemma":[0.9982899,0.00035300755,0.00006152491,0.00047742564,0.0007598503,0.00005830518],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014035325,0.0017784381,0.0014551011,0.0041818745,0.0021000968,0.0034745387,0.0026459617,0.00095571234,0.06024344],"category_scores_gemma":[0.0039793225,0.0008726178,0.001818066,0.0048580286,0.0008709818,0.0021336926,0.002319217,0.0018208382,0.04914012],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008181122,0.000060258313,0.0004902053,0.00055258,0.00013099478,0.000074236355,0.00035001,0.0074797804,0.008780005,0.10798239,0.21786733,0.65615034],"study_design_scores_gemma":[0.000025051711,0.00007350878,0.0020641536,0.00019685559,0.00012607542,0.00055683404,0.00049923017,0.076159306,0.022325054,0.15526056,0.7425942,0.00011923361],"about_ca_topic_score_codex":0.0023751052,"about_ca_topic_score_gemma":0.0037411419,"teacher_disagreement_score":0.06024344,"about_ca_system_score_codex":0.001366155,"about_ca_system_score_gemma":0.001808319,"threshold_uncertainty_score":0.20153439},"labels":[],"label_agreement":null},{"id":"W4253573012","doi":"10.1137/1.9781611976335.ch2","title":"Chapter 2: Data Types","year":2020,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Computer science; Data type; Unstructured data; Cluster analysis; Information retrieval; Data mining; Artificial intelligence; Big data; Programming language","score_opus":0.18043614749240725,"score_gpt":0.302795166268206,"score_spread":0.12235901877579877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4253573012","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027661007,0.06717976,0.37647387,0.0117522115,0.010291374,0.0032217845,0.109671354,0.015599142,0.40304434],"genre_scores_gemma":[0.02039236,0.06699351,0.43039948,0.011100367,0.005649242,0.004798843,0.12689033,0.011270673,0.3225053],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99559575,0.000770345,0.00060023885,0.0008051983,0.002093979,0.00013458927],"domain_scores_gemma":[0.992822,0.0036168243,0.00033433424,0.0013001545,0.0016800748,0.0002466882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031980341,0.0013623798,0.0013614207,0.0060277255,0.0019305713,0.009711809,0.0026789184,0.0019090036,0.106617175],"category_scores_gemma":[0.015957123,0.0009153002,0.0012053325,0.01108103,0.0011246852,0.010198363,0.003033177,0.0037167263,0.07965548],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007749418,0.00005302707,0.0004462409,0.0020676253,0.000025060423,0.00022768728,0.00050325657,0.0011875359,0.0009758097,0.19103517,0.5239041,0.27949697],"study_design_scores_gemma":[0.000003162337,0.000005206761,0.000096233045,0.0003447025,0.000004257476,0.0002367264,0.000050923572,0.0002166483,0.00022005202,0.022357337,0.9764538,0.000011070067],"about_ca_topic_score_codex":0.0015006391,"about_ca_topic_score_gemma":0.0013273573,"teacher_disagreement_score":0.106617175,"about_ca_system_score_codex":0.001935249,"about_ca_system_score_gemma":0.0020842815,"threshold_uncertainty_score":0.35667002},"labels":[],"label_agreement":null},{"id":"W4254373892","doi":"10.1007/978-1-4614-6170-8_100953","title":"Graph Clustering","year":2014,"lang":"en","type":"book-chapter","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.02848160426955035,"score_gpt":0.2753965071202259,"score_spread":0.24691490285067555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4254373892","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022205303,0.010311133,0.7327801,0.0016882878,0.0021093227,0.0002447128,0.001662551,0.007469678,0.24151364],"genre_scores_gemma":[0.030125523,0.013253235,0.5689875,0.0011880012,0.0013432567,0.00031150325,0.0073370454,0.0042094267,0.37324452],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995047,0.000060380906,0.000014689921,0.0001734532,0.00021636167,0.000030420899],"domain_scores_gemma":[0.99947757,0.000100350786,0.000017821103,0.00019320498,0.00017554878,0.00003554758],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036740126,0.001530834,0.000995048,0.0027474288,0.0011263625,0.0017353566,0.0018830989,0.0010578396,0.05009547],"category_scores_gemma":[0.0011885047,0.0007525761,0.0010555565,0.0040330715,0.0007091948,0.002473463,0.0015728133,0.0017875462,0.041965995],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025724145,0.000050626724,0.00010768577,0.00041194996,0.00004957844,0.000058185822,0.00009044844,0.011399954,0.0035848222,0.094364166,0.29771972,0.5921372],"study_design_scores_gemma":[0.00001076981,0.00003296775,0.0004695911,0.0001594664,0.000042048214,0.00045557908,0.000096511365,0.035930716,0.0067602447,0.19893734,0.75705785,0.00004708637],"about_ca_topic_score_codex":0.0019013063,"about_ca_topic_score_gemma":0.003855649,"teacher_disagreement_score":0.05009547,"about_ca_system_score_codex":0.0010762602,"about_ca_system_score_gemma":0.00085903367,"threshold_uncertainty_score":0.16758603},"labels":[],"label_agreement":null},{"id":"W4281975883","doi":"10.1145/3805034","title":"Streaming Hierarchical Clustering Based on Point-Set Kernel","year":2022,"lang":"en","type":"preprint","venue":"ACM Transactions on Knowledge Discovery from Data","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"State Key Laboratory of Novel Software Technology; Nanjing University; National Natural Science Foundation of China","keywords":"Hierarchical clustering; Cluster analysis; Computer science; Data mining; Single-linkage clustering; Brown clustering; Tree (set theory); Consensus clustering; Correlation clustering; Hierarchical clustering of networks; CURE data clustering algorithm; Scalability; Kernel (algebra); Tree structure; Constrained clustering; Artificial intelligence; Mathematics; Algorithm; Database; Binary tree","score_opus":0.07734526681048454,"score_gpt":0.3535982704819712,"score_spread":0.27625300367148664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281975883","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.050538886,0.00021713287,0.9464798,0.00010702428,0.000040184048,0.00010922219,0.00019568244,0.001368327,0.0009438687],"genre_scores_gemma":[0.58594936,0.00014389881,0.4111653,0.000053462216,0.000038884056,0.00009514342,0.0008726843,0.0001427719,0.0015384282],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982589,0.00034042078,0.00010976511,0.00037048926,0.0007787325,0.00014179459],"domain_scores_gemma":[0.995756,0.00096117775,0.00041913008,0.00091892376,0.0017053123,0.00023951083],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001455392,0.00052677875,0.0013158765,0.0027596767,0.0008972445,0.0013143646,0.0021902649,0.0007687507,0.0013163214],"category_scores_gemma":[0.006048637,0.0003819089,0.00083369366,0.0033975872,0.00070325227,0.0019222613,0.0017318833,0.0009941638,0.0006816812],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003153714,0.00017599946,0.0050422433,0.00016345392,0.00013648116,0.00016745464,0.00035411984,0.73250055,0.012197332,0.023124266,0.0045285816,0.22129416],"study_design_scores_gemma":[0.000004346829,0.000012988229,0.00030588167,0.000002004946,0.000003995419,0.000017980254,0.000014963267,0.9952342,0.0010524103,0.0030114013,0.00033286016,0.0000070317733],"about_ca_topic_score_codex":0.0141627025,"about_ca_topic_score_gemma":0.010351281,"teacher_disagreement_score":0.0141627025,"about_ca_system_score_codex":0.0018896104,"about_ca_system_score_gemma":0.0015597786,"threshold_uncertainty_score":0.028160512},"labels":[],"label_agreement":null},{"id":"W4285364251","doi":"10.1609/aaai.v26i1.8282","title":"Weighted Clustering","year":2021,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Strategic Research Council; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; National Research Foundation; Sino-Danish Center; Danmarks Grundforskningsfond; National Science Foundation","keywords":"Cluster analysis; Computer science; Generalization; CURE data clustering algorithm; Correlation clustering; Data mining; Single-linkage clustering; Artificial intelligence; Consensus clustering; Hierarchical clustering; Canopy clustering algorithm; Fuzzy clustering; Selection (genetic algorithm); Pattern recognition (psychology); Mathematics","score_opus":0.02217365358364736,"score_gpt":0.2965578701439754,"score_spread":0.274384216560328,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285364251","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006223437,0.0005139405,0.98748624,0.00027596075,0.00011516159,0.00014715463,0.0003088775,0.00024749633,0.004681703],"genre_scores_gemma":[0.210643,0.0014089036,0.7695887,0.00068069337,0.00044076992,0.0006161916,0.0027710923,0.00052823,0.0133224055],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99009013,0.003124344,0.00049059023,0.0026886615,0.0031137397,0.0004924854],"domain_scores_gemma":[0.9866412,0.0046974183,0.001156429,0.0043661203,0.0027319784,0.00040680243],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065275505,0.0015425611,0.001478994,0.0039089597,0.0014959166,0.003878735,0.004768128,0.0018914595,0.0056765843],"category_scores_gemma":[0.025475137,0.0007257585,0.0015934358,0.0052352794,0.0019789857,0.0057374993,0.0040991777,0.0019805206,0.002632196],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018063009,0.00015457362,0.0027059899,0.0006249604,0.0004388449,0.0001612359,0.0003756223,0.18948033,0.004812484,0.54479784,0.017118398,0.2391491],"study_design_scores_gemma":[0.000031164353,0.00009946995,0.00089665817,0.00008124193,0.00008993804,0.00024027166,0.00015125194,0.52855366,0.0040692524,0.43383598,0.03190219,0.000048947775],"about_ca_topic_score_codex":0.0024083997,"about_ca_topic_score_gemma":0.0033888444,"teacher_disagreement_score":0.0065275505,"about_ca_system_score_codex":0.001988338,"about_ca_system_score_gemma":0.0019930492,"threshold_uncertainty_score":0.03452146},"labels":[],"label_agreement":null},{"id":"W4285464664","doi":"10.32920/ryerson.14661465.v1","title":"Kernel k-MACE: hypercube unsupervised clustering method","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; University of Toronto","funders":"","keywords":"Cluster analysis; Kernel (algebra); Pattern recognition (psychology); Computer science; Artificial intelligence; Kernel method; Initialization; Kernel embedding of distributions; Feature vector; Mace; Variable kernel density estimation; Mathematics; Support vector machine; Combinatorics","score_opus":0.0463775866907028,"score_gpt":0.3578731729497401,"score_spread":0.3114955862590373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285464664","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019232426,0.00018549289,0.9961661,0.00009882346,0.000037025096,0.00007733586,0.000108486565,0.0006501373,0.0007532873],"genre_scores_gemma":[0.093520924,0.0002859973,0.89893126,0.00023170818,0.00008538761,0.00073937816,0.0010950877,0.0004739115,0.004636334],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99717915,0.0011645763,0.00015063732,0.00062146405,0.000750656,0.00013352768],"domain_scores_gemma":[0.99748343,0.00078934856,0.0001975614,0.00055349746,0.0008840815,0.000092114795],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022759647,0.0010449219,0.0018528363,0.001984247,0.0011459488,0.0015661438,0.003051281,0.001811291,0.003725919],"category_scores_gemma":[0.007206761,0.0006844355,0.0016064988,0.002559613,0.0009874932,0.0019345576,0.0024400835,0.0021184254,0.002543226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031307587,0.0001273818,0.0013279476,0.0003569545,0.00038125739,0.00014019056,0.00032880926,0.56922734,0.007284055,0.04242767,0.017847857,0.3602375],"study_design_scores_gemma":[0.000014446242,0.000017756749,0.00023340149,0.000012957669,0.000008972169,0.00006916317,0.000025967971,0.98060834,0.0021771537,0.012381157,0.004428669,0.000022029653],"about_ca_topic_score_codex":0.005193781,"about_ca_topic_score_gemma":0.005115139,"teacher_disagreement_score":0.005193781,"about_ca_system_score_codex":0.0011162391,"about_ca_system_score_gemma":0.0022887199,"threshold_uncertainty_score":0.012464404},"labels":[],"label_agreement":null},{"id":"W4285781363","doi":"10.1002/widm.1343","title":"Density‐based clustering","year":2019,"lang":"en","type":"article","venue":"Wiley Interdisciplinary Reviews Data Mining and Knowledge Discovery","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":116,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Cluster analysis; Artificial intelligence","score_opus":0.06673601290054507,"score_gpt":0.3667671650858934,"score_spread":0.30003115218534837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285781363","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040998016,0.0026688452,0.98237497,0.00070114725,0.00016996235,0.00019555364,0.0010053081,0.0009771847,0.0078072967],"genre_scores_gemma":[0.20043573,0.007931127,0.7709983,0.00074840675,0.0005505387,0.0006167406,0.0065507623,0.0009230143,0.011245314],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99492633,0.0012834019,0.0003007831,0.0012619956,0.0019757403,0.0002517191],"domain_scores_gemma":[0.99357325,0.0022821906,0.00053903507,0.0012189341,0.0022264374,0.00016009365],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038384811,0.0012923804,0.0022087302,0.0072391215,0.0019355798,0.0052343654,0.0032644598,0.002048587,0.005479062],"category_scores_gemma":[0.016204963,0.00097157236,0.0018768721,0.008456771,0.0020686453,0.004136845,0.0035055035,0.0022255648,0.0045903563],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014081207,0.000099331286,0.007816515,0.0012927133,0.0005144815,0.00024788562,0.0010633135,0.19869708,0.004724952,0.3514377,0.05352438,0.38044095],"study_design_scores_gemma":[0.000026631833,0.000044436092,0.004098123,0.00037599553,0.00010593246,0.00045592512,0.00046996865,0.58124644,0.004608578,0.31401497,0.094393164,0.00015974484],"about_ca_topic_score_codex":0.008484232,"about_ca_topic_score_gemma":0.0060327016,"teacher_disagreement_score":0.008484232,"about_ca_system_score_codex":0.0031001726,"about_ca_system_score_gemma":0.0029888065,"threshold_uncertainty_score":0.022493422},"labels":[],"label_agreement":null},{"id":"W4286239843","doi":"10.5281/zenodo.6866942","title":"Datasets for Data-Centric Classification and Clustering","year":2022,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Data science; Artificial intelligence","score_opus":0.09582514426084024,"score_gpt":0.30527602719829605,"score_spread":0.2094508829374558,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4286239843","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036620302,0.0010862603,0.014742072,0.0008141314,0.0007126522,0.0009660902,0.95795214,0.015308066,0.0047565443],"genre_scores_gemma":[0.003325475,0.0002447498,0.022516048,0.00023157975,0.000048442,0.0014809031,0.9701351,0.0005680025,0.0014496369],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9944171,0.00081961113,0.0010652934,0.0014875768,0.0017620999,0.0004482374],"domain_scores_gemma":[0.9930214,0.0011435191,0.0004369602,0.0025959234,0.0023942895,0.00040789184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039028393,0.0036473747,0.0020239132,0.005892353,0.0019708327,0.0029707993,0.005452669,0.0033196928,0.022001954],"category_scores_gemma":[0.014373788,0.00069773215,0.0035980358,0.007549184,0.00096406345,0.0021357222,0.0038296175,0.0038730502,0.044005986],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038702646,0.00033772312,0.0015714888,0.0015211409,0.00013398787,0.000112655995,0.00008479498,0.0052266517,0.0014933058,0.002641395,0.94707865,0.039411124],"study_design_scores_gemma":[0.00042451167,0.00017685801,0.0079112435,0.00059648074,0.0000971281,0.00043173085,0.00036363496,0.021090817,0.0064835143,0.013892089,0.94831556,0.00021638843],"about_ca_topic_score_codex":0.010810929,"about_ca_topic_score_gemma":0.017034328,"teacher_disagreement_score":0.022001954,"about_ca_system_score_codex":0.0024672153,"about_ca_system_score_gemma":0.0032685646,"threshold_uncertainty_score":0.07360387},"labels":[],"label_agreement":null},{"id":"W4292397078","doi":"10.1007/s11222-022-10128-9","title":"A joint latent factor analyzer and functional subspace model for clustering multivariate functional data","year":2022,"lang":"en","type":"article","venue":"Statistics and Computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Functional principal component analysis; Principal component analysis; Multivariate statistics; Univariate; Cluster analysis; Subspace topology; Initialization; Projection pursuit; Mathematics; Computer science; Model selection; Multivariate normal distribution; Artificial intelligence; Pattern recognition (psychology); Statistics","score_opus":0.1475347888853481,"score_gpt":0.3320264717340741,"score_spread":0.18449168284872602,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4292397078","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015989662,0.000035533223,0.9979055,0.000030697876,0.000007507086,0.000014862868,0.0000578118,0.00020723055,0.00014184257],"genre_scores_gemma":[0.1356281,0.0002241743,0.8594215,0.00010677337,0.000064318716,0.00029618156,0.0011349447,0.00043742775,0.00268651],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968141,0.0017295881,0.00013571272,0.0005666984,0.0005799818,0.00017392155],"domain_scores_gemma":[0.9968907,0.0011851392,0.00024518138,0.00065662264,0.0008573592,0.00016509056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047442834,0.0011412264,0.0017331013,0.001969611,0.0010521205,0.0021432687,0.0022158972,0.0013575429,0.002815341],"category_scores_gemma":[0.008414732,0.0007055969,0.0020844382,0.003036797,0.0012705439,0.0031563863,0.0024826606,0.0022541948,0.0018726046],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058926776,0.00036073572,0.0040140557,0.0002534411,0.00052144006,0.00016323566,0.000518252,0.35083678,0.011664643,0.18342656,0.008476183,0.43917537],"study_design_scores_gemma":[0.000008293217,0.00003644102,0.0003350485,0.000010913701,0.000021541966,0.000044620756,0.00003301689,0.9654746,0.00054514204,0.03260608,0.00086028373,0.00002408703],"about_ca_topic_score_codex":0.0060343603,"about_ca_topic_score_gemma":0.007377584,"teacher_disagreement_score":0.0060343603,"about_ca_system_score_codex":0.0008215189,"about_ca_system_score_gemma":0.0026335532,"threshold_uncertainty_score":0.025090456},"labels":[],"label_agreement":null},{"id":"W4292585402","doi":"10.48550/arxiv.1607.06187","title":"Exploring Differences in Interpretation of Words Essential in Medical\\n Expert-Patient Communication","year":2016,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Jaccard index; Context (archaeology); Interpretation (philosophy); Quality of life (healthcare); Sample (material); Medicine; Psychology; Computer science; Artificial intelligence; Nursing","score_opus":0.14636189583854994,"score_gpt":0.2538733988732508,"score_spread":0.10751150303470083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4292585402","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96072215,0.00033927214,0.02908329,0.00078164676,0.000071262555,0.00020110505,0.00037437855,0.00005552024,0.008371296],"genre_scores_gemma":[0.9907025,0.00013455804,0.008097662,0.00017170563,0.000015716709,0.00012919112,0.00021656229,0.000025371666,0.0005068092],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.97356164,0.018733123,0.0017199374,0.0016275882,0.0037788735,0.0005788081],"domain_scores_gemma":[0.8316432,0.14380331,0.01186213,0.0048400946,0.0071416837,0.0007095581],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016180601,0.00042344193,0.00039526608,0.0018886557,0.00091409,0.004946789,0.0007908153,0.001115544,0.0032218655],"category_scores_gemma":[0.16356893,0.00024193831,0.00053221994,0.001691134,0.0026806898,0.0038341968,0.002502168,0.0012075418,0.0006038875],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022801778,0.00041427402,0.20981123,0.0020205802,0.00049459835,0.0007781618,0.5266379,0.008103913,0.015151621,0.037695874,0.0033468497,0.19326477],"study_design_scores_gemma":[0.00025993213,0.0018708422,0.39158466,0.0017472776,0.0005489932,0.0019514703,0.37152168,0.06865117,0.0140794795,0.10589847,0.041079465,0.0008065768],"about_ca_topic_score_codex":0.002580159,"about_ca_topic_score_gemma":0.00206503,"teacher_disagreement_score":0.016180601,"about_ca_system_score_codex":0.0020613687,"about_ca_system_score_gemma":0.0012947188,"threshold_uncertainty_score":0.08557224},"labels":[],"label_agreement":null},{"id":"W4293863504","doi":"10.1109/siu55565.2022.9864873","title":"Semantic Segmentation with the Mixup Data Augmentation Method","year":2022,"lang":"en","type":"article","venue":"2022 30th Signal Processing and Communications Applications Conference (SIU)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Stantec (Canada)","funders":"","keywords":"Computer science; Segmentation; Artificial intelligence; Image segmentation; Computer vision; Natural language processing","score_opus":0.07743756159890856,"score_gpt":0.37187996519312605,"score_spread":0.29444240359421747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293863504","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028649747,0.0002758023,0.96418965,0.00016197444,0.00010072167,0.0001613896,0.00033874618,0.0040261606,0.0020958742],"genre_scores_gemma":[0.15927455,0.00023149628,0.835565,0.00014238319,0.000052961354,0.00027742411,0.0010163112,0.00051512447,0.002924794],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987723,0.00023166802,0.00009807833,0.00037792537,0.00042110393,0.00009881457],"domain_scores_gemma":[0.9982596,0.00048389446,0.00014787787,0.00059342635,0.00045461667,0.000060547147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018410637,0.001197481,0.0009055449,0.002097548,0.00060084264,0.0018761891,0.001376195,0.0012223148,0.0038542135],"category_scores_gemma":[0.0034638152,0.00065783923,0.0015274514,0.0014443136,0.000837376,0.0024634427,0.0025716985,0.0013607051,0.001787672],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000722288,0.00019872223,0.0027239644,0.0003617567,0.00019817227,0.00022417856,0.0004619137,0.05321391,0.103467286,0.008642747,0.0048593413,0.8249257],"study_design_scores_gemma":[0.000035406203,0.0003090171,0.0020311421,0.000043951422,0.00011754743,0.0005088465,0.00021095977,0.8129612,0.15693858,0.007022051,0.01973891,0.000082313716],"about_ca_topic_score_codex":0.0015268167,"about_ca_topic_score_gemma":0.00261331,"teacher_disagreement_score":0.0038542135,"about_ca_system_score_codex":0.00061691925,"about_ca_system_score_gemma":0.00091258675,"threshold_uncertainty_score":0.012893617},"labels":[],"label_agreement":null},{"id":"W4294311511","doi":"10.1109/tetci.2022.3201620","title":"Viewpoint-Based Kernel Fuzzy Clustering With Weight Information Granules","year":2022,"lang":"en","type":"article","venue":"IEEE Transactions on Emerging Topics in Computational Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":71,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Anhui Provincial Key Research and Development Plan; Natural Science Foundation of Anhui Province; Central University Basic Research Fund of China; National Natural Science Foundation of China","keywords":"Cluster analysis; Fuzzy clustering; Mathematics; Artificial intelligence; Data mining; Pattern recognition (psychology); FLAME clustering; Correlation clustering; CURE data clustering algorithm; Computer science","score_opus":0.025033645151595774,"score_gpt":0.2962129579615027,"score_spread":0.2711793128099069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294311511","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019702217,0.00016938134,0.9789505,0.000048612408,0.000017637958,0.000032763626,0.00003455127,0.00016189122,0.00088249124],"genre_scores_gemma":[0.5978384,0.0004597622,0.399384,0.000055684068,0.000052288895,0.00014859326,0.0003904661,0.000096846255,0.0015739626],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990783,0.00015155811,0.000061484294,0.0002577139,0.0003676947,0.000083339524],"domain_scores_gemma":[0.9991061,0.00022910125,0.00010401023,0.00013733463,0.00038090887,0.000042504584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000880613,0.0008190622,0.0010789457,0.001431293,0.0005362485,0.0014151885,0.0012754143,0.0008182704,0.0007562274],"category_scores_gemma":[0.003564575,0.00036460377,0.0012386244,0.0015041785,0.00074520934,0.002006944,0.0012905329,0.00075660227,0.00027878108],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029535757,0.00007597348,0.0029051725,0.00026363158,0.00016709186,0.00018470407,0.0005725664,0.5872093,0.018107044,0.058867365,0.0023229811,0.32902887],"study_design_scores_gemma":[0.000010958152,0.000029147424,0.0005152081,0.0000087780745,0.000017818329,0.000046513578,0.000035758996,0.9869528,0.0023642583,0.0092798555,0.0007185021,0.000020527676],"about_ca_topic_score_codex":0.0055828732,"about_ca_topic_score_gemma":0.0031177846,"teacher_disagreement_score":0.0055828732,"about_ca_system_score_codex":0.0010983187,"about_ca_system_score_gemma":0.0009318774,"threshold_uncertainty_score":0.011100769},"labels":[],"label_agreement":null},{"id":"W4308319351","doi":"10.1007/s11721-022-00219-8","title":"Blending multiple algorithmic granular components: a recipe for clustering","year":2022,"lang":"en","type":"article","venue":"Swarm Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Cluster analysis; Computer science; DBSCAN; Algorithm; Swarm intelligence; Canopy clustering algorithm; Correlation clustering; Data mining; Artificial intelligence; Particle swarm optimization","score_opus":0.05911746107355961,"score_gpt":0.31609172751656583,"score_spread":0.25697426644300625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308319351","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024461057,0.000049316346,0.9942967,0.000094334915,0.00004645593,0.000053265692,0.000024848903,0.00019402047,0.00279497],"genre_scores_gemma":[0.07019589,0.00013809784,0.9246031,0.00010191935,0.000046522513,0.00018186614,0.000080971826,0.00025755135,0.0043940004],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994722,0.00009915409,0.000039315524,0.000100418794,0.00023903075,0.000049937174],"domain_scores_gemma":[0.9991491,0.00025570576,0.00005365039,0.00025960646,0.00021551998,0.00006655779],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010835233,0.0007270932,0.0010894664,0.001339577,0.0012975644,0.0022651958,0.002160482,0.0014736251,0.0070921225],"category_scores_gemma":[0.0040768017,0.00079106964,0.0012360711,0.0019147971,0.0014524609,0.002484585,0.0023644194,0.002398583,0.001869534],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009961803,0.000100420984,0.0008402383,0.00028069512,0.00011160181,0.0001566806,0.00043650268,0.33277145,0.013792608,0.44780695,0.006096996,0.19750631],"study_design_scores_gemma":[0.000019274094,0.000042599404,0.00032279312,0.000031440024,0.000029890618,0.000088542285,0.00009658163,0.78050125,0.0026375724,0.20725894,0.008936379,0.000034833338],"about_ca_topic_score_codex":0.003155446,"about_ca_topic_score_gemma":0.0042774775,"teacher_disagreement_score":0.0070921225,"about_ca_system_score_codex":0.00084059895,"about_ca_system_score_gemma":0.0011682615,"threshold_uncertainty_score":0.02372551},"labels":[],"label_agreement":null},{"id":"W4310371105","doi":"10.3390/e24121749","title":"Nonparametric Clustering of Mixed Data Using Modified Chi-Squared Tests","year":2022,"lang":"en","type":"article","venue":"Entropy","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Nonparametric statistics; Cluster analysis; Benchmark (surveying); Computer science; Sample space; Data mining; Cluster (spacecraft); Algorithm; Continuous variable; Parametric statistics; Mathematics; Hierarchical clustering; Product (mathematics); Statistics; Artificial intelligence","score_opus":0.09565957577078021,"score_gpt":0.3471802942440409,"score_spread":0.2515207184732607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310371105","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010702391,0.00011973884,0.9881467,0.000058824004,0.000041297844,0.000077097226,0.000072300274,0.0004965153,0.00028510517],"genre_scores_gemma":[0.30260503,0.00013601738,0.69449437,0.00010523718,0.00013506835,0.0005268618,0.00083703356,0.0002880948,0.0008722409],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9835322,0.008617476,0.0009973123,0.002603249,0.0039093336,0.00034048225],"domain_scores_gemma":[0.95902795,0.027670257,0.0030565907,0.0049100444,0.004782185,0.00055292546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014528259,0.001195958,0.0023110611,0.0051001236,0.0010399717,0.0023700465,0.0039287754,0.0014458424,0.002627326],"category_scores_gemma":[0.059999377,0.0006500971,0.0018428575,0.0042270008,0.002449954,0.0024960628,0.002559148,0.0021236087,0.00095871737],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015872696,0.0004094957,0.030755179,0.0007021796,0.001739946,0.00053401187,0.0007535759,0.15431601,0.01712043,0.05568798,0.0044016144,0.73199236],"study_design_scores_gemma":[0.00009456539,0.00038057024,0.0095300805,0.000045913897,0.00009883411,0.00034122705,0.00013001217,0.9399052,0.0068429294,0.039736316,0.0027785355,0.000115822615],"about_ca_topic_score_codex":0.0016055569,"about_ca_topic_score_gemma":0.0016484226,"teacher_disagreement_score":0.014528259,"about_ca_system_score_codex":0.0010273746,"about_ca_system_score_gemma":0.0017769758,"threshold_uncertainty_score":0.076833725},"labels":[],"label_agreement":null},{"id":"W4312723533","doi":"10.1109/access.2022.3228238","title":"Piecemeal Clustering: a Self-Driven Data Clustering Algorithm","year":2022,"lang":"en","type":"article","venue":"IEEE Access","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"Natural Sciences and Engineering Research Council of Canada; Memorial University of Newfoundland; Equinor","keywords":"Cluster analysis; CURE data clustering algorithm; Computer science; Canopy clustering algorithm; Correlation clustering; Data stream clustering; Fuzzy clustering; Data mining; Determining the number of clusters in a data set; Single-linkage clustering; Clustering high-dimensional data; Algorithm; Constrained clustering; Data set; Affinity propagation; Artificial intelligence","score_opus":0.09243514291304206,"score_gpt":0.3723603590328441,"score_spread":0.27992521611980203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312723533","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004733793,0.00022628004,0.9922982,0.00013734479,0.0000766673,0.0001420328,0.00010794327,0.001439418,0.00083828456],"genre_scores_gemma":[0.053207926,0.00016536255,0.94270015,0.00019970398,0.0000746245,0.00029191244,0.00068667735,0.0004160633,0.002257695],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970709,0.00057301717,0.000142101,0.0008293331,0.0012312421,0.00015327375],"domain_scores_gemma":[0.9972658,0.0005946855,0.00028202485,0.0006936789,0.0010460095,0.000117794065],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024125492,0.0014609158,0.0018541313,0.0043597175,0.0023940487,0.0018148058,0.0049895034,0.0028297715,0.0029793438],"category_scores_gemma":[0.0076236236,0.0008108875,0.0018233957,0.0045507858,0.001489165,0.002968573,0.0035040162,0.0020980083,0.0023463692],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040761582,0.00020003543,0.002919369,0.0003647863,0.0004143955,0.00017018754,0.0008821908,0.18787076,0.017166022,0.020040115,0.019908503,0.749656],"study_design_scores_gemma":[0.00004157165,0.000088470355,0.000792915,0.00003591059,0.000039123108,0.00027234547,0.00014106654,0.95690304,0.0074793734,0.016118582,0.0180004,0.00008725106],"about_ca_topic_score_codex":0.005333617,"about_ca_topic_score_gemma":0.0074947304,"teacher_disagreement_score":0.005333617,"about_ca_system_score_codex":0.0012407298,"about_ca_system_score_gemma":0.0017914507,"threshold_uncertainty_score":0.012758911},"labels":[],"label_agreement":null},{"id":"W4319653775","doi":"10.48550/arxiv.2302.03391","title":"Sparse and geometry-aware generalisation of the mutual information for joint discriminative clustering and feature selection","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre hospitalier de l'Université Laval","funders":"Agence Nationale de la Recherche","keywords":"Cluster analysis; Discriminative model; Feature selection; Pattern recognition (psychology); Computer science; Scalability; Selection (genetic algorithm); Feature (linguistics); Artificial intelligence; Relevance (law); Clustering high-dimensional data; Data mining; Machine learning; Mathematics","score_opus":0.12688043526973647,"score_gpt":0.22954914912027388,"score_spread":0.10266871385053741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4319653775","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0050524366,0.00014971272,0.9936186,0.00017632786,0.000016959815,0.000044879096,0.00006515513,0.00037687973,0.0004990608],"genre_scores_gemma":[0.23779039,0.00038387577,0.75634634,0.00044474265,0.00018914563,0.00039584062,0.0011653975,0.00043699756,0.0028472391],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9971545,0.0012708952,0.00011561578,0.0006567638,0.00064004894,0.00016219374],"domain_scores_gemma":[0.9961786,0.0020292727,0.00036046843,0.00073758414,0.0005002223,0.000193861],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004245994,0.001810529,0.0023215152,0.0019298604,0.00094890466,0.0015941095,0.0034112113,0.0025288453,0.0021710123],"category_scores_gemma":[0.013461909,0.0009818607,0.0018224221,0.002850422,0.0022585003,0.0027338073,0.0039661885,0.002813865,0.0012160578],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002365425,0.0001235115,0.0014029343,0.00022519076,0.00015496896,0.0001390925,0.000240533,0.7622107,0.008336188,0.050300986,0.005685898,0.1709434],"study_design_scores_gemma":[0.000015476713,0.000053689007,0.0003246931,0.000009885185,0.000010471749,0.00007658611,0.000019126252,0.973409,0.0013307084,0.02356977,0.0011631151,0.000017556793],"about_ca_topic_score_codex":0.002653602,"about_ca_topic_score_gemma":0.003924415,"teacher_disagreement_score":0.004245994,"about_ca_system_score_codex":0.0016089819,"about_ca_system_score_gemma":0.0012570752,"threshold_uncertainty_score":0.022455215},"labels":[],"label_agreement":null},{"id":"W4319660509","doi":"10.1016/j.eswa.2023.119655","title":"Rule-based fuzzy neural networks realized with the aid of linear function Prototype-driven fuzzy clustering and layer Reconstruction-based network design strategy","year":2023,"lang":"en","type":"article","venue":"Expert Systems with Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Fuzzy clustering; Fuzzy logic; Centroid; Computer science; Cluster analysis; Euclidean distance; Neuro-fuzzy; Artificial neural network; Data mining; Artificial intelligence; Fuzzy classification; Fuzzy number; Defuzzification; Mathematics; Fuzzy set; Pattern recognition (psychology); Fuzzy control system","score_opus":0.035932059205595356,"score_gpt":0.2865825160744831,"score_spread":0.25065045686888776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4319660509","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006775454,0.000079280944,0.9911945,0.00003822484,0.000025631514,0.000043537722,0.000016943659,0.00030761058,0.0015188567],"genre_scores_gemma":[0.45845714,0.00015534456,0.5388267,0.00006090201,0.000021487807,0.000199698,0.00008838605,0.000048152884,0.0021422417],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997291,0.0000625257,0.00002291181,0.00006646382,0.00009894717,0.000020086009],"domain_scores_gemma":[0.9996588,0.00007895216,0.00002901359,0.000044641245,0.00017742378,0.000011180414],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007501653,0.00046296732,0.0005193019,0.00028541996,0.00039259414,0.0007217907,0.0011746595,0.0008799512,0.0013575637],"category_scores_gemma":[0.0013937206,0.000308779,0.00048456283,0.00037274108,0.00033292247,0.00084595085,0.00037139343,0.00075861294,0.00048028736],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023480166,0.00011830135,0.0005384976,0.00027968874,0.000095671974,0.00012078706,0.0001920541,0.6463929,0.053527128,0.03493656,0.0020767802,0.26148695],"study_design_scores_gemma":[0.0000068326485,0.000026894506,0.00007762992,0.0000075885127,0.000012539341,0.000026118683,0.0000042922265,0.99277896,0.0050309347,0.0015249619,0.00049552857,0.000007764588],"about_ca_topic_score_codex":0.0031809125,"about_ca_topic_score_gemma":0.0038266699,"teacher_disagreement_score":0.0031809125,"about_ca_system_score_codex":0.00049699435,"about_ca_system_score_gemma":0.0007205338,"threshold_uncertainty_score":0.0063248277},"labels":[],"label_agreement":null},{"id":"W4320007135","doi":"10.18280/mmep.090622","title":"Block-Based K-Medoids Partitioning Method with Standardized Data to Improve Clustering Accuracy","year":2022,"lang":"en","type":"article","venue":"Mathematical Modelling and Engineering Problems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Medoid; Cluster analysis; Block (permutation group theory); Computer science; k-medoids; Data mining; Algorithm; Correlation clustering; CURE data clustering algorithm; Pattern recognition (psychology); Mathematics; Artificial intelligence","score_opus":0.055049099934319605,"score_gpt":0.304892631881267,"score_spread":0.24984353194694736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320007135","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008391249,0.00030294663,0.9897877,0.000064580636,0.000050285857,0.00012434242,0.00011098104,0.0005176233,0.0006501792],"genre_scores_gemma":[0.12486239,0.00039787963,0.8718176,0.00007844759,0.00004569964,0.0004527581,0.00072783325,0.0002819338,0.001335557],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99653924,0.0010665258,0.00038288743,0.0008185561,0.0010198908,0.00017294254],"domain_scores_gemma":[0.9958788,0.0014734359,0.00030586825,0.0005273883,0.0017273509,0.00008718275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003350323,0.0015533284,0.0020187236,0.0031640518,0.0013271007,0.0016395343,0.0017943573,0.0012756026,0.0024040549],"category_scores_gemma":[0.01212861,0.0007355427,0.0017344614,0.003636073,0.0008462814,0.0024515162,0.0018841282,0.001398613,0.0015469755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005416876,0.00014265995,0.0054257773,0.0009258314,0.0004061803,0.00015310581,0.0016284644,0.374593,0.02654128,0.018549992,0.008583622,0.5625084],"study_design_scores_gemma":[0.000053997668,0.00010678394,0.0015308576,0.000058677168,0.000087030334,0.00016988761,0.00025385054,0.96942174,0.010838458,0.008802349,0.0085973805,0.00007893391],"about_ca_topic_score_codex":0.008930439,"about_ca_topic_score_gemma":0.006982524,"teacher_disagreement_score":0.008930439,"about_ca_system_score_codex":0.00095767796,"about_ca_system_score_gemma":0.002089127,"threshold_uncertainty_score":0.017756939},"labels":[],"label_agreement":null},{"id":"W4323896842","doi":"10.1109/tbdata.2023.3255003","title":"Approximate Clustering Ensemble Method for Big Data","year":2023,"lang":"en","type":"article","venue":"IEEE Transactions on Big Data","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Disjoint sets; Big data; Correlation clustering; Data mining; Component (thermodynamics); Cluster (spacecraft); Artificial intelligence; Mathematics; Combinatorics","score_opus":0.3223194882196442,"score_gpt":0.40855782386772044,"score_spread":0.08623833564807626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323896842","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031196969,0.0003664529,0.9942153,0.000123945,0.0000798064,0.00004489962,0.00013835046,0.0009362323,0.00097533205],"genre_scores_gemma":[0.21615851,0.00082984153,0.7743323,0.0003058753,0.00036167452,0.00047192012,0.00200196,0.0005813942,0.0049564647],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977398,0.0005044582,0.000100516416,0.00056045054,0.0009542068,0.00014058096],"domain_scores_gemma":[0.9975706,0.0006663449,0.00017333413,0.0006122164,0.00084850995,0.00012896463],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00227493,0.0012398384,0.00194772,0.0020161567,0.0011663318,0.0016603841,0.0028685578,0.0013274091,0.002977135],"category_scores_gemma":[0.0063682,0.00047465312,0.0015429836,0.00296577,0.0006738655,0.0028853086,0.00205053,0.0021511079,0.0014916344],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022873803,0.00008809231,0.0020339806,0.00020360801,0.00031967097,0.00013896347,0.00018597495,0.70526505,0.00422882,0.047043182,0.011025988,0.22923793],"study_design_scores_gemma":[0.000005956249,0.000013225837,0.00018107108,0.000005652746,0.000012451001,0.000021358972,0.000017002018,0.9845483,0.0007616364,0.012343407,0.0020805753,0.000009413166],"about_ca_topic_score_codex":0.006591692,"about_ca_topic_score_gemma":0.006989103,"teacher_disagreement_score":0.006591692,"about_ca_system_score_codex":0.0013264163,"about_ca_system_score_gemma":0.0014711154,"threshold_uncertainty_score":0.013106644},"labels":[],"label_agreement":null},{"id":"W4360979800","doi":"10.1007/s00453-023-01113-3","title":"Preface to the Special Issue on the 17th Algorithms and Data Structures Symposium (WADS 2021)","year":2023,"lang":"en","type":"article","venue":"Algorithmica","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; University of Waterloo; Dalhousie University","funders":"Division of Mathematical Sciences; Dalhousie University; Fields Institute for Research in Mathematical Sciences; Pacific Institute for the Mathematical Sciences","keywords":"Theory of computation; Computer science; Algorithm; Data science","score_opus":0.04072615141270493,"score_gpt":0.3320292424290909,"score_spread":0.29130309101638596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4360979800","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00032557358,0.010144954,0.004627579,0.024650797,0.9145392,0.00018291277,0.0017127048,0.00067945075,0.04313686],"genre_scores_gemma":[0.001967545,0.00942934,0.0026983735,0.00873963,0.5279624,0.00024579573,0.004214131,0.0017155226,0.44302732],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978555,0.0002769149,0.0001496282,0.000353622,0.001156894,0.0002073277],"domain_scores_gemma":[0.9850809,0.0023851574,0.0007131961,0.0008593678,0.0076501537,0.003311318],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0027372455,0.0025703264,0.0033503168,0.004240115,0.0024078763,0.008073142,0.0021874905,0.0030926396,0.31503662],"category_scores_gemma":[0.009798291,0.00083118875,0.0016598942,0.003042334,0.00066996313,0.0043913005,0.0028522413,0.0055895355,0.2416478],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022233236,0.000017164888,0.000021836002,0.00005419171,0.0000037173754,0.000009250337,0.000002794254,0.000040480794,0.00008338017,0.00023936629,0.9886177,0.010887858],"study_design_scores_gemma":[0.000025793981,0.00006412334,0.00047007963,0.00015273519,0.000016872651,0.00005554401,0.000020551874,0.0004530306,0.00017324272,0.0020490014,0.99650264,0.00001629957],"about_ca_topic_score_codex":0.0020622832,"about_ca_topic_score_gemma":0.003921583,"teacher_disagreement_score":0.31503662,"about_ca_system_score_codex":0.0021996326,"about_ca_system_score_gemma":0.0023200258,"threshold_uncertainty_score":0.9770171},"labels":[],"label_agreement":null},{"id":"W4375858861","doi":"10.32473/flairs.36.133230","title":"Biogeography-based optimization for feature selection","year":2023,"lang":"en","type":"article","venue":"Proceedings of the ... International Florida Artificial Intelligence Research Society Conference","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Data mining; Computer science; Cluster (spacecraft); Selection (genetic algorithm); Feature selection; Biogeography; Feature (linguistics); Machine learning; Artificial intelligence; Ecology; Biology","score_opus":0.13065043082825348,"score_gpt":0.38871743987603846,"score_spread":0.258067009047785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4375858861","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012049184,0.00046178387,0.9851564,0.00029735273,0.00005057819,0.000071627976,0.000080937716,0.00023454997,0.0015975711],"genre_scores_gemma":[0.3818566,0.00055143575,0.61257064,0.00043218237,0.00009682309,0.0005622531,0.00059070246,0.00020314501,0.003136313],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99934345,0.00023015613,0.00004246743,0.00014932505,0.00016335634,0.00007128374],"domain_scores_gemma":[0.9993912,0.0003096936,0.00006690935,0.000037929563,0.0001615034,0.00003262245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012825009,0.0010020752,0.0011973497,0.0011419272,0.00052371627,0.00076372124,0.0010212512,0.0009857086,0.0021665897],"category_scores_gemma":[0.0029782958,0.0004384042,0.00088660244,0.001490578,0.0007142761,0.0009323495,0.00084160117,0.00080134376,0.00048269186],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000092547605,0.00008603003,0.0024392852,0.00015636685,0.000112409,0.00011904684,0.000060622293,0.8522724,0.0058343555,0.011314773,0.003376202,0.124135934],"study_design_scores_gemma":[0.000017411665,0.000026108508,0.00041084774,0.000008871652,0.0000091724505,0.000028603,0.000009166749,0.99438924,0.0006550921,0.0034964478,0.0009422797,0.0000067588144],"about_ca_topic_score_codex":0.004385454,"about_ca_topic_score_gemma":0.0032261638,"teacher_disagreement_score":0.004385454,"about_ca_system_score_codex":0.0007668819,"about_ca_system_score_gemma":0.0013442988,"threshold_uncertainty_score":0.008719802},"labels":[],"label_agreement":null},{"id":"W4375869758","doi":"10.48550/arxiv.1907.01136","title":"Finding Outliers in Gaussian Model-Based Clustering","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Outlier; Cluster analysis; Gaussian; Computer science; Mixture model; Pattern recognition (psychology); Artificial intelligence; Statistics; Econometrics; Mathematics; Data mining; Physics","score_opus":0.10265674291270344,"score_gpt":0.2399083167034491,"score_spread":0.13725157379074565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4375869758","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006954926,0.00078971393,0.9906792,0.00023872551,0.00004967651,0.00003704171,0.00010025174,0.00066006265,0.00049033586],"genre_scores_gemma":[0.29811117,0.00192963,0.69461155,0.00036499483,0.00026457923,0.00028577825,0.001594755,0.0006096341,0.0022278752],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9930714,0.0024827183,0.00032756556,0.0012501894,0.0025785111,0.00028958716],"domain_scores_gemma":[0.9891166,0.00572969,0.0010598092,0.0016825012,0.0021049557,0.0003064606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070186346,0.0012461779,0.0025238958,0.00398581,0.0018353505,0.0029181521,0.0035121976,0.0029796173,0.0008766217],"category_scores_gemma":[0.0368637,0.0011181937,0.0013010249,0.004737682,0.0027494782,0.0029914093,0.003622176,0.0030417223,0.0009520982],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023974245,0.00007879166,0.00715805,0.00041822015,0.00026505406,0.00020768736,0.0007883848,0.7065602,0.002189224,0.08123506,0.008992808,0.19186682],"study_design_scores_gemma":[0.000009470974,0.000023501523,0.0006753136,0.000044426855,0.000016592461,0.000076415854,0.00010425135,0.9173184,0.00090780476,0.078199156,0.0025941555,0.000030509853],"about_ca_topic_score_codex":0.010620437,"about_ca_topic_score_gemma":0.008371744,"teacher_disagreement_score":0.010620437,"about_ca_system_score_codex":0.002133963,"about_ca_system_score_gemma":0.0019133065,"threshold_uncertainty_score":0.037118554},"labels":[],"label_agreement":null},{"id":"W4377249775","doi":"10.1007/978-3-031-33271-5_19","title":"Interpretable Clustering via Soft Clustering Trees","year":2023,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster analysis; Computer science; Interpretability; Constrained clustering; Correlation clustering; Artificial intelligence; Data mining; CURE data clustering algorithm; Fuzzy clustering; Brown clustering; Conceptual clustering; Machine learning; Consensus clustering; Partition (number theory); Pattern recognition (psychology); Mathematics","score_opus":0.025300078755108986,"score_gpt":0.2867354810263392,"score_spread":0.26143540227123024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377249775","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021251624,0.00021697978,0.9933739,0.00016571463,0.000056231213,0.00002194193,0.00015630959,0.000652561,0.003231143],"genre_scores_gemma":[0.11586852,0.0007149776,0.8716701,0.00027095698,0.00019283814,0.00016610659,0.0014234976,0.0008967495,0.00879622],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985275,0.0005170314,0.000100792444,0.00030253036,0.00048613356,0.00006590409],"domain_scores_gemma":[0.9968323,0.0017410546,0.00018282273,0.00076764874,0.00039961108,0.000076482705],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001378679,0.00085179805,0.0009009543,0.0017681733,0.00075216463,0.0025217612,0.0016795932,0.0011159026,0.008318464],"category_scores_gemma":[0.0071922457,0.0008123353,0.001288866,0.0022457542,0.0014751619,0.0037290973,0.0029643883,0.002834754,0.0027246303],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017199814,0.00004901236,0.00038246074,0.00047650674,0.000089443696,0.0002322377,0.0005959786,0.06875112,0.0071509047,0.5723922,0.01579959,0.33390853],"study_design_scores_gemma":[0.000011343442,0.000015110064,0.00016124704,0.000056518074,0.000019231766,0.00010952972,0.0000788097,0.29876444,0.002064763,0.6895245,0.009176516,0.000017999122],"about_ca_topic_score_codex":0.0008100781,"about_ca_topic_score_gemma":0.0014353209,"teacher_disagreement_score":0.008318464,"about_ca_system_score_codex":0.0008479406,"about_ca_system_score_gemma":0.00049981324,"threshold_uncertainty_score":0.027827978},"labels":[],"label_agreement":null},{"id":"W4378087396","doi":"10.3390/app13106342","title":"VSFCM: A Novel Viewpoint-Driven Subspace Fuzzy C-Means Algorithm","year":2023,"lang":"en","type":"article","venue":"Applied Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Anhui Provincial Key Research and Development Plan; National Key Research and Development Program of China; Natural Science Foundation of Anhui Province; National Natural Science Foundation of China","keywords":"Initialization; Cluster analysis; Computer science; Subspace topology; Fuzzy clustering; Data mining; Pattern recognition (psychology); Correlation clustering; Algorithm; Artificial intelligence; Fuzzy logic","score_opus":0.04843435855030365,"score_gpt":0.31999287265577925,"score_spread":0.2715585141054756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378087396","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0056220735,0.00021601653,0.9924947,0.000074284515,0.000047416754,0.00005782636,0.000068357585,0.0005137631,0.0009054907],"genre_scores_gemma":[0.11652115,0.00027124118,0.88000536,0.00015726256,0.000060830687,0.00025460153,0.00051438547,0.00016663005,0.0020485618],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99905163,0.0001580199,0.000040335926,0.00020748696,0.00045320834,0.00008928423],"domain_scores_gemma":[0.999413,0.00011151701,0.00004153014,0.00006981958,0.00032540187,0.00003868876],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008991782,0.0011135908,0.0012499599,0.0016379039,0.0011106774,0.0010420599,0.0023357759,0.0015329205,0.0017703326],"category_scores_gemma":[0.0022599236,0.00048285665,0.0013937465,0.0017103407,0.0006833171,0.0011788472,0.0015331395,0.0013883329,0.0008339241],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022122245,0.00007598621,0.0013818006,0.00013505168,0.00013987004,0.000087772845,0.00021333025,0.2913458,0.012895304,0.012889663,0.007176013,0.67343825],"study_design_scores_gemma":[0.000017592754,0.00003544141,0.0002560147,0.000009969821,0.000010487036,0.000048546444,0.000018972209,0.99140596,0.002341432,0.0035383967,0.0022952713,0.000021932552],"about_ca_topic_score_codex":0.015802171,"about_ca_topic_score_gemma":0.012649616,"teacher_disagreement_score":0.015802171,"about_ca_system_score_codex":0.0011161312,"about_ca_system_score_gemma":0.0024527304,"threshold_uncertainty_score":0.03142041},"labels":[],"label_agreement":null},{"id":"W4384438355","doi":"10.1007/s10489-023-04827-7","title":"A three-way adaptive density peak clustering (3W-ADPC) method","year":2023,"lang":"en","type":"article","venue":"Applied Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Computer science; Cluster (spacecraft); Process (computing); Algorithm; Data mining; Boundary (topology); Correlation clustering; Artificial intelligence; Mathematics","score_opus":0.06696534723128361,"score_gpt":0.3432592515649392,"score_spread":0.2762939043336556,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384438355","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019303587,0.00011668684,0.9961111,0.00006890514,0.00008138562,0.000053515014,0.000072628776,0.0005415593,0.0010238736],"genre_scores_gemma":[0.053766754,0.00016776547,0.94160706,0.00011806724,0.000050564897,0.0001943971,0.0003688831,0.00030860913,0.0034178735],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987143,0.0002929634,0.00006407491,0.00027027223,0.0005602177,0.00009827486],"domain_scores_gemma":[0.9986941,0.00024026704,0.000059790775,0.00029844994,0.00063663005,0.000070838665],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011945616,0.0007893047,0.001120301,0.0018810128,0.0015675424,0.0015722563,0.0034395924,0.0017841732,0.0055764047],"category_scores_gemma":[0.0029238795,0.00057360297,0.0012836056,0.0029570311,0.00089522364,0.0013514735,0.0022274707,0.0017428087,0.0031859798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002571022,0.00015561805,0.0011983599,0.00042296047,0.00019648716,0.00015897665,0.00024141381,0.17705116,0.0212176,0.0470886,0.015211494,0.7368003],"study_design_scores_gemma":[0.00001995965,0.00002657362,0.00032461717,0.000014412681,0.000025643614,0.00011807925,0.00003726163,0.9774162,0.0052959337,0.009221688,0.0074611595,0.000038501978],"about_ca_topic_score_codex":0.00810193,"about_ca_topic_score_gemma":0.0086679105,"teacher_disagreement_score":0.00810193,"about_ca_system_score_codex":0.0007244867,"about_ca_system_score_gemma":0.0026899981,"threshold_uncertainty_score":0.018654943},"labels":[],"label_agreement":null},{"id":"W4384937323","doi":"10.1016/j.neucom.2023.126547","title":"On k-means iterations and Gaussian clusters","year":2023,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Innovate UK; Royal Society","keywords":"Cluster analysis; Computer science; Set (abstract data type); k-means clustering; Algorithm; Gaussian; Data set; Data mining; Determining the number of clusters in a data set; Pattern recognition (psychology); Correlation clustering; Artificial intelligence; CURE data clustering algorithm","score_opus":0.02419776443647915,"score_gpt":0.3013819124988729,"score_spread":0.2771841480623938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384937323","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036865566,0.004671337,0.98507243,0.0009777168,0.0001990601,0.00006746255,0.00008203488,0.00024178838,0.0050016437],"genre_scores_gemma":[0.1547505,0.010647904,0.8239689,0.00085163116,0.0015905218,0.0005607046,0.00048655877,0.00048670708,0.006656555],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9868945,0.006582581,0.00055943313,0.002057186,0.0034232787,0.0004830984],"domain_scores_gemma":[0.9596193,0.031184934,0.0016433517,0.0021619112,0.0048675165,0.00052291737],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014522545,0.002161469,0.0024465863,0.0044816313,0.0020164829,0.003999617,0.0031721096,0.004385348,0.00370776],"category_scores_gemma":[0.06427027,0.0014283853,0.0015745776,0.007018981,0.006718937,0.006104438,0.0046132472,0.006504504,0.0024365576],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029731134,0.00008546894,0.0019235638,0.00046258478,0.00015546293,0.0001778949,0.0008031671,0.351216,0.0009316977,0.52790093,0.00778224,0.10826374],"study_design_scores_gemma":[0.000030427738,0.00008247016,0.0006029967,0.00016052385,0.000022806415,0.00010093234,0.000078543475,0.58685035,0.0009019539,0.40083045,0.010261581,0.00007688823],"about_ca_topic_score_codex":0.0069834203,"about_ca_topic_score_gemma":0.004284978,"teacher_disagreement_score":0.014522545,"about_ca_system_score_codex":0.003583228,"about_ca_system_score_gemma":0.0026902372,"threshold_uncertainty_score":0.076803446},"labels":[],"label_agreement":null},{"id":"W4385232475","doi":"10.1109/tpami.2023.3298629","title":"Knowledge-Induced Multiple Kernel Fuzzy Clustering","year":2023,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"Cluster analysis; Fuzzy clustering; Computer science; Data mining; Artificial intelligence; Fuzzy logic; Knowledge extraction; Kernel (algebra); Domain knowledge; Correlation clustering; Data stream clustering; Pattern recognition (psychology); CURE data clustering algorithm; Machine learning; Mathematics","score_opus":0.050388673218625786,"score_gpt":0.3325815652432935,"score_spread":0.28219289202466774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385232475","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011682626,0.00019684306,0.9865493,0.0000744117,0.000020635616,0.000036359004,0.00004402571,0.00029103394,0.0011048579],"genre_scores_gemma":[0.61361927,0.00038387752,0.38242266,0.00008537403,0.000046232217,0.00014386953,0.000481325,0.00009286417,0.002724565],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979107,0.00038483719,0.0001579081,0.00048338936,0.00089438935,0.00016888753],"domain_scores_gemma":[0.9976044,0.00063673797,0.00020845306,0.00036998923,0.0011166316,0.00006382974],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015306096,0.0006788656,0.0011572388,0.0018119472,0.0009125568,0.0015400149,0.0017450359,0.0012155771,0.0011348525],"category_scores_gemma":[0.006612902,0.00033635763,0.0012280873,0.0019184109,0.00092123216,0.0024085417,0.001700474,0.0010248233,0.00046173105],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000184443,0.00010119441,0.0017427268,0.0002596659,0.00015250288,0.00014080625,0.00035718654,0.5833345,0.008695056,0.03991163,0.0023725256,0.36274776],"study_design_scores_gemma":[0.000006143562,0.000019440704,0.000388583,0.000007919849,0.00001097244,0.00003827016,0.000025285575,0.98780614,0.0029242893,0.007873156,0.00088109006,0.000018706945],"about_ca_topic_score_codex":0.008070629,"about_ca_topic_score_gemma":0.0049103866,"teacher_disagreement_score":0.008070629,"about_ca_system_score_codex":0.0016169178,"about_ca_system_score_gemma":0.0019087538,"threshold_uncertainty_score":0.016047299},"labels":[],"label_agreement":null},{"id":"W4385768263","doi":"10.24963/ijcai.2023/225","title":"Optimal Decision Trees For Interpretable Clustering with Constraints","year":2023,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; University of Toronto","funders":"Vector Institute; Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research","keywords":"Cluster analysis; Interpretability; Constrained clustering; Computer science; Conceptual clustering; Data mining; Artificial intelligence; Machine learning; Correlation clustering; CURE data clustering algorithm","score_opus":0.02759134347042677,"score_gpt":0.32498078392234264,"score_spread":0.29738944045191584,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385768263","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0074921693,0.00024394806,0.9901826,0.00033325262,0.00002285232,0.00007295648,0.00021075757,0.0002594947,0.0011821251],"genre_scores_gemma":[0.19782518,0.0004106444,0.7971126,0.00043566377,0.000104240746,0.00044615264,0.0017280211,0.0002636359,0.0016738927],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9964573,0.0017711759,0.0001942317,0.00076352723,0.0006270307,0.00018671885],"domain_scores_gemma":[0.9887611,0.008749801,0.0006597916,0.0006655986,0.0009371418,0.00022642613],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045823674,0.0014790476,0.0016522629,0.0019581227,0.0012931696,0.0019573467,0.0019284537,0.002451929,0.0042576827],"category_scores_gemma":[0.019483957,0.0008359458,0.001637708,0.0024378642,0.0020590199,0.002937367,0.001973084,0.003669152,0.00089563825],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016195144,0.00010473326,0.00091043627,0.00033582794,0.000089550056,0.00015213522,0.00032786347,0.805447,0.0018455131,0.09645905,0.0062797307,0.08788621],"study_design_scores_gemma":[0.00002351836,0.000023858986,0.0000977244,0.000030072693,0.000009885978,0.000027630573,0.00003631011,0.88834953,0.00045843885,0.1100181,0.0009154362,0.000009566169],"about_ca_topic_score_codex":0.0028134568,"about_ca_topic_score_gemma":0.0056333654,"teacher_disagreement_score":0.0045823674,"about_ca_system_score_codex":0.001840836,"about_ca_system_score_gemma":0.0021897722,"threshold_uncertainty_score":0.024234116},"labels":[],"label_agreement":null},{"id":"W4385832543","doi":"10.1007/978-3-031-39821-6_28","title":"Subspace Clustering Technique Using Multi-objective Functions for Multi-class Categorical Data","year":2023,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Cluster analysis; Computer science; Categorical variable; Data mining; CURE data clustering algorithm; Correlation clustering; Fuzzy clustering; Clustering high-dimensional data; Entropy (arrow of time); Artificial intelligence; Class (philosophy); Pattern recognition (psychology); Subspace topology; Data stream clustering; Machine learning","score_opus":0.1494450529088616,"score_gpt":0.3744778443600943,"score_spread":0.2250327914512327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385832543","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045847693,0.00022227825,0.9942152,0.00004872841,0.000042589152,0.00002436217,0.0000697134,0.0003168341,0.0004756767],"genre_scores_gemma":[0.09400249,0.00039234594,0.90138173,0.000055607117,0.00006580362,0.00014615554,0.000727758,0.00015034326,0.003077775],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99867463,0.0003493342,0.00008553691,0.00025284357,0.0005445892,0.00009315613],"domain_scores_gemma":[0.999102,0.00021506839,0.000054961,0.00014171378,0.00044182627,0.000044490742],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014274612,0.0007176114,0.0016652293,0.0016174543,0.0010262429,0.001216275,0.0015097787,0.0008621669,0.0022753999],"category_scores_gemma":[0.0017691549,0.0002890385,0.0019701454,0.0034413827,0.0004409155,0.0014127984,0.0013281419,0.0014776539,0.0011342095],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016383034,0.00016343022,0.0013058712,0.00022758024,0.00019094578,0.000080576574,0.00019849086,0.112164676,0.021718789,0.014864025,0.0068490226,0.8420728],"study_design_scores_gemma":[0.0000061482497,0.000070745846,0.0009767313,0.000013166333,0.0000361147,0.00013354549,0.00009563116,0.98237014,0.0052302354,0.007742299,0.0032888318,0.000036393092],"about_ca_topic_score_codex":0.0031996015,"about_ca_topic_score_gemma":0.0038915293,"teacher_disagreement_score":0.0031996015,"about_ca_system_score_codex":0.0005658665,"about_ca_system_score_gemma":0.0009771278,"threshold_uncertainty_score":0.00761199},"labels":[],"label_agreement":null},{"id":"W4386124589","doi":"10.2139/ssrn.4551320","title":"Approximation Schemes for Min-Sum K-Clustering","year":2023,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Mathematics; Combinatorics; Statistics","score_opus":0.04373845239734463,"score_gpt":0.33118047490163116,"score_spread":0.2874420225042865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386124589","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010511341,0.001358166,0.9812044,0.00080810237,0.0002417862,0.000105939886,0.00040301675,0.0007980756,0.00456908],"genre_scores_gemma":[0.23494208,0.0015089466,0.7455414,0.00060311746,0.00045873437,0.00035990923,0.0013218867,0.0006109574,0.014653079],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9967668,0.0009531631,0.00019965289,0.0005312789,0.0011894354,0.00035967247],"domain_scores_gemma":[0.9910279,0.0038492042,0.00038759844,0.0030641083,0.0011954914,0.00047578197],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039917226,0.0015388861,0.0026653165,0.0018250619,0.0020078763,0.0044054617,0.008219071,0.0026508435,0.00830487],"category_scores_gemma":[0.020091062,0.0009085343,0.0014791721,0.0058991024,0.0017041658,0.0070621665,0.006260589,0.004982927,0.0031726088],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001016433,0.00030192354,0.0008186366,0.0006455471,0.0001483609,0.00007474375,0.00052931474,0.40269744,0.0027721773,0.261483,0.027976422,0.30153605],"study_design_scores_gemma":[0.000051038947,0.000054818083,0.00015913715,0.000042470198,0.000035583897,0.000101095866,0.00010243437,0.7612104,0.00102237,0.2321711,0.0050243856,0.000025164944],"about_ca_topic_score_codex":0.0041839937,"about_ca_topic_score_gemma":0.006599762,"teacher_disagreement_score":0.00830487,"about_ca_system_score_codex":0.0038664439,"about_ca_system_score_gemma":0.0026073623,"threshold_uncertainty_score":0.028053164},"labels":[],"label_agreement":null},{"id":"W4386283145","doi":"10.18280/mmep.100420","title":"Performance Evaluation of Some Clustering Algorithms under Different Validity Indices","year":2023,"lang":"en","type":"article","venue":"Mathematical Modelling and Engineering Problems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Computer science; Algorithm; Data mining; Artificial intelligence","score_opus":0.12362484919823169,"score_gpt":0.3054019343034031,"score_spread":0.18177708510517143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386283145","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.65423244,0.0067964005,0.3186748,0.0011107883,0.00037519363,0.00089310796,0.0018145681,0.005570924,0.010531803],"genre_scores_gemma":[0.7225534,0.0012148967,0.27114162,0.00012337856,0.00008920555,0.00039192277,0.0031006292,0.00049658935,0.0008882883],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9756139,0.011414976,0.0031415476,0.003145893,0.0055745803,0.0011090628],"domain_scores_gemma":[0.8896825,0.07582045,0.0064923777,0.007767414,0.018863687,0.0013736075],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.036553282,0.002278309,0.002368544,0.008534407,0.0033944033,0.005031363,0.0026256659,0.0033239424,0.0010003586],"category_scores_gemma":[0.097831406,0.00058779755,0.002032685,0.0062287464,0.002155653,0.0031860538,0.002813596,0.0014714033,0.00075162255],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038907402,0.00088765146,0.05791361,0.0016654843,0.0014915507,0.00024540123,0.0016236885,0.57327086,0.00526107,0.009878486,0.0042699915,0.33960146],"study_design_scores_gemma":[0.00013655933,0.00069589494,0.0094114905,0.00016143113,0.0002173209,0.00027193862,0.0007701419,0.9686493,0.013643744,0.0046296734,0.0012935784,0.000118919284],"about_ca_topic_score_codex":0.011561386,"about_ca_topic_score_gemma":0.0064544627,"teacher_disagreement_score":0.036553282,"about_ca_system_score_codex":0.0040528327,"about_ca_system_score_gemma":0.004089598,"threshold_uncertainty_score":0.19331455},"labels":[],"label_agreement":null},{"id":"W4386846972","doi":"10.14293/p2199-8442.1.sop-.pquwpp.v1","title":"A study on density based spatial clustering of applications with noise","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"DBSCAN; Cluster analysis; Computer science; Outlier; Noise (video); Data mining; Pattern recognition (psychology); Presentation (obstetrics); Artificial intelligence; Image (mathematics); Correlation clustering; CURE data clustering algorithm","score_opus":0.0700196497098633,"score_gpt":0.34811029555545936,"score_spread":0.27809064584559606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386846972","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0560323,0.010712032,0.9169827,0.0023913654,0.00034417096,0.000196474,0.0001732857,0.00037629783,0.012791436],"genre_scores_gemma":[0.7181122,0.015898453,0.25152615,0.00083235797,0.0010905891,0.00029160525,0.00051302335,0.00032470506,0.0114109125],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99427265,0.0019705934,0.00024009372,0.0007523347,0.002451489,0.00031280398],"domain_scores_gemma":[0.9893861,0.0071342154,0.00041910837,0.0006909619,0.0021530555,0.00021665476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039347196,0.0010753112,0.0010490193,0.00199004,0.0013422422,0.0036085444,0.0020376155,0.0017842598,0.0017448474],"category_scores_gemma":[0.02114648,0.00068082585,0.0010838514,0.0048909774,0.001991957,0.0032828145,0.0018442438,0.002255498,0.00039718946],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027483457,0.00015406762,0.006546382,0.0010862638,0.00018814027,0.00077639444,0.0018012467,0.46008015,0.009244071,0.33660126,0.008904751,0.17434242],"study_design_scores_gemma":[0.0000095134965,0.00015353771,0.0020572967,0.00010891369,0.000053300195,0.00048498964,0.00046238856,0.923418,0.0052043158,0.048755426,0.01923717,0.00005511561],"about_ca_topic_score_codex":0.01178174,"about_ca_topic_score_gemma":0.005016885,"teacher_disagreement_score":0.01178174,"about_ca_system_score_codex":0.0024065536,"about_ca_system_score_gemma":0.0017737453,"threshold_uncertainty_score":0.023426354},"labels":[],"label_agreement":null},{"id":"W4387790211","doi":"10.2139/ssrn.4602038","title":"Parsimonious Bayesian Model-Based Clustering with Dissimilarities","year":2023,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Cluster analysis; Bayesian probability; Computer science; Artificial intelligence; Econometrics; Data mining; Statistics; Mathematics","score_opus":0.025785571967865857,"score_gpt":0.2935396233393311,"score_spread":0.26775405137146524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387790211","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013517514,0.00027525923,0.9839094,0.0003467252,0.000042826523,0.00012500184,0.00035285618,0.0003926448,0.0010378154],"genre_scores_gemma":[0.3598899,0.00048066382,0.62878853,0.0004558948,0.0002236322,0.0007102196,0.0036921953,0.00077518524,0.004983815],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98926824,0.0061762603,0.0005926609,0.0023049281,0.0012846574,0.00037319848],"domain_scores_gemma":[0.97691965,0.016107334,0.0009787558,0.0040169857,0.0014523466,0.0005249952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011261906,0.0019177482,0.0040648333,0.0052843234,0.0040593185,0.0046085506,0.00927022,0.006173999,0.006465621],"category_scores_gemma":[0.0544468,0.002802058,0.0039775674,0.0064995633,0.002896485,0.0073751025,0.0064547807,0.0058520855,0.0032769223],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009133246,0.00029956503,0.0044482374,0.00060251175,0.0007420676,0.00036688862,0.0010099657,0.7048637,0.0021131816,0.13208173,0.006693497,0.14586525],"study_design_scores_gemma":[0.000044903223,0.00003027074,0.00037359932,0.000038512557,0.00006073627,0.00008284692,0.00004207476,0.832808,0.0003123168,0.16555116,0.0006278787,0.000027665816],"about_ca_topic_score_codex":0.0060233762,"about_ca_topic_score_gemma":0.00794682,"teacher_disagreement_score":0.011261906,"about_ca_system_score_codex":0.0024665159,"about_ca_system_score_gemma":0.0023358993,"threshold_uncertainty_score":0.059559405},"labels":[],"label_agreement":null},{"id":"W4388305482","doi":"10.1080/00401706.2023.2262896","title":"Post-Shrinkage Strategies in Statistical and Machine Learning for High Dimensional DataPost-Shrinkage Strategies in Statistical and Machine Learning for High Dimensional Data, Syed Ejaz Ahmed, Feryaal Ahmed, and Bahadir Yüzbaşı, New York: Chapman and Hall/CRC Press, 2023, 408 pp., ISBN 9780367763442","year":2023,"lang":"en","type":"article","venue":"Technometrics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Shrinkage; Statistical learning; Statistical analysis; Computer science; Artificial intelligence; Mathematics; Machine learning; Statistics","score_opus":0.05985652472270515,"score_gpt":0.317590329561149,"score_spread":0.2577338048384439,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388305482","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015912633,0.057185724,0.9338542,0.0020951268,0.00087569194,0.000036473248,0.00012561516,0.00051709404,0.0037188362],"genre_scores_gemma":[0.05784618,0.13329217,0.7720951,0.0015833849,0.004793635,0.00056795875,0.0010552306,0.0011271663,0.027639188],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99524224,0.0022281823,0.00034397977,0.0006645323,0.0014038663,0.000117195945],"domain_scores_gemma":[0.99113256,0.006894834,0.0003432572,0.0007385379,0.0007950992,0.00009582265],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069639604,0.0023745967,0.002618621,0.0018842948,0.0004836178,0.0028439779,0.0019315695,0.0023070471,0.0068704803],"category_scores_gemma":[0.011815236,0.0014302187,0.0016046276,0.0044107246,0.0038033503,0.0050638285,0.0026350003,0.005546344,0.005861181],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014712151,0.00009492947,0.0006920609,0.0016274721,0.00031816334,0.00034684673,0.00053382554,0.03818004,0.0035316017,0.3382428,0.06282395,0.5534612],"study_design_scores_gemma":[0.00004598293,0.00022141189,0.00171138,0.00047412803,0.00011783457,0.000617409,0.00014378708,0.244477,0.004657324,0.60729444,0.14005114,0.00018822406],"about_ca_topic_score_codex":0.0011294284,"about_ca_topic_score_gemma":0.0013286498,"teacher_disagreement_score":0.0069639604,"about_ca_system_score_codex":0.000937629,"about_ca_system_score_gemma":0.0009973623,"threshold_uncertainty_score":0.036829352},"labels":[],"label_agreement":null},{"id":"W4388517820","doi":"10.1109/fuzz52849.2023.10309720","title":"Incremental Cluster Validity Index for Predicting Early Signs of Change in Data Streams","year":2023,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Data stream mining; Cluster analysis; Outlier; Data mining; Data stream; Index (typography); Change detection; STREAMS; Cluster (spacecraft); Artificial intelligence","score_opus":0.3202045578532634,"score_gpt":0.4126508453781036,"score_spread":0.09244628752484019,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388517820","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25737596,0.0017979913,0.7316473,0.0005189064,0.0002630931,0.00041322358,0.0016958702,0.0017897916,0.004497847],"genre_scores_gemma":[0.83609205,0.00041623667,0.15988085,0.000105224455,0.000117661846,0.00017712206,0.0019928836,0.00008845135,0.0011294358],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99805605,0.00026815353,0.00021557267,0.0003319497,0.00096453883,0.00016375179],"domain_scores_gemma":[0.9901147,0.0041058837,0.0011592464,0.000792438,0.0033754646,0.0004523082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027310383,0.00073122006,0.0011967449,0.005048699,0.0007665842,0.0016678704,0.002092932,0.0007560751,0.0007551822],"category_scores_gemma":[0.017758027,0.00025977596,0.0006186524,0.0036653155,0.000775775,0.0024633128,0.0014652642,0.0010349222,0.00038723223],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011452098,0.00043875535,0.22294062,0.00045522032,0.00043719157,0.00035487738,0.00043214718,0.26206014,0.011216265,0.011276405,0.007385894,0.4818572],"study_design_scores_gemma":[0.000022461803,0.00017404667,0.01866711,0.000029597315,0.000045707682,0.0001684865,0.0001356545,0.9697381,0.004487993,0.0048170355,0.0016541807,0.00005961303],"about_ca_topic_score_codex":0.00800067,"about_ca_topic_score_gemma":0.00717177,"teacher_disagreement_score":0.00800067,"about_ca_system_score_codex":0.0014694352,"about_ca_system_score_gemma":0.0017706663,"threshold_uncertainty_score":0.015908182},"labels":[],"label_agreement":null},{"id":"W4390419450","doi":"10.2478/ecce-2023-0005","title":"Optimized Centroid-Based Clustering of Dense Nearly-square Point Clouds by the Hexagonal Pattern","year":2023,"lang":"en","type":"article","venue":"Electrical Control and Communication Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Northwestern Polytechnic","funders":"","keywords":"Centroid; Cluster analysis; Euclidean distance; Hexagonal crystal system; Hexagonal lattice; Square lattice; Combinatorics; Euclidean geometry; Algorithm; Square (algebra); Unit square; Computer science; Mathematics; Partition (number theory); Data mining; Geometry; Artificial intelligence; Physics; Crystallography; Statistical physics","score_opus":0.009776132249794716,"score_gpt":0.2415935204913045,"score_spread":0.2318173882415098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390419450","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026391704,0.00009858149,0.9719505,0.0000727266,0.000028684608,0.0000677084,0.00012526114,0.00047320468,0.0007916718],"genre_scores_gemma":[0.2932062,0.000115547875,0.70462394,0.000047854977,0.00001810611,0.000122013545,0.00048228406,0.00014788851,0.0012361575],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99925953,0.00021796417,0.000046221918,0.00013291293,0.00026397975,0.00007935192],"domain_scores_gemma":[0.99887747,0.00029122442,0.000114721224,0.00019887868,0.00046401832,0.000053706437],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007956283,0.0005627606,0.0010728826,0.0011885439,0.0007483681,0.0009415635,0.0014244099,0.0007179487,0.0011931537],"category_scores_gemma":[0.0030577297,0.0004953525,0.0006717188,0.0020568762,0.0005205976,0.0009914757,0.000975205,0.00059732684,0.0006120973],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015757362,0.00004765766,0.0017986293,0.00007397797,0.000057641795,0.000056644112,0.00011840692,0.89567506,0.006164399,0.007837129,0.0025259624,0.08548689],"study_design_scores_gemma":[0.000006833138,0.000009841736,0.00019214895,0.0000023523705,0.0000019100432,0.000011824836,0.000017402199,0.99676836,0.0009333756,0.0017285895,0.00032183187,0.000005433138],"about_ca_topic_score_codex":0.01699693,"about_ca_topic_score_gemma":0.015644405,"teacher_disagreement_score":0.01699693,"about_ca_system_score_codex":0.0010839404,"about_ca_system_score_gemma":0.0014056119,"threshold_uncertainty_score":0.033795953},"labels":[],"label_agreement":null},{"id":"W4390444745","doi":"10.18280/ts.400626","title":"Enhanced Cross-Validation Methods Leveraging Clustering Techniques","year":2023,"lang":"en","type":"article","venue":"Traitement du signal","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Computer science; Cross-validation; Data mining; Artificial intelligence","score_opus":0.060675995641990804,"score_gpt":0.4088140946180655,"score_spread":0.34813809897607473,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390444745","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015909744,0.0011838711,0.98012286,0.00015749819,0.00012295836,0.00016983392,0.00008374356,0.001384869,0.0008645738],"genre_scores_gemma":[0.3013434,0.0007303697,0.6934018,0.00032324888,0.00015777918,0.00048377912,0.0010787845,0.0005123735,0.001968403],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9836209,0.008727928,0.001472138,0.002253703,0.0035473858,0.00037791752],"domain_scores_gemma":[0.960504,0.021149598,0.002013475,0.006016036,0.009959552,0.00035734536],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.028038023,0.0020025945,0.0019163952,0.003605958,0.000896387,0.0023174435,0.0036385062,0.002375999,0.0017210534],"category_scores_gemma":[0.056716725,0.00071715616,0.0017566844,0.0023367906,0.0011284283,0.0024599745,0.0026166765,0.0030293323,0.001475749],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005700237,0.0004293898,0.0106430035,0.0006852089,0.0011819188,0.00019573627,0.0006343625,0.3998977,0.011634989,0.013070334,0.004613756,0.55644363],"study_design_scores_gemma":[0.000036922862,0.00021241521,0.0026503874,0.00013122865,0.0000703809,0.0001530864,0.000075563374,0.98098075,0.005746873,0.0066049783,0.0032827116,0.000054714674],"about_ca_topic_score_codex":0.0030821841,"about_ca_topic_score_gemma":0.0041278577,"teacher_disagreement_score":0.028038023,"about_ca_system_score_codex":0.000909916,"about_ca_system_score_gemma":0.0020340686,"threshold_uncertainty_score":0.14828104},"labels":[],"label_agreement":null},{"id":"W4390597136","doi":"10.5383/juspn.16.01.006","title":"An Enhanced Fuzzy Clustering to Pattern Recognition for Cloud Computing, by using Model Aggregation and Model Selection","year":2022,"lang":"en","type":"article","venue":"Journal of Ubiquitous Systems and Pervasive Networks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Université du Québec à Rimouski","funders":"","keywords":"Cluster analysis; Cloud computing; Computer science; Fuzzy clustering; Data mining; Fuzzy logic; Adaptive resonance theory; Artificial intelligence; Machine learning; Model selection","score_opus":0.03468317124047883,"score_gpt":0.30431706792857116,"score_spread":0.26963389668809234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390597136","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007940685,0.00023906832,0.990473,0.00010053779,0.000047749276,0.00006829704,0.000025072228,0.00022790032,0.0008777358],"genre_scores_gemma":[0.3936372,0.0005566384,0.6027569,0.00013074443,0.000074008945,0.00026715305,0.00016431854,0.000070508264,0.0023424951],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99954635,0.00013193066,0.00002843012,0.00009817906,0.00016277083,0.000032288168],"domain_scores_gemma":[0.9996464,0.00013191241,0.00003234426,0.00005294221,0.00011987543,0.00001649263],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008669449,0.0006373148,0.00071375864,0.00091514445,0.00057649333,0.0009004728,0.00096501777,0.00062182086,0.0011316496],"category_scores_gemma":[0.0017521166,0.00023153804,0.0009213994,0.0011447202,0.0003500751,0.0009070824,0.0008518865,0.0007126171,0.00029327633],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010873907,0.000121743884,0.0010645855,0.0001656448,0.000111156165,0.00011704334,0.00014656741,0.6900983,0.009406239,0.019969271,0.0025623061,0.27612832],"study_design_scores_gemma":[0.0000023394712,0.000013088155,0.000098376906,0.0000035959729,0.0000050348294,0.00001488931,0.0000061848596,0.9968561,0.00067481434,0.0018070708,0.0005149008,0.0000036651177],"about_ca_topic_score_codex":0.007787261,"about_ca_topic_score_gemma":0.0057320185,"teacher_disagreement_score":0.007787261,"about_ca_system_score_codex":0.0008452338,"about_ca_system_score_gemma":0.00097117765,"threshold_uncertainty_score":0.015483856},"labels":[],"label_agreement":null},{"id":"W4392136634","doi":"10.31234/osf.io/fq6e9","title":"Ensemble clustering: A practical tutorial","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"","keywords":"Cluster analysis; Computer science; Consensus clustering; Data mining; Ensemble learning; Machine learning; Clustering high-dimensional data; CURE data clustering algorithm; Correlation clustering; Robustness (evolution); Stability (learning theory); Artificial intelligence","score_opus":0.062162072007400676,"score_gpt":0.38882062716574867,"score_spread":0.32665855515834796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392136634","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007701771,0.023040826,0.93745446,0.003136025,0.0033175552,0.00024058386,0.0011726016,0.0055126757,0.025354983],"genre_scores_gemma":[0.012552525,0.04460588,0.8706423,0.0030993,0.007231402,0.0010077511,0.003943175,0.0033625471,0.053555217],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990133,0.0003334445,0.0000665822,0.00021954313,0.0003194282,0.00004771407],"domain_scores_gemma":[0.998154,0.0010533254,0.00006769548,0.00019032766,0.00043395592,0.00010063796],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020691769,0.0023944424,0.0012585482,0.002198402,0.00088214263,0.0023769168,0.0019306733,0.0019657004,0.04171675],"category_scores_gemma":[0.0062060226,0.0010973451,0.0020087126,0.003061156,0.0007032453,0.0032500483,0.0019158362,0.0038242948,0.03133716],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004724104,0.00011424926,0.00049609994,0.0010467502,0.00011879387,0.0003421064,0.0002516166,0.021511586,0.0020095238,0.094041094,0.42304718,0.45697382],"study_design_scores_gemma":[0.000016803995,0.000055271274,0.00059469446,0.00036810304,0.000027450284,0.000820453,0.000077474404,0.038137,0.00090451847,0.14437343,0.81454784,0.00007693881],"about_ca_topic_score_codex":0.002083407,"about_ca_topic_score_gemma":0.0030399617,"teacher_disagreement_score":0.04171675,"about_ca_system_score_codex":0.000909043,"about_ca_system_score_gemma":0.00095580786,"threshold_uncertainty_score":0.13955647},"labels":[],"label_agreement":null},{"id":"W4392197543","doi":"10.28924/2291-8639-22-2024-36","title":"Impact of Dataset Scaling on Hierarchical Clustering: A Comparative Analysis of Distance-Based and Ratio-Based Methods","year":2024,"lang":"en","type":"article","venue":"International Journal of Analysis and Applications","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Hierarchical clustering; Mathematics; Cluster analysis; Scaling; Multidimensional scaling; Data mining; Statistics; Computer science; Geometry","score_opus":0.0474050211507275,"score_gpt":0.4728101699367248,"score_spread":0.42540514878599733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392197543","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3538649,0.022421205,0.6042799,0.0015941701,0.00079637504,0.0014541312,0.001068193,0.0024584457,0.0120626325],"genre_scores_gemma":[0.5563169,0.00350414,0.43693712,0.00019455398,0.000164442,0.00045658517,0.0012706397,0.00045391553,0.0007016733],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.96323884,0.021514088,0.0023020683,0.0028159358,0.009686881,0.00044216533],"domain_scores_gemma":[0.8957323,0.07627199,0.004961959,0.0080205435,0.014244575,0.0007686123],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.038602322,0.0014774231,0.0016594724,0.007135326,0.0016338752,0.0032228257,0.0022184986,0.0011876627,0.0012797439],"category_scores_gemma":[0.13194382,0.00037135498,0.0017028562,0.008904997,0.0011075685,0.0045752847,0.0025776974,0.0010702258,0.0005867872],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002069269,0.0005568943,0.044017658,0.0043101823,0.0025436769,0.0003102353,0.0036092272,0.09808402,0.006658649,0.02010697,0.007700441,0.8100328],"study_design_scores_gemma":[0.00041459673,0.0024901403,0.065364994,0.0013392846,0.0018096678,0.0012405084,0.007975265,0.8422154,0.015969431,0.03343921,0.027271308,0.00047026452],"about_ca_topic_score_codex":0.0047507337,"about_ca_topic_score_gemma":0.0052306373,"teacher_disagreement_score":0.038602322,"about_ca_system_score_codex":0.001542274,"about_ca_system_score_gemma":0.002061511,"threshold_uncertainty_score":0.20415103},"labels":[],"label_agreement":null},{"id":"W4392715283","doi":"10.1016/j.ijar.2024.109169","title":"An auto-weighted enhanced horizontal collaborative fuzzy clustering algorithm with knowledge adaption mechanism","year":2024,"lang":"en","type":"article","venue":"International Journal of Approximate Reasoning","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Fuzzy clustering; Partition (number theory); Fuzzy logic; Entropy (arrow of time); Machine learning; Artificial intelligence; Mathematics","score_opus":0.010357393417794491,"score_gpt":0.3068244848556005,"score_spread":0.296467091437806,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392715283","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033707477,0.00027190195,0.9624189,0.000083025996,0.000073426236,0.00007173702,0.00005585617,0.00048490328,0.0028327988],"genre_scores_gemma":[0.49517298,0.00019801506,0.49826512,0.00010179816,0.000057341527,0.00015410768,0.00021776691,0.00005686936,0.0057760742],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99906653,0.00012314887,0.00006775925,0.00022271751,0.00042388827,0.00009596883],"domain_scores_gemma":[0.9992823,0.00013173414,0.000050715098,0.00014127485,0.00035248924,0.000041422427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008532522,0.00049720804,0.0011015582,0.0010533169,0.0009197338,0.0009247322,0.0025212897,0.0010644662,0.002413262],"category_scores_gemma":[0.0015065337,0.00034306492,0.00072416366,0.0015309805,0.00044288504,0.0013685329,0.0017633775,0.0005015733,0.00051306054],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042195828,0.00027273726,0.002063173,0.00014197415,0.0002829356,0.00019678404,0.00020720887,0.23630807,0.021977834,0.014537907,0.0038474153,0.719742],"study_design_scores_gemma":[0.000025897147,0.0000624165,0.0004253459,0.000006534138,0.000046419183,0.00011883492,0.00003645207,0.9906832,0.004479382,0.002856426,0.0012403044,0.000018750108],"about_ca_topic_score_codex":0.0054454356,"about_ca_topic_score_gemma":0.005537073,"teacher_disagreement_score":0.0054454356,"about_ca_system_score_codex":0.0005275272,"about_ca_system_score_gemma":0.0012122154,"threshold_uncertainty_score":0.010827422},"labels":[],"label_agreement":null},{"id":"W4392738528","doi":"10.1080/03610918.2024.2328166","title":"An ensemble approach to determine the number of latent dimensions and assess its reliability","year":2024,"lang":"en","type":"article","venue":"Communications in Statistics - Simulation and Computation","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Reliability (semiconductor); Computer science; Statistics; Reliability engineering; Mathematics; Engineering; Physics","score_opus":0.19477715379534086,"score_gpt":0.47886232169168147,"score_spread":0.28408516789634064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392738528","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017070124,0.0003753587,0.98145884,0.00012518401,0.00003970492,0.000040736955,0.00013835216,0.00019003513,0.00056156097],"genre_scores_gemma":[0.36620104,0.0007974109,0.6302848,0.00013543689,0.00019062759,0.0003342126,0.00089098595,0.00014942467,0.0010161124],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99511826,0.0023947423,0.00033154502,0.0009016859,0.0010626165,0.00019105017],"domain_scores_gemma":[0.97179925,0.018848412,0.0015842248,0.003841719,0.0034711983,0.00045521534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009865895,0.0012457572,0.0016809602,0.003701646,0.0009975045,0.0018158408,0.0015755361,0.0015161927,0.0016242121],"category_scores_gemma":[0.03326762,0.00044233486,0.0016756807,0.0024152936,0.00092890824,0.0028397415,0.0021184485,0.0026670839,0.00054094766],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003680171,0.00019814512,0.042160787,0.0003566432,0.0012830435,0.0001820252,0.0009006523,0.4037912,0.007900493,0.04432973,0.00397316,0.49455622],"study_design_scores_gemma":[0.000010885149,0.00007792115,0.003403948,0.00006169645,0.000068068075,0.0001080159,0.000101071324,0.9638906,0.0018612372,0.028942367,0.0014298242,0.000044396565],"about_ca_topic_score_codex":0.002835082,"about_ca_topic_score_gemma":0.0034427845,"teacher_disagreement_score":0.009865895,"about_ca_system_score_codex":0.0007016453,"about_ca_system_score_gemma":0.0010379036,"threshold_uncertainty_score":0.052176476},"labels":[],"label_agreement":null},{"id":"W4393552393","doi":"10.5281/zenodo.6577860","title":"SpatialSort: Supporting Datasets","year":2022,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Terry Fox Research Institute; BC Cancer Agency","funders":"","keywords":"Computer science","score_opus":0.03703822458885673,"score_gpt":0.296882333592253,"score_spread":0.2598441090033963,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393552393","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013074072,0.000056989393,0.00036192464,0.00007478281,0.000028290739,0.000026692262,0.9974348,0.0013346527,0.00055117725],"genre_scores_gemma":[0.00041418965,0.000048272133,0.0013635196,0.00007169031,0.0000075963403,0.00020103362,0.9971154,0.00035199686,0.00042635188],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978114,0.00044786825,0.0002832047,0.0006433868,0.0005511633,0.0002630106],"domain_scores_gemma":[0.9965808,0.0011121196,0.00023541428,0.0010977782,0.0006708504,0.00030294718],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00263435,0.0030195164,0.0020337657,0.0047087725,0.0018027647,0.003344059,0.005141891,0.002994736,0.13030113],"category_scores_gemma":[0.01352989,0.0011716282,0.002074706,0.008162767,0.00074883783,0.0023887178,0.0031875356,0.00307754,0.1516397],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006022696,0.000022556978,0.0004830561,0.00045278735,0.000032634503,0.000027297147,0.000026059144,0.000620159,0.00017780362,0.0007272304,0.9952911,0.002079077],"study_design_scores_gemma":[0.0003608891,0.00002490037,0.0020155755,0.00028317244,0.000053138425,0.00012100137,0.00013126411,0.0013080981,0.00088315987,0.007512298,0.98724383,0.00006281433],"about_ca_topic_score_codex":0.016493998,"about_ca_topic_score_gemma":0.035305414,"teacher_disagreement_score":0.13030113,"about_ca_system_score_codex":0.0017561301,"about_ca_system_score_gemma":0.0032723073,"threshold_uncertainty_score":0.4359007},"labels":[],"label_agreement":null},{"id":"W4393622469","doi":"10.5281/zenodo.7812052","title":"Data from: The problem with missingness in mixed-type datasets and what we can do about it","year":2023,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Missing data; Type (biology); Statistics; Computer science; Mathematics; Data science; Biology; Ecology","score_opus":0.0762632156264673,"score_gpt":0.30655359672369104,"score_spread":0.23029038109722372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393622469","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017880218,0.00087909866,0.005557866,0.0036200902,0.00073989306,0.00016939857,0.9826187,0.0032560513,0.0013709075],"genre_scores_gemma":[0.006893065,0.00034193916,0.018193945,0.0014670899,0.00016357044,0.00096447044,0.96931714,0.0012207028,0.0014381515],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9886535,0.0034748877,0.0022303553,0.00223358,0.0028099937,0.0005977658],"domain_scores_gemma":[0.938447,0.0330392,0.0032980824,0.018563678,0.005619434,0.0010326268],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.017878313,0.0016643126,0.0021729607,0.0028740484,0.0019552426,0.005757481,0.0049097976,0.005063353,0.03793137],"category_scores_gemma":[0.10908413,0.001517432,0.0027595784,0.0068233768,0.0018334526,0.0046530073,0.0035969317,0.0039838133,0.02786556],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024290143,0.00004831001,0.0026852426,0.0022582163,0.00015404557,0.00007623511,0.00009438752,0.00082188554,0.00039244533,0.002675385,0.98206884,0.008482116],"study_design_scores_gemma":[0.0011717265,0.00008058071,0.00815767,0.0015011885,0.00015583259,0.0005195068,0.0003908119,0.004743556,0.002892963,0.03082501,0.9493752,0.00018596495],"about_ca_topic_score_codex":0.004285615,"about_ca_topic_score_gemma":0.009658647,"teacher_disagreement_score":0.9821217,"about_ca_system_score_codex":0.0015756892,"about_ca_system_score_gemma":0.0040096096,"threshold_uncertainty_score":0.1268931},"labels":[],"label_agreement":null},{"id":"W4393700520","doi":"10.5281/zenodo.7812051","title":"Data from: The problem with missingness in mixed-type datasets and what we can do about it","year":2023,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Missing data; Type (biology); Computer science; Data type; Data science; Data mining; Statistics; Mathematics; Machine learning; Geology; Programming language","score_opus":0.0762632156264673,"score_gpt":0.30655359672369104,"score_spread":0.23029038109722372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393700520","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017880218,0.00087909866,0.005557866,0.0036200902,0.00073989306,0.00016939857,0.9826187,0.0032560513,0.0013709075],"genre_scores_gemma":[0.006893065,0.00034193916,0.018193945,0.0014670899,0.00016357044,0.00096447044,0.96931714,0.0012207028,0.0014381515],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9886535,0.0034748877,0.0022303553,0.00223358,0.0028099937,0.0005977658],"domain_scores_gemma":[0.938447,0.0330392,0.0032980824,0.018563678,0.005619434,0.0010326268],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.017878313,0.0016643126,0.0021729607,0.0028740484,0.0019552426,0.005757481,0.0049097976,0.005063353,0.03793137],"category_scores_gemma":[0.10908413,0.001517432,0.0027595784,0.0068233768,0.0018334526,0.0046530073,0.0035969317,0.0039838133,0.02786556],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024290143,0.00004831001,0.0026852426,0.0022582163,0.00015404557,0.00007623511,0.00009438752,0.00082188554,0.00039244533,0.002675385,0.98206884,0.008482116],"study_design_scores_gemma":[0.0011717265,0.00008058071,0.00815767,0.0015011885,0.00015583259,0.0005195068,0.0003908119,0.004743556,0.002892963,0.03082501,0.9493752,0.00018596495],"about_ca_topic_score_codex":0.004285615,"about_ca_topic_score_gemma":0.009658647,"teacher_disagreement_score":0.9821217,"about_ca_system_score_codex":0.0015756892,"about_ca_system_score_gemma":0.0040096096,"threshold_uncertainty_score":0.1268931},"labels":[],"label_agreement":null},{"id":"W4393758208","doi":"10.5281/zenodo.5061257","title":"GVI: Sample data for computing VGVI. Vancouver, BC and Manchester.","year":2021,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sample (material); Computer science; Geography; Physics","score_opus":0.0718370294873503,"score_gpt":0.3051840397421036,"score_spread":0.23334701025475332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393758208","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00055061514,0.00017305712,0.0009814929,0.00013921877,0.000106391504,0.00004056659,0.9829516,0.004956614,0.010100539],"genre_scores_gemma":[0.0033690676,0.00029657732,0.0046539484,0.00008951944,0.000024647257,0.0001539483,0.97294766,0.0075973696,0.010867178],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987643,0.000090175934,0.00007039128,0.00028938506,0.00057832454,0.00020752999],"domain_scores_gemma":[0.9967259,0.00031015434,0.00013692869,0.0005182678,0.0020551276,0.00025366113],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008992232,0.0015623099,0.0015715827,0.0039605447,0.0014249803,0.004996044,0.0024196035,0.0010873362,0.16837521],"category_scores_gemma":[0.0063948315,0.0012334613,0.000890083,0.009649288,0.00046424245,0.0018850215,0.0021400321,0.0019636473,0.17885149],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027835966,0.000007644465,0.0011645813,0.00028017643,0.000030962747,0.00001823928,0.000062805215,0.00033462144,0.0001301657,0.0004360465,0.9905737,0.0069332924],"study_design_scores_gemma":[0.0000438359,0.000003718146,0.0058918977,0.0002376978,0.000021082658,0.000026262676,0.0001415823,0.00051958783,0.0003950943,0.00081599876,0.9918676,0.000035645844],"about_ca_topic_score_codex":0.41397074,"about_ca_topic_score_gemma":0.58434093,"teacher_disagreement_score":0.41397074,"about_ca_system_score_codex":0.0036303645,"about_ca_system_score_gemma":0.0060157073,"threshold_uncertainty_score":0.8231223},"labels":[],"label_agreement":null},{"id":"W4396712800","doi":"10.1109/tfuzz.2024.3397808","title":"CBCG: A Clustering Algorithm Based on Bidirectional Conical Information Granularity","year":2024,"lang":"en","type":"article","venue":"IEEE Transactions on Fuzzy Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Basic and Applied Basic Research Foundation of Guangdong Province; Natural Science Foundation of Hunan Province","keywords":"Cluster analysis; Data mining; Granularity; Fuzzy clustering; Computer science; Sorting; Cluster (spacecraft); Algorithm; Fuzzy logic; Single-linkage clustering; Correlation clustering; CURE data clustering algorithm; k-medians clustering; FLAME clustering; Complete-linkage clustering; Pattern recognition (psychology); Artificial intelligence","score_opus":0.020597308030298498,"score_gpt":0.2820249380215247,"score_spread":0.26142762999122615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396712800","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010997812,0.00052448816,0.98501915,0.00011922345,0.00007544095,0.00015988802,0.0001686179,0.0016181987,0.0013172862],"genre_scores_gemma":[0.13523558,0.00042861496,0.8609292,0.00014212877,0.00008080833,0.00029765765,0.00088645914,0.0003076001,0.0016919717],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99731255,0.00029490973,0.00019005571,0.00061147346,0.0013461335,0.00024481866],"domain_scores_gemma":[0.9980319,0.00037884066,0.00022155144,0.00032719082,0.0009449598,0.00009552957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001456393,0.0011463069,0.0016810114,0.006622667,0.002381427,0.002098818,0.0031955906,0.0017079423,0.0017817247],"category_scores_gemma":[0.0041677877,0.000539382,0.001399447,0.007177463,0.0012955614,0.0030266233,0.0020686865,0.0015023543,0.0011181718],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005894578,0.00018170981,0.004549995,0.00034436883,0.00025847682,0.00014704424,0.0006140186,0.164597,0.0186997,0.021839732,0.010902839,0.7772757],"study_design_scores_gemma":[0.0001190311,0.0001487151,0.0023590387,0.000057164765,0.000087716144,0.00032782962,0.00028330085,0.9478556,0.01687384,0.017619584,0.0141154975,0.00015281228],"about_ca_topic_score_codex":0.016562564,"about_ca_topic_score_gemma":0.00919345,"teacher_disagreement_score":0.016562564,"about_ca_system_score_codex":0.002163598,"about_ca_system_score_gemma":0.0029212185,"threshold_uncertainty_score":0.03293228},"labels":[],"label_agreement":null},{"id":"W4396844948","doi":"10.1186/s13104-024-06791-y","title":"An inversion-based clustering approach for complex clusters","year":2024,"lang":"en","type":"article","venue":"BMC Research Notes","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Similarity measure; Profiling (computer programming); Feature (linguistics); Similarity (geometry); Ranking (information retrieval); Artificial intelligence; Inversion (geology); Machine learning; Pattern recognition (psychology)","score_opus":0.30026848375425247,"score_gpt":0.4587841597762987,"score_spread":0.15851567602204625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396844948","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006509891,0.00013481038,0.99193114,0.00010339228,0.000023178702,0.00008258582,0.000051944993,0.00026149384,0.00090157054],"genre_scores_gemma":[0.135247,0.00018377048,0.8619716,0.00013878538,0.000075419324,0.00023824477,0.0004643338,0.00020287563,0.0014779023],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99796104,0.0004978601,0.00012722041,0.0004979555,0.00076858915,0.00014734706],"domain_scores_gemma":[0.99582183,0.0015502089,0.00045462974,0.00050796807,0.0015235466,0.00014183826],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030215497,0.0009923356,0.0013835436,0.0038126044,0.0015327068,0.0018328105,0.002899826,0.0017697344,0.0022996857],"category_scores_gemma":[0.0076195984,0.00053498265,0.0019506188,0.0029992627,0.0014769898,0.001662245,0.0026641567,0.001817149,0.0009642806],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021725321,0.00015181168,0.003843613,0.00029527294,0.00027018628,0.00023952936,0.0008464469,0.6230756,0.00788233,0.059301842,0.0062591485,0.297617],"study_design_scores_gemma":[0.000010419606,0.000021402084,0.00029490504,0.0000126036,0.000015670437,0.00009379935,0.0000580605,0.98169184,0.0008945687,0.015286901,0.0015979557,0.000021814318],"about_ca_topic_score_codex":0.009235373,"about_ca_topic_score_gemma":0.009400881,"teacher_disagreement_score":0.009235373,"about_ca_system_score_codex":0.0019038608,"about_ca_system_score_gemma":0.0024322683,"threshold_uncertainty_score":0.018363237},"labels":[],"label_agreement":null},{"id":"W4398203199","doi":"10.1016/j.ins.2024.120761","title":"An axiomatic framework for three-way clustering","year":2024,"lang":"en","type":"article","venue":"Information Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"China Scholarship Council; National Natural Science Foundation of China","keywords":"Cluster analysis; Computer science; Axiom; Axiomatic system; Data mining; Mathematics; Artificial intelligence","score_opus":0.054430876012930327,"score_gpt":0.39105383680820793,"score_spread":0.3366229607952776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398203199","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016704284,0.00013904124,0.99011594,0.0007731585,0.00007503252,0.000033871675,0.0001563536,0.0001844641,0.006851647],"genre_scores_gemma":[0.09963619,0.00039967475,0.8919525,0.0008950664,0.00020466275,0.00042949626,0.00074881414,0.00017287798,0.005560807],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99367744,0.0024794396,0.0005496865,0.0012456042,0.0016547631,0.0003932381],"domain_scores_gemma":[0.9913033,0.003349984,0.0003985347,0.0022017292,0.0022621583,0.00048431757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070496355,0.0009650214,0.0010970164,0.0029885478,0.0054395227,0.0066496967,0.004972012,0.0031964183,0.008382871],"category_scores_gemma":[0.011170665,0.0009309203,0.0036964128,0.0036372654,0.007687851,0.009575742,0.0063592303,0.005692362,0.002819049],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006420022,0.0000068684726,0.00008501163,0.000036137557,0.000009571299,0.00002285817,0.000112350055,0.0014137587,0.00015354967,0.9923298,0.0012155187,0.004608222],"study_design_scores_gemma":[0.000011010754,0.0000075907888,0.00007709669,0.000020606492,0.0000128212705,0.00010059775,0.0000686739,0.0141391205,0.00026808365,0.97733384,0.00794573,0.000014937169],"about_ca_topic_score_codex":0.0036751202,"about_ca_topic_score_gemma":0.004085636,"teacher_disagreement_score":0.008382871,"about_ca_system_score_codex":0.002570467,"about_ca_system_score_gemma":0.0023062069,"threshold_uncertainty_score":0.037282467},"labels":[],"label_agreement":null},{"id":"W4399424565","doi":"10.48550/arxiv.2406.02465","title":"An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Danmarks Grundforskningsfond; Government of Canada; Canadian Institute for Advanced Research","keywords":"Cluster analysis; Computer science; Artificial intelligence; Encoder; Empirical research; Pattern recognition (psychology); Data mining; Machine learning; Mathematics; Statistics","score_opus":0.07352471772403338,"score_gpt":0.27488410284662,"score_spread":0.20135938512258664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399424565","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82475907,0.0072241034,0.14762169,0.0020184582,0.00045112427,0.00060842076,0.004730116,0.006295461,0.0062915958],"genre_scores_gemma":[0.9057577,0.0007884055,0.07702891,0.00044523846,0.00011360993,0.00022632678,0.012838873,0.0007792099,0.0020216932],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98597926,0.007405377,0.00070459675,0.0036573247,0.0016924221,0.0005609497],"domain_scores_gemma":[0.9410829,0.030943953,0.0028619706,0.017768867,0.0063390615,0.0010032086],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019735094,0.0022987593,0.0013534462,0.0026208875,0.0013921991,0.00215769,0.0034041796,0.0025555429,0.0012578263],"category_scores_gemma":[0.081715554,0.0009478864,0.001259522,0.0027050173,0.0029296945,0.006098903,0.002625305,0.003394342,0.0012925275],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020096905,0.001411906,0.09163068,0.0014206807,0.0012533296,0.0002750869,0.0013082108,0.54052377,0.005779489,0.0068479716,0.03500847,0.31253064],"study_design_scores_gemma":[0.00008715437,0.0005652547,0.012789215,0.00016583232,0.00009580601,0.00032782735,0.00040469866,0.9657071,0.008047716,0.008616279,0.0031188524,0.00007426042],"about_ca_topic_score_codex":0.011703318,"about_ca_topic_score_gemma":0.017420474,"teacher_disagreement_score":0.019735094,"about_ca_system_score_codex":0.0032364863,"about_ca_system_score_gemma":0.0013210884,"threshold_uncertainty_score":0.104370415},"labels":[],"label_agreement":null},{"id":"W4399995656","doi":"10.1109/tpami.2024.3418776","title":"Simplex Clustering via sBeta With Applications to Online Adjustment of Black-Box Predictions","year":2024,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"École de Technologie Supérieure; Institut National de la Recherche Scientifique","funders":"Compute Canada","keywords":"Cluster analysis; Computer science; Artificial intelligence; Simplex; Black box; Combinatorics; Algorithm; Pattern recognition (psychology); Mathematics","score_opus":0.02565515967990565,"score_gpt":0.32014705448452946,"score_spread":0.2944918948046238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399995656","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041624815,0.00010331412,0.99365145,0.000121001925,0.00002941808,0.000036576916,0.000055992918,0.00090989115,0.00092985324],"genre_scores_gemma":[0.31060398,0.00028217884,0.68130785,0.00031508802,0.00009950542,0.00041149696,0.0005522833,0.00149561,0.004932027],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99842095,0.00064504927,0.00007529894,0.00037416737,0.00034794342,0.00013655973],"domain_scores_gemma":[0.9956892,0.0021302875,0.00045413288,0.0008315172,0.00063346454,0.00026137094],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036191894,0.001613042,0.0017872662,0.0012581629,0.0010677228,0.0022343146,0.003388028,0.0018884679,0.0062333466],"category_scores_gemma":[0.014813496,0.0010615941,0.0011953714,0.0015827006,0.0018150416,0.0025838297,0.0044965055,0.003426538,0.0018825022],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021124785,0.00008485562,0.0009668666,0.00010720661,0.00009213311,0.00007739084,0.00016915097,0.83564466,0.0021076319,0.056929205,0.0030554773,0.100554146],"study_design_scores_gemma":[0.0000059695644,0.000009787409,0.000043132794,0.0000074698796,0.0000031627171,0.00000901244,0.000007041685,0.98205525,0.0007805924,0.016482065,0.00058833434,0.000008063583],"about_ca_topic_score_codex":0.0051995274,"about_ca_topic_score_gemma":0.006911568,"teacher_disagreement_score":0.0062333466,"about_ca_system_score_codex":0.002042514,"about_ca_system_score_gemma":0.002889121,"threshold_uncertainty_score":0.020852566},"labels":[],"label_agreement":null},{"id":"W4400811537","doi":"10.1109/csci62032.2023.00024","title":"Three-Way Clustering: An Advanced Soft Clustering Approach","year":2023,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Computer science; Data mining; Artificial intelligence","score_opus":0.054600297286088305,"score_gpt":0.32227804524714937,"score_spread":0.2676777479610611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400811537","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015227493,0.0017444585,0.99253756,0.00029668453,0.00013181627,0.000086443506,0.00014182935,0.0005718989,0.0029665509],"genre_scores_gemma":[0.07246771,0.004225839,0.9145951,0.0005815704,0.0002992422,0.00039357328,0.00094884314,0.0005485299,0.005939604],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9936392,0.0016015016,0.00047925013,0.0012779401,0.0026925013,0.00030966537],"domain_scores_gemma":[0.9947364,0.0014872797,0.00041560642,0.0007823236,0.0023315097,0.0002468235],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038138237,0.0016504577,0.0024579908,0.007959868,0.002712095,0.005334553,0.00388434,0.0028142503,0.0039335666],"category_scores_gemma":[0.0070246323,0.00096506835,0.0033871569,0.008983684,0.0024873791,0.004729087,0.0042897575,0.0034243928,0.0035610688],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026534774,0.00013427422,0.002749313,0.001991606,0.0006908573,0.00035861853,0.0015433504,0.10897666,0.01506797,0.20897165,0.019009074,0.6402412],"study_design_scores_gemma":[0.000040453306,0.00013047895,0.0023334785,0.0004837552,0.00023281759,0.00094488537,0.0007834869,0.549789,0.01431712,0.304619,0.1259034,0.00042212452],"about_ca_topic_score_codex":0.0053785574,"about_ca_topic_score_gemma":0.0043701553,"teacher_disagreement_score":0.007959868,"about_ca_system_score_codex":0.0026606836,"about_ca_system_score_gemma":0.0030182265,"threshold_uncertainty_score":0.020169616},"labels":[],"label_agreement":null},{"id":"W4400929535","doi":"10.24132/csrn.3401.27","title":"LVCluster: Bounded Clustering using Laguerre Voronoi Diagram","year":2024,"lang":"en","type":"article","venue":"Computer Science Research Notes","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Voronoi diagram; Laguerre polynomials; Bounded function; Cluster analysis; Centroidal Voronoi tessellation; Power diagram; Computer science; Diagram; Weighted Voronoi diagram; Mathematics; Artificial intelligence; Mathematical analysis; Geometry","score_opus":0.1833181193907332,"score_gpt":0.46084896332075936,"score_spread":0.27753084393002614,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400929535","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006410291,0.00028540788,0.99017745,0.00009484377,0.00003902922,0.00005807217,0.00022005156,0.0012984937,0.0014163825],"genre_scores_gemma":[0.2701101,0.00048883475,0.72185916,0.00020011254,0.00007927737,0.00038488995,0.001950573,0.001039309,0.0038877812],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9987406,0.00033036555,0.00007101324,0.00030723028,0.00043076946,0.00012012348],"domain_scores_gemma":[0.9983033,0.0006473802,0.0001866679,0.00026970977,0.00047396563,0.000118856056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011470463,0.0010115809,0.0014878035,0.0023537613,0.0012012321,0.0025170422,0.0026971125,0.0013233961,0.0033934696],"category_scores_gemma":[0.005626609,0.00054052885,0.0010208207,0.002035673,0.0011561426,0.002286069,0.0024307892,0.0011558766,0.002026979],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027653552,0.000063953754,0.0022282486,0.00025295757,0.00011508679,0.00013124387,0.00035711523,0.6437268,0.0071323393,0.088231824,0.013154758,0.24432904],"study_design_scores_gemma":[0.000013568075,0.000019703082,0.00016879456,0.000015592934,0.000006765337,0.0000476753,0.000036683938,0.96093106,0.0024245933,0.031350184,0.0049643973,0.00002098993],"about_ca_topic_score_codex":0.009783866,"about_ca_topic_score_gemma":0.012727745,"teacher_disagreement_score":0.009783866,"about_ca_system_score_codex":0.0018869432,"about_ca_system_score_gemma":0.002446756,"threshold_uncertainty_score":0.019453824},"labels":[],"label_agreement":null},{"id":"W4400985398","doi":"10.54254/2755-2721/79/20241631","title":"Analysis of clustering algorithms in Iris and breast cancer datasets","year":2024,"lang":"en","type":"article","venue":"Applied and Computational Engineering","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Cluster analysis; DBSCAN; Expansive; Computer science; Noise (video); Data mining; Domain (mathematical analysis); Data science; Artificial intelligence; CURE data clustering algorithm; Correlation clustering; Mathematics; Image (mathematics)","score_opus":0.009490061202234312,"score_gpt":0.2762931799248556,"score_spread":0.26680311872262125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400985398","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8620221,0.008205457,0.08428194,0.0027894361,0.0004422343,0.0005275964,0.026137264,0.0037651751,0.011828756],"genre_scores_gemma":[0.74690175,0.0021487577,0.18705429,0.00029105355,0.00012959396,0.00036011578,0.059811495,0.00034875702,0.00295417],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99610823,0.0009865684,0.00033913855,0.0006911473,0.0015493925,0.00032555548],"domain_scores_gemma":[0.9938876,0.0025446818,0.00055990665,0.00088011107,0.0019947633,0.00013291517],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005729587,0.00083236344,0.00082362466,0.005378052,0.0012034206,0.0015196638,0.0012278518,0.0011110451,0.00087981165],"category_scores_gemma":[0.015226041,0.0001760186,0.0012378059,0.006120843,0.000510924,0.00096970017,0.0008300506,0.0009848638,0.0007100225],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019172962,0.0008751011,0.09877487,0.001626293,0.0011894488,0.00043584887,0.00094050023,0.28262106,0.008395379,0.019213147,0.08289592,0.50111514],"study_design_scores_gemma":[0.00009664997,0.00043855957,0.1360286,0.00016760085,0.00022799411,0.000618479,0.0013364938,0.79983807,0.01517404,0.011857635,0.034093514,0.0001223945],"about_ca_topic_score_codex":0.015609762,"about_ca_topic_score_gemma":0.02139819,"teacher_disagreement_score":0.015609762,"about_ca_system_score_codex":0.0022006596,"about_ca_system_score_gemma":0.0013788536,"threshold_uncertainty_score":0.031037807},"labels":[],"label_agreement":null},{"id":"W4401011434","doi":"10.1016/b978-0-443-24010-2.00005-6","title":"Clustering","year":2024,"lang":"en","type":"book-chapter","venue":"Elsevier eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Cluster analysis; Computer science; Artificial intelligence","score_opus":0.024099824071557665,"score_gpt":0.2847893070594668,"score_spread":0.26068948298790917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401011434","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032134159,0.0030836258,0.5566765,0.0012433782,0.0018002419,0.0005891218,0.0077280197,0.025130035,0.40053576],"genre_scores_gemma":[0.02958018,0.0033304084,0.3980939,0.0010277313,0.0005969243,0.0006432752,0.026217142,0.0061715296,0.5343389],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998789,0.00012631992,0.000056230598,0.00035887246,0.0005672474,0.00010236042],"domain_scores_gemma":[0.99892837,0.00013732539,0.000030758634,0.00047161485,0.00037586896,0.00005608633],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0008276141,0.0014872489,0.0010807146,0.0036100927,0.0018760979,0.0037228102,0.0028287342,0.0015361279,0.16120067],"category_scores_gemma":[0.002679042,0.0008743956,0.0014490099,0.0046818336,0.00059210125,0.002783134,0.0023519346,0.0015122556,0.12780404],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000768502,0.000059677786,0.00023880154,0.00038264165,0.00004788198,0.00007866747,0.00013376276,0.007991416,0.004897806,0.052927554,0.36698076,0.5661842],"study_design_scores_gemma":[0.000024185947,0.0000426366,0.00053197215,0.00013588472,0.00003354004,0.0004228929,0.00018141122,0.029305954,0.008311736,0.065618835,0.8953434,0.00004743971],"about_ca_topic_score_codex":0.003329522,"about_ca_topic_score_gemma":0.005051098,"teacher_disagreement_score":0.83879936,"about_ca_system_score_codex":0.0011273726,"about_ca_system_score_gemma":0.001568748,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4401331642","doi":"10.1109/fuzz-ieee60900.2024.10611833","title":"Generalized Deep Embedded Fuzzy C-Means for Clustering High-Dimensional Data","year":2024,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Cluster analysis; Fuzzy clustering; Fuzzy logic; Artificial intelligence; Data mining; Clustering high-dimensional data; Pattern recognition (psychology)","score_opus":0.06002803338243091,"score_gpt":0.35236343863696323,"score_spread":0.29233540525453233,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401331642","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00623983,0.00025990876,0.9923935,0.00009216406,0.000027728278,0.000030737927,0.000053474676,0.00048948824,0.00041322148],"genre_scores_gemma":[0.1923576,0.0003530413,0.8045283,0.00018737816,0.000051672687,0.00015925313,0.0004670412,0.00014864096,0.0017471005],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99894196,0.00019351901,0.00006732929,0.0003223627,0.0003907749,0.00008403754],"domain_scores_gemma":[0.9986619,0.00037508376,0.0001537152,0.00023038313,0.0005236207,0.00005540649],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015610608,0.0011825957,0.0012367624,0.0016174413,0.0010402175,0.001014526,0.0025138874,0.0018085574,0.0011722137],"category_scores_gemma":[0.0048383465,0.00048650507,0.0012153138,0.0020255048,0.0010242352,0.0015699742,0.0012292964,0.0023380027,0.00051790714],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009960984,0.00006940597,0.0009002154,0.000118095406,0.00012669811,0.000059448448,0.00015621274,0.72151124,0.004659708,0.017676998,0.0031579714,0.2514644],"study_design_scores_gemma":[0.0000023820014,0.000009352524,0.00009521143,0.0000063132675,0.0000046044065,0.000014783266,0.0000081074,0.9924615,0.0010416753,0.00590138,0.0004454576,0.000009355246],"about_ca_topic_score_codex":0.022819595,"about_ca_topic_score_gemma":0.024564356,"teacher_disagreement_score":0.022819595,"about_ca_system_score_codex":0.0023984895,"about_ca_system_score_gemma":0.0026682725,"threshold_uncertainty_score":0.04537356},"labels":[],"label_agreement":null},{"id":"W4401769675","doi":"10.18280/isi.290405","title":"Distributed K-means Clustering Using Topological Relationships","year":2024,"lang":"en","type":"article","venue":"Ingénierie des systèmes d information","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Topology (electrical circuits); Computer science; Mathematics; Artificial intelligence; Combinatorics","score_opus":0.054677586115813914,"score_gpt":0.30345414908231905,"score_spread":0.24877656296650513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401769675","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011880965,0.00030736852,0.98547274,0.00014139133,0.000035963974,0.00006965677,0.00006269083,0.00047092163,0.001558214],"genre_scores_gemma":[0.4623018,0.00060073164,0.5332524,0.00009486172,0.00012255245,0.00031030958,0.00056204916,0.00013408391,0.0026212279],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978483,0.0005822259,0.00011018396,0.00053783396,0.0007811436,0.00014037895],"domain_scores_gemma":[0.9980253,0.0006649111,0.0003265923,0.000331848,0.00055829756,0.000093032024],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012376545,0.00087176973,0.001163439,0.0029016442,0.0014562468,0.0018111961,0.0019543727,0.0013390618,0.0013473032],"category_scores_gemma":[0.005630612,0.0005199351,0.0008970893,0.0035443536,0.0011135808,0.0027503292,0.0021996736,0.00087784627,0.00083284016],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032438192,0.00020232007,0.003761937,0.00042362898,0.00025067513,0.00018954046,0.00064156245,0.5257129,0.014951608,0.06649129,0.0049750903,0.3820751],"study_design_scores_gemma":[0.00003284065,0.00009460104,0.0008488863,0.000016460233,0.000041180952,0.00011813781,0.00015566034,0.95665365,0.0034379212,0.034882743,0.0036723183,0.000045547135],"about_ca_topic_score_codex":0.003078608,"about_ca_topic_score_gemma":0.003594615,"teacher_disagreement_score":0.003078608,"about_ca_system_score_codex":0.0012145026,"about_ca_system_score_gemma":0.0016327508,"threshold_uncertainty_score":0.008811891},"labels":[],"label_agreement":null},{"id":"W4401831247","doi":"10.18280/ria.380412","title":"A Comprehensive Study of Ensemble Models to Improve the Performance of Cluster Algorithms","year":2024,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster (spacecraft); Computer science; Algorithm","score_opus":0.0697253121005116,"score_gpt":0.32701849358913027,"score_spread":0.2572931814886187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401831247","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06794039,0.0046942662,0.9212402,0.00036985995,0.00016706815,0.00008876232,0.00011339099,0.00061152066,0.0047744587],"genre_scores_gemma":[0.7265729,0.0037498423,0.26562807,0.00015347081,0.0002980869,0.00017644011,0.00048818483,0.00020813946,0.0027248142],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977335,0.00090254546,0.00009886008,0.0003116806,0.00081040565,0.00014301809],"domain_scores_gemma":[0.9951414,0.0023159096,0.00018355713,0.00063130126,0.0016451846,0.00008270113],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043881806,0.0010948229,0.0015204425,0.0016332616,0.00073349976,0.0014294412,0.001074319,0.00089827477,0.0007433157],"category_scores_gemma":[0.009709549,0.00037763806,0.0013783921,0.0023455692,0.00040098952,0.0026355756,0.00083343557,0.0015607465,0.00026433385],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011082445,0.00014867504,0.005704585,0.00017935445,0.0003889549,0.000054000295,0.00014288288,0.6922716,0.002653391,0.018237816,0.0025984058,0.27750945],"study_design_scores_gemma":[0.0000028787335,0.000073338255,0.00092158135,0.000015495023,0.0000420874,0.000028398563,0.000017669194,0.99375546,0.0011899411,0.0027835665,0.001159073,0.000010472445],"about_ca_topic_score_codex":0.004986626,"about_ca_topic_score_gemma":0.0027744411,"teacher_disagreement_score":0.004986626,"about_ca_system_score_codex":0.0010023959,"about_ca_system_score_gemma":0.001331396,"threshold_uncertainty_score":0.023207188},"labels":[],"label_agreement":null},{"id":"W4401835829","doi":"10.18280/ria.380416","title":"Optimizing the Evaluation of K-means Clustering Using the Weight Product","year":2024,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Product (mathematics); k-means clustering; Computer science; Mathematics; Statistics","score_opus":0.13743874341883913,"score_gpt":0.37872133203304276,"score_spread":0.24128258861420362,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401835829","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14850312,0.0014772968,0.84470946,0.00036886378,0.00013836245,0.00030968245,0.00021607714,0.00216554,0.0021114994],"genre_scores_gemma":[0.45961568,0.00040152122,0.5369596,0.00013358466,0.000040736435,0.000324372,0.0009850615,0.00042396676,0.0011154474],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99446857,0.0021606605,0.000521524,0.00104818,0.0014968792,0.0003041109],"domain_scores_gemma":[0.9909904,0.00425508,0.0006095246,0.00069415924,0.0032438112,0.00020696362],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008994771,0.0023998215,0.0019182395,0.0025838716,0.0013294695,0.0025861915,0.002180518,0.0019915607,0.0009344968],"category_scores_gemma":[0.026745714,0.00088387163,0.0018931996,0.002161281,0.0011129486,0.0025372053,0.001674884,0.0014425768,0.0007339138],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005591461,0.00029151246,0.009913943,0.0005851776,0.0005552267,0.000100673606,0.00049188157,0.70632136,0.008667584,0.0034092055,0.003097385,0.26600683],"study_design_scores_gemma":[0.000020600677,0.00011556863,0.0014106627,0.000028808103,0.000037148504,0.00003885793,0.00010427402,0.99161375,0.0047208876,0.0014006735,0.0004831409,0.000025631209],"about_ca_topic_score_codex":0.012744181,"about_ca_topic_score_gemma":0.0103244735,"teacher_disagreement_score":0.012744181,"about_ca_system_score_codex":0.0018280883,"about_ca_system_score_gemma":0.0027559835,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4402654931","doi":"10.1007/978-3-031-70239-6_27","title":"SERIEMA: A Framework to Enhance Clustering Stability, Compactness, and Separation by Fusing Multimodal Data","year":2024,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Compact space; Cluster analysis; Stability (learning theory); Separation (statistics); Artificial intelligence; Data mining; Machine learning; Mathematics","score_opus":0.03724753028036763,"score_gpt":0.3658450307763174,"score_spread":0.32859750049594977,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402654931","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001785464,0.00015828096,0.9961583,0.000047027897,0.000037340935,0.000023727443,0.00006488915,0.0012845462,0.00044043968],"genre_scores_gemma":[0.05125758,0.00027973688,0.9430624,0.00017403762,0.00009535656,0.00014256855,0.00058751623,0.0008457236,0.0035551006],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99840945,0.0004284921,0.00006360696,0.00040552166,0.00057220884,0.00012081455],"domain_scores_gemma":[0.9982395,0.00054794864,0.00013936225,0.00046079652,0.0004798852,0.00013241381],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035388714,0.001817502,0.0017722744,0.001960161,0.0010200513,0.0025905455,0.0037156385,0.0018096467,0.0048478474],"category_scores_gemma":[0.005709068,0.0008756925,0.002078439,0.0017258773,0.001294106,0.0038440973,0.005090983,0.0026911497,0.0024131956],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007430439,0.00022712229,0.0011989394,0.00044828138,0.00046213815,0.0002626048,0.0005979688,0.17565958,0.059640054,0.062001,0.017986586,0.6807727],"study_design_scores_gemma":[0.000041007308,0.00012184841,0.0004905178,0.000030756113,0.000070175076,0.00016983025,0.000083163635,0.9330352,0.016770815,0.0349229,0.014201641,0.00006208422],"about_ca_topic_score_codex":0.0029288325,"about_ca_topic_score_gemma":0.007604114,"teacher_disagreement_score":0.0048478474,"about_ca_system_score_codex":0.0008181644,"about_ca_system_score_gemma":0.0013945292,"threshold_uncertainty_score":0.01871562},"labels":[],"label_agreement":null},{"id":"W4402728512","doi":"10.3390/math12182884","title":"Imputing Missing Data in One-Shot Devices Using Unsupervised Learning Approach","year":2024,"lang":"en","type":"article","venue":"Mathematics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Oakland University","keywords":"Computer science; Cluster analysis; Missing data; Imputation (statistics); Data mining; Reliability (semiconductor); Crash; DBSCAN; Mobile device; Sampling (signal processing); Machine learning; Artificial intelligence; Fuzzy clustering","score_opus":0.29114768685540554,"score_gpt":0.4016962559138063,"score_spread":0.11054856905840077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402728512","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034902252,0.00028034544,0.9630024,0.0002693578,0.000057010086,0.00006748372,0.00039521474,0.00067956117,0.00034632342],"genre_scores_gemma":[0.57630783,0.00032907136,0.41525614,0.00039026572,0.0001927116,0.00029715715,0.0042544957,0.00016348671,0.0028088766],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956886,0.001732263,0.00028832527,0.0013749198,0.00058524014,0.00033060505],"domain_scores_gemma":[0.98016214,0.012927427,0.0014653291,0.002876402,0.0022446907,0.00032399202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008312138,0.000931561,0.0023094686,0.0019334549,0.0011238192,0.0013597757,0.0055888593,0.0023151713,0.0010387246],"category_scores_gemma":[0.020067742,0.0008453631,0.001814348,0.002712331,0.0012469805,0.0019132476,0.0014696032,0.0025752932,0.0005448574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004895749,0.00043913804,0.02493399,0.00031228797,0.00049357046,0.00039693626,0.0004937934,0.753624,0.0014394266,0.012753828,0.0054812455,0.1991421],"study_design_scores_gemma":[0.000011606613,0.000030106126,0.0011743518,0.00001307357,0.000022024957,0.000035258534,0.000030097253,0.9897664,0.0005421106,0.007897007,0.00045924512,0.000018679311],"about_ca_topic_score_codex":0.009371083,"about_ca_topic_score_gemma":0.010328826,"teacher_disagreement_score":0.009371083,"about_ca_system_score_codex":0.0012431222,"about_ca_system_score_gemma":0.0015615299,"threshold_uncertainty_score":0.04395932},"labels":[],"label_agreement":null},{"id":"W4403170718","doi":"10.1007/s00357-024-09493-z","title":"Mixed-Type Distance Shrinkage and Selection for Clustering via Kernel Metric Learning","year":2024,"lang":"en","type":"article","venue":"Journal of Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"University of British Columbia","keywords":"Cluster analysis; Shrinkage; Selection (genetic algorithm); Mathematics; Kernel (algebra); Artificial intelligence; Metric (unit); Pattern recognition (psychology); Type (biology); Statistics; Computer science; Combinatorics; Geology; Engineering","score_opus":0.04542069984512903,"score_gpt":0.3289202876242669,"score_spread":0.28349958777913786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403170718","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003714703,0.00009850148,0.99566084,0.000069353766,0.000026521631,0.000027758855,0.000024463414,0.00020801082,0.00016974597],"genre_scores_gemma":[0.13789101,0.00022456536,0.8571922,0.00013950511,0.00015389909,0.0003422742,0.00045010148,0.00036839006,0.0032379907],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9951515,0.0023687084,0.00029071863,0.00082270894,0.0011664433,0.00019972172],"domain_scores_gemma":[0.9909478,0.0040112548,0.0005419038,0.0015878131,0.0026073144,0.0003038938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007872747,0.0011073637,0.002477767,0.0015312902,0.001195905,0.0016469904,0.004303803,0.0022038575,0.0021279259],"category_scores_gemma":[0.02221743,0.0010288502,0.0018655875,0.0018169873,0.001509908,0.0033915685,0.0035039214,0.0031627843,0.0015336038],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007978904,0.0003578387,0.002410116,0.0004994697,0.00043353686,0.00013158552,0.00042369255,0.2844838,0.016833313,0.101807825,0.008891745,0.5829292],"study_design_scores_gemma":[0.000021901324,0.00006680658,0.0002389184,0.000012341484,0.00002146158,0.000044306194,0.00001923242,0.98083824,0.0024917922,0.015045536,0.0011793106,0.000020201167],"about_ca_topic_score_codex":0.0017727695,"about_ca_topic_score_gemma":0.0028495383,"teacher_disagreement_score":0.007872747,"about_ca_system_score_codex":0.0011371798,"about_ca_system_score_gemma":0.0019498412,"threshold_uncertainty_score":0.041635573},"labels":[],"label_agreement":null},{"id":"W4403344678","doi":"10.1287/ijoc.2024.0938","title":"A Column Generation Algorithm with Dynamic Constraint Aggregation for Minimum Sum-of-Squares Clustering","year":2025,"lang":"en","type":"preprint","venue":"INFORMS journal on computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Polytechnique Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Column (typography); Column generation; Cluster analysis; Constraint (computer-aided design); Explained sum of squares; Mathematical optimization; Algorithm; Computer science; Least-squares function approximation; Mathematics; Combinatorics; Artificial intelligence; Statistics","score_opus":0.030440264737981348,"score_gpt":0.32754724705520116,"score_spread":0.2971069823172198,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403344678","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004605762,0.00017412567,0.9914205,0.00014107974,0.00006845574,0.000083424355,0.00011914825,0.0010412623,0.0023463517],"genre_scores_gemma":[0.092677094,0.00014302233,0.90277004,0.00019773481,0.000074421754,0.00025694133,0.0006703038,0.00040419344,0.0028061029],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999343,0.00017047903,0.00003172525,0.00013896404,0.00023386408,0.0000820627],"domain_scores_gemma":[0.9988681,0.00050855713,0.00008234333,0.00014795587,0.00031849663,0.0000744499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007311755,0.0014185997,0.0010548267,0.00093552034,0.0008620269,0.001161712,0.0014957642,0.0011250848,0.006464105],"category_scores_gemma":[0.002677581,0.00053062686,0.0008900915,0.0015945361,0.000568983,0.0009292786,0.0014925906,0.0016678046,0.0019681493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012495248,0.00010668219,0.00061643636,0.00015916718,0.00006264648,0.00015730156,0.00014702514,0.73762274,0.006682961,0.0130909495,0.016017279,0.2252118],"study_design_scores_gemma":[0.000020026024,0.000023660668,0.000056854118,0.0000061160263,0.000005727367,0.000032438988,0.000015305786,0.9941016,0.000971538,0.0029681514,0.0017905527,0.000007955097],"about_ca_topic_score_codex":0.0071677957,"about_ca_topic_score_gemma":0.010229808,"teacher_disagreement_score":0.0071677957,"about_ca_system_score_codex":0.0007394105,"about_ca_system_score_gemma":0.0019160139,"threshold_uncertainty_score":0.021624565},"labels":[],"label_agreement":null},{"id":"W4403730407","doi":"10.23952/jnva.8.2024.6.09","title":"Solving net-constrained clustering problem","year":2024,"lang":"en","type":"article","venue":"Journal of Nonlinear and Variational Analysis","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Net (polyhedron); Computer science; Mathematics; Artificial intelligence","score_opus":0.01568507956756476,"score_gpt":0.2976116037681624,"score_spread":0.28192652420059766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403730407","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009342831,0.00016455632,0.9841133,0.00027056443,0.000045606914,0.00006110007,0.00020987565,0.00020262945,0.0055896193],"genre_scores_gemma":[0.2809417,0.0005303194,0.69793195,0.00039948974,0.00012465863,0.00046973006,0.0018004929,0.00043823675,0.017363383],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993057,0.00018102369,0.000025949508,0.00025585317,0.00014478705,0.0000866397],"domain_scores_gemma":[0.99916005,0.0004569192,0.00007019309,0.000069525595,0.00019667279,0.00004670248],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009595997,0.0010820198,0.001269353,0.00070164766,0.0008416397,0.0011777825,0.0016917574,0.0022217594,0.007595622],"category_scores_gemma":[0.0025890872,0.0005038617,0.00087279995,0.0012707514,0.0007459092,0.0014303725,0.0015429418,0.0012953448,0.0009634803],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000039441275,0.00003237974,0.0003135314,0.00015607406,0.000028233007,0.00009519204,0.000044658223,0.93710345,0.0011737711,0.031064598,0.003513479,0.026435174],"study_design_scores_gemma":[0.000007617101,0.00001489072,0.00009259894,0.000009716337,0.000005322286,0.000037829577,0.00002955181,0.98211884,0.00051166094,0.015519092,0.0016451934,0.00000771344],"about_ca_topic_score_codex":0.010240817,"about_ca_topic_score_gemma":0.0088344645,"teacher_disagreement_score":0.010240817,"about_ca_system_score_codex":0.0011974843,"about_ca_system_score_gemma":0.0018400126,"threshold_uncertainty_score":0.025409937},"labels":[],"label_agreement":null},{"id":"W4403793086","doi":"10.1007/s11634-024-00608-3","title":"A sparse exponential family latent block model for co-clustering","year":2024,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Amirkabir University of Technology","keywords":"Exponential family; Cluster analysis; Exponential function; Block (permutation group theory); Computer science; Mathematics; Statistics; Combinatorics","score_opus":0.1106521907296518,"score_gpt":0.39391825692417415,"score_spread":0.2832660661945223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403793086","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003486987,0.0002607537,0.9950067,0.00018511293,0.00004285424,0.000038037353,0.00031447987,0.00018685966,0.00047825623],"genre_scores_gemma":[0.40751758,0.0020517693,0.5575198,0.0006490221,0.00044479262,0.0012419606,0.0056081284,0.0006342183,0.024332674],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9969138,0.0015458305,0.0001262636,0.00063965836,0.0004936605,0.00028076835],"domain_scores_gemma":[0.9931178,0.003804082,0.00047755643,0.0013037693,0.0010007496,0.00029604993],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047492348,0.0010308773,0.002345404,0.0015780224,0.0012815697,0.0019984818,0.005507947,0.0032469048,0.0057915724],"category_scores_gemma":[0.013773707,0.0012311883,0.0021023152,0.0037997242,0.0018565005,0.004344972,0.0029428154,0.004161143,0.004105047],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036498514,0.00018687542,0.0019324301,0.00021772872,0.00019895591,0.00014403477,0.00037618013,0.6011391,0.0021991143,0.29669362,0.0110188965,0.08552806],"study_design_scores_gemma":[0.000011023415,0.000013577731,0.00014235821,0.000010258249,0.000014354157,0.000027311668,0.000014410435,0.9650204,0.00013621249,0.033500988,0.0010952312,0.000013830325],"about_ca_topic_score_codex":0.013713874,"about_ca_topic_score_gemma":0.01758404,"teacher_disagreement_score":0.013713874,"about_ca_system_score_codex":0.0018239425,"about_ca_system_score_gemma":0.0024128053,"threshold_uncertainty_score":0.027268112},"labels":[],"label_agreement":null},{"id":"W4403821441","doi":"10.48550/arxiv.2410.00802","title":"A practical guide to fitting correlation functions from lattice data","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Science and Technology Facilities Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Correlation; Statistical physics; Lattice (music); Computer science; Mathematics; Physics; Geometry","score_opus":0.20254420063618953,"score_gpt":0.3094327341186903,"score_spread":0.1068885334825008,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403821441","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00053131895,0.0012213433,0.8999256,0.0008727011,0.0007418985,0.00017058213,0.015278779,0.060843263,0.020414487],"genre_scores_gemma":[0.0058053355,0.0014039725,0.9158108,0.0013609118,0.00030414588,0.000731001,0.0135507295,0.032279484,0.028753635],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.998329,0.00047589917,0.00014402227,0.00019594593,0.00077478384,0.00008039205],"domain_scores_gemma":[0.993767,0.002837564,0.00021726821,0.0010984706,0.0018079416,0.0002717726],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026752073,0.0019238153,0.0015609106,0.0029950507,0.0013469614,0.0034597258,0.004218271,0.002485371,0.19715579],"category_scores_gemma":[0.01855726,0.0022309036,0.0013485472,0.0039047028,0.0007058524,0.0031847034,0.0016406415,0.0048385374,0.16518559],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053285352,0.00008332421,0.0004277503,0.00059000833,0.00005097448,0.00017288854,0.00014844743,0.009935646,0.002365378,0.035334274,0.7947268,0.15611129],"study_design_scores_gemma":[0.00007667554,0.00003780621,0.0006916837,0.00031689927,0.000014820452,0.00064979336,0.00013358585,0.050452206,0.0029111488,0.11128999,0.8332703,0.00015510556],"about_ca_topic_score_codex":0.004437267,"about_ca_topic_score_gemma":0.0085113235,"teacher_disagreement_score":0.19715579,"about_ca_system_score_codex":0.00077129726,"about_ca_system_score_gemma":0.0019069884,"threshold_uncertainty_score":0.65955186},"labels":[],"label_agreement":null},{"id":"W4404308404","doi":"10.48550/arxiv.2410.17976","title":"metasnf: Meta Clustering with Similarity Network Fusion in R","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Canadian Institutes of Health Research; Campbell Family Mental Health Research Institute; University of Toronto; Hospital for Sick Children; Natural Sciences and Engineering Research Council of Canada; Fondation Brain Canada; Centre for Addiction and Mental Health Foundation; Canada Research Chairs","keywords":"Similarity (geometry); Cluster analysis; Artificial intelligence; Fusion; Computer science; Pattern recognition (psychology); Philosophy; Linguistics","score_opus":0.12070957064102666,"score_gpt":0.22956725980234727,"score_spread":0.1088576891613206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404308404","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001789633,0.0005286302,0.6101317,0.000497121,0.00034555956,0.00022458087,0.029675756,0.3537969,0.0030101475],"genre_scores_gemma":[0.040910296,0.00085776515,0.7381474,0.0007813024,0.00018184529,0.0024932306,0.050133567,0.16129276,0.00520183],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949941,0.001771417,0.00041351348,0.0011384224,0.0013962537,0.00028629805],"domain_scores_gemma":[0.99035954,0.0057184873,0.00078258,0.0017919578,0.0010605921,0.00028679363],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008165547,0.004678395,0.0036939953,0.005054195,0.0014570513,0.004820064,0.005654874,0.001854203,0.07148521],"category_scores_gemma":[0.035169505,0.0026851338,0.0052926335,0.003775096,0.0014340892,0.0041367724,0.0048076077,0.0048585883,0.054796435],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009471124,0.00014194829,0.006498569,0.004036512,0.0027802372,0.00067742285,0.0009651085,0.03846535,0.0066342605,0.051407523,0.7411281,0.14631787],"study_design_scores_gemma":[0.00095524907,0.00020774595,0.0042755515,0.0007871901,0.00087273296,0.001045784,0.00031754147,0.280711,0.019346189,0.1952281,0.49565336,0.00059960294],"about_ca_topic_score_codex":0.0057535353,"about_ca_topic_score_gemma":0.007255606,"teacher_disagreement_score":0.07148521,"about_ca_system_score_codex":0.0016120678,"about_ca_system_score_gemma":0.0036708817,"threshold_uncertainty_score":0.23914182},"labels":[],"label_agreement":null},{"id":"W4404851989","doi":"10.1016/j.nepr.2024.104219","title":"Latent cluster analysis: Another method in search of a research question?","year":2024,"lang":"en","type":"editorial","venue":"Nurse Education in Practice","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Journal of Communication (Canada)","funders":"","keywords":"Cluster (spacecraft); Information retrieval; Psychology; Computer science","score_opus":0.03807364250420745,"score_gpt":0.5090188835520983,"score_spread":0.4709452410478909,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404851989","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000030732914,0.01471945,0.0017176716,0.078112185,0.90459335,0.000022349737,0.000058909212,0.0000544251,0.00069087656],"genre_scores_gemma":[0.00055366015,0.007621548,0.0012806023,0.02656215,0.96068466,0.000080269034,0.000029006907,0.00008432838,0.0031039014],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9711891,0.012757889,0.0031280478,0.0025639606,0.009847741,0.00051338866],"domain_scores_gemma":[0.7880178,0.1497818,0.0053942776,0.005228747,0.044420272,0.0071570976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.042328246,0.0042171036,0.006680249,0.013414062,0.006324599,0.018727046,0.0058361306,0.02593459,0.008670673],"category_scores_gemma":[0.17370366,0.0015997902,0.0036619666,0.0077465996,0.012407864,0.009563518,0.0029121384,0.028434096,0.0073228613],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031207568,0.000010171036,0.00004475538,0.0006405064,0.000068806134,0.00006144845,0.000054222586,0.000028736,0.000024028242,0.0020341491,0.98666346,0.010338408],"study_design_scores_gemma":[0.00016130182,0.00004542049,0.00055423693,0.0025980466,0.00035385173,0.00059824,0.0005008114,0.0011036163,0.00013542439,0.02720689,0.96664476,0.00009739529],"about_ca_topic_score_codex":0.0036288162,"about_ca_topic_score_gemma":0.011396544,"teacher_disagreement_score":0.042328246,"about_ca_system_score_codex":0.005894806,"about_ca_system_score_gemma":0.007854689,"threshold_uncertainty_score":0.22385585},"labels":[],"label_agreement":null},{"id":"W4404973628","doi":"10.3390/a17120551","title":"Ellipsoidal K-Means: An Automatic Clustering Approach for Non-Uniform Data Distributions","year":2024,"lang":"en","type":"article","venue":"Algorithms","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"King Abdulaziz City for Science and Technology","keywords":"Cluster analysis; Ellipsoid; Computer science; Mathematics; Artificial intelligence; Physics","score_opus":0.06264586508325108,"score_gpt":0.3495320024963194,"score_spread":0.2868861374130683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404973628","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019010886,0.000104241364,0.9969566,0.00003790091,0.000013350582,0.00003163241,0.000031176944,0.0006099939,0.0003140276],"genre_scores_gemma":[0.0841183,0.00025956435,0.91317755,0.00012518371,0.000040553776,0.00026734828,0.00045837444,0.00044920822,0.0011038651],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99644107,0.0010966663,0.00022935291,0.0008495914,0.0011897318,0.00019351378],"domain_scores_gemma":[0.99713886,0.0009146067,0.00032002575,0.0005510734,0.0009823252,0.00009315687],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027437222,0.0016284164,0.0016846877,0.003255447,0.0014383051,0.0014633887,0.0031512743,0.001627759,0.001268063],"category_scores_gemma":[0.0066732448,0.0011176042,0.0017247365,0.0028207141,0.0013388047,0.0019110853,0.0023988124,0.0019087545,0.000975153],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028720082,0.00011190732,0.0020788158,0.000445556,0.0002804709,0.00013422221,0.000755955,0.5348838,0.024676705,0.026530912,0.00979535,0.40001914],"study_design_scores_gemma":[0.000012112926,0.000023263396,0.0003313905,0.000019714993,0.000014369826,0.000057008612,0.000044177414,0.981729,0.004899774,0.009615311,0.003222113,0.00003183209],"about_ca_topic_score_codex":0.0076035983,"about_ca_topic_score_gemma":0.011299825,"teacher_disagreement_score":0.0076035983,"about_ca_system_score_codex":0.001361378,"about_ca_system_score_gemma":0.002636216,"threshold_uncertainty_score":0.015118659},"labels":[],"label_agreement":null},{"id":"W4405631822","doi":"10.1016/j.rines.2024.100051","title":"Spatial characterisation of groundwater systems using fuzzy c-mean clustering: A multi-parameter approach in crystalline aquifers","year":2024,"lang":"en","type":"article","venue":"Results in Earth Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Energy","funders":"","keywords":"Aquifer; Groundwater; Cluster analysis; Fuzzy logic; Environmental science; Groundwater resources; Water resource management; Hydrology (agriculture); Geology; Data mining; Soil science; Computer science; Geotechnical engineering; Artificial intelligence","score_opus":0.07866996826638287,"score_gpt":0.33222634178589294,"score_spread":0.2535563735195101,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405631822","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5265643,0.00033658784,0.47023457,0.0001305859,0.0000144615815,0.00018689966,0.00043012775,0.00034236113,0.0017602354],"genre_scores_gemma":[0.9165428,0.00008302342,0.08263022,0.000010931914,0.0000064868104,0.000058034853,0.00029615973,0.000019980325,0.00035233086],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995572,0.00009145233,0.000041922194,0.000113733004,0.00013139266,0.00006425467],"domain_scores_gemma":[0.9993393,0.00025243175,0.00010638345,0.00004792143,0.00022355343,0.000030355011],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00084950426,0.0004240523,0.00050069916,0.0048136683,0.0007543194,0.0012545878,0.0007947982,0.00075454946,0.00048120593],"category_scores_gemma":[0.0018345298,0.00025874816,0.0008015591,0.0028609885,0.0004716792,0.0005992102,0.0005078653,0.00029358687,0.000089115674],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022362103,0.00011864016,0.04180609,0.00022916606,0.00016223782,0.00029083327,0.0009810831,0.7779115,0.016342692,0.0049841576,0.0008773866,0.15607257],"study_design_scores_gemma":[0.0000034480606,0.000024808878,0.012488752,0.000013136537,0.000020089192,0.00005100447,0.00035570952,0.98316056,0.001991595,0.0015743533,0.0002843583,0.000032151405],"about_ca_topic_score_codex":0.031665005,"about_ca_topic_score_gemma":0.01988997,"teacher_disagreement_score":0.031665005,"about_ca_system_score_codex":0.0011033745,"about_ca_system_score_gemma":0.0009940701,"threshold_uncertainty_score":0.0629614},"labels":[],"label_agreement":null},{"id":"W4405765923","doi":"10.4230/lipics.esa.2022.56","title":"Improved Polynomial-Time Approximations for Clustering with Minimum Sum of Radii or Diameters","year":2022,"lang":"en","type":"preprint","venue":"DROPS (Schloss Dagstuhl – Leibniz Center for Informatics)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Algorithm; Mathematics; Combinatorics; Computer science; Statistics","score_opus":0.026285380078038455,"score_gpt":0.2950712382600269,"score_spread":0.26878585818198847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405765923","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012408086,0.00081941177,0.9792586,0.0005295644,0.00014870941,0.0000980606,0.00029061048,0.001783214,0.0046638213],"genre_scores_gemma":[0.1507498,0.00055055687,0.8413237,0.00038314986,0.00018696005,0.00027893813,0.001068186,0.0007528174,0.0047058607],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9955167,0.0010672131,0.00023013397,0.0008329414,0.0016647774,0.00068827294],"domain_scores_gemma":[0.99438864,0.0022485785,0.000434677,0.0018399847,0.0007621671,0.00032592213],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031120337,0.0023232873,0.0024860366,0.001837988,0.0013454811,0.002855286,0.004319447,0.002557053,0.005711228],"category_scores_gemma":[0.015833631,0.0009949052,0.00216172,0.0033904004,0.001313645,0.004880109,0.00370473,0.0048709963,0.0032756424],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087553845,0.0002861789,0.0010828234,0.00041490595,0.00011726701,0.00013781332,0.00040702787,0.6902141,0.008751944,0.10582566,0.020101933,0.17178479],"study_design_scores_gemma":[0.000060936825,0.000055944587,0.00017359252,0.000026877755,0.00001830664,0.00007907771,0.000051838077,0.94747144,0.0021110813,0.046107672,0.0038239122,0.000019284122],"about_ca_topic_score_codex":0.006060365,"about_ca_topic_score_gemma":0.008078489,"teacher_disagreement_score":0.006060365,"about_ca_system_score_codex":0.0036314533,"about_ca_system_score_gemma":0.0023031048,"threshold_uncertainty_score":0.026348174},"labels":[],"label_agreement":null},{"id":"W4406141871","doi":"10.1137/1.9781611978322.102","title":"The Johnson-Lindenstrauss Lemma for Clustering and Subspace Approximation: From Coresets to Dimension Reduction","year":2025,"lang":"en","type":"book-chapter","venue":"Society for Industrial and Applied Mathematics eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Sciences and Engineering Research Council of Canada","funders":"","keywords":"Lemma (botany); Dimensionality reduction; Subspace topology; Dimension (graph theory); Mathematics; Reduction (mathematics); Cluster analysis; Combinatorics; Pure mathematics; Computer science; Mathematical analysis; Artificial intelligence; Geometry; Statistics","score_opus":0.06832685645952044,"score_gpt":0.29021760091077076,"score_spread":0.22189074445125032,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406141871","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030028606,0.0014884577,0.9459703,0.00086836243,0.00017278179,0.00006302126,0.00010786784,0.00047368815,0.020826925],"genre_scores_gemma":[0.42737344,0.0026974038,0.55202967,0.0008132906,0.00035992474,0.0002351626,0.00043786946,0.0008581742,0.015195034],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976654,0.00076100766,0.00007268171,0.00035150535,0.0009672954,0.00018217423],"domain_scores_gemma":[0.9943639,0.0031236256,0.00036083255,0.0013437846,0.0005903339,0.00021755841],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003330098,0.001244071,0.0015047101,0.0015624819,0.0012656389,0.00258432,0.0018505703,0.0012898006,0.0038697917],"category_scores_gemma":[0.010820143,0.0006213884,0.0013911308,0.0021134766,0.0037206358,0.005148274,0.003930707,0.0044499384,0.0010446396],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014671801,0.00007358916,0.0005386314,0.0002076215,0.000047689322,0.00007585493,0.00024760852,0.0788535,0.008725234,0.83179283,0.006531201,0.072759576],"study_design_scores_gemma":[0.000026845117,0.00016036564,0.00067537813,0.00005952386,0.00003275391,0.00024978808,0.00018454867,0.47321644,0.019041033,0.49602756,0.010281237,0.000044577653],"about_ca_topic_score_codex":0.001619369,"about_ca_topic_score_gemma":0.0017731552,"teacher_disagreement_score":0.0038697917,"about_ca_system_score_codex":0.0019017416,"about_ca_system_score_gemma":0.0010864685,"threshold_uncertainty_score":0.017611444},"labels":[],"label_agreement":null},{"id":"W4406170506","doi":"10.1109/tfuzz.2025.3526978","title":"A Vertical Federated Multiview Fuzzy Clustering Method for Incomplete Data","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Fuzzy Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"China Scholarship Council; National Natural Science Foundation of China","keywords":"Computer science; Cluster analysis; Raw data; Flexibility (engineering); Data mining; Process (computing); Fuzzy clustering; Fuzzy logic; Machine learning; Distributed computing; Artificial intelligence","score_opus":0.08335018997250195,"score_gpt":0.3831114108535232,"score_spread":0.29976122088102125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406170506","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027681938,0.00008910649,0.9965257,0.000037081085,0.00001578756,0.000018314431,0.000026551606,0.00018709425,0.00033214095],"genre_scores_gemma":[0.2341489,0.00025992986,0.7616657,0.000118122625,0.000062451385,0.0001318437,0.000432204,0.00012553671,0.0030553667],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99893886,0.00015642511,0.00007101229,0.00030612125,0.0003927366,0.00013488534],"domain_scores_gemma":[0.9991628,0.00016248942,0.00009207579,0.00014374856,0.00038345857,0.00005546929],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013747534,0.0007419403,0.0012251291,0.0015110832,0.0009998307,0.0010452295,0.002427102,0.0013684786,0.0017048742],"category_scores_gemma":[0.0025172064,0.00045048242,0.0013841367,0.001603003,0.00058936776,0.0016291521,0.0016780747,0.0011604175,0.00053244724],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019554305,0.00007948242,0.0011986669,0.00016247352,0.00013304557,0.00013238602,0.00033112592,0.47569188,0.01477392,0.024352554,0.003972467,0.47897646],"study_design_scores_gemma":[0.0000052779233,0.000020545673,0.000138363,0.0000072295074,0.000010883589,0.00004561496,0.00003380425,0.99301106,0.0020165325,0.0036835785,0.0010146054,0.000012569155],"about_ca_topic_score_codex":0.014704871,"about_ca_topic_score_gemma":0.013603128,"teacher_disagreement_score":0.014704871,"about_ca_system_score_codex":0.0013368177,"about_ca_system_score_gemma":0.002438588,"threshold_uncertainty_score":0.029238582},"labels":[],"label_agreement":null},{"id":"W4406266350","doi":"10.1109/vtc2024-fall63153.2024.10757898","title":"A Robust Clustering Scheme for Vehicular Communication Networks","year":2024,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Computer science; Scheme (mathematics); Cluster analysis; Computer network; Distributed computing; Artificial intelligence; Mathematics","score_opus":0.052503773740987955,"score_gpt":0.31585285650315836,"score_spread":0.2633490827621704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406266350","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043514497,0.0007143349,0.991452,0.0001624588,0.00015714005,0.0001252611,0.000114015646,0.0007013279,0.002222022],"genre_scores_gemma":[0.43251914,0.0015895201,0.5555781,0.00022177114,0.00017445239,0.00046165916,0.0008208687,0.00017200313,0.00846259],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991047,0.00021937714,0.000062450075,0.00017598423,0.00035766716,0.00007988118],"domain_scores_gemma":[0.9993636,0.00008949764,0.0000939338,0.00015298638,0.00026717354,0.000032790354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00074733864,0.00068940624,0.0006955426,0.0009901673,0.000947416,0.00076328794,0.0020858075,0.0010038276,0.0013209941],"category_scores_gemma":[0.0020574294,0.00022859297,0.0006034611,0.001364991,0.0005190369,0.0010996374,0.0012050408,0.00088871224,0.0010448293],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019283577,0.000041341966,0.0003663329,0.00027923542,0.000113901544,0.00018119944,0.00019510924,0.5927626,0.030389879,0.08730835,0.012464766,0.27570453],"study_design_scores_gemma":[0.00003071377,0.0001458516,0.00026685468,0.000027862347,0.000033574,0.00024233847,0.0000590857,0.9445329,0.008435099,0.018565902,0.027595889,0.000063956555],"about_ca_topic_score_codex":0.0034858203,"about_ca_topic_score_gemma":0.003127311,"teacher_disagreement_score":0.0034858203,"about_ca_system_score_codex":0.0013361317,"about_ca_system_score_gemma":0.0011169688,"threshold_uncertainty_score":0.009694338},"labels":[],"label_agreement":null},{"id":"W4407352320","doi":"10.2139/ssrn.5131398","title":"Fast Online $\\Ell_0$ Elastic Net Subspace Clustering Via a Novel Dictionary Update Strategy","year":2025,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Elastic net regularization; Subspace topology; Cluster analysis; Computer science; Net (polyhedron); Artificial intelligence; Pattern recognition (psychology); Mathematics; Feature selection","score_opus":0.017270484703722992,"score_gpt":0.298309612447457,"score_spread":0.281039127743734,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407352320","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010072887,0.00022997837,0.98062474,0.00027631113,0.0002516288,0.00011093751,0.0003058497,0.003414479,0.0047132233],"genre_scores_gemma":[0.14895672,0.00022846935,0.83082676,0.00037197088,0.00022743328,0.00021826329,0.0015461202,0.0006557632,0.01696851],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99916387,0.000121970384,0.000044342272,0.00015664086,0.00042107463,0.000092117865],"domain_scores_gemma":[0.9989625,0.00025591365,0.00005064571,0.00032501054,0.00030962986,0.00009627122],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056605716,0.0011536765,0.0013222886,0.0010518428,0.0010198841,0.001704526,0.0023765517,0.0013671487,0.015399893],"category_scores_gemma":[0.0026838228,0.0006146175,0.00052997936,0.0016684558,0.0006525304,0.002287086,0.0033323953,0.0014661094,0.008078913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006899718,0.00025068392,0.0006278013,0.00015905412,0.00007951526,0.00014311778,0.000102903636,0.07757324,0.015869802,0.016351081,0.033804208,0.8543486],"study_design_scores_gemma":[0.000054054646,0.000057282607,0.00024688017,0.000008976108,0.0000171251,0.00012702169,0.000056704128,0.9781334,0.0064961594,0.009230298,0.005549099,0.000022967022],"about_ca_topic_score_codex":0.0059224893,"about_ca_topic_score_gemma":0.0146289775,"teacher_disagreement_score":0.015399893,"about_ca_system_score_codex":0.0004777589,"about_ca_system_score_gemma":0.0012651044,"threshold_uncertainty_score":0.051517785},"labels":[],"label_agreement":null},{"id":"W4407641753","doi":"10.32388/6yejp9","title":"Review of: \"Fixed-Sized Clusters k-Means\"","year":2025,"lang":"en","type":"peer-review","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"","keywords":"Mathematics","score_opus":0.03736909423499118,"score_gpt":0.3719601192432709,"score_spread":0.3345910250082797,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407641753","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001057379,0.6443605,0.0683941,0.09900623,0.15818544,0.00024259348,0.00049702945,0.00066253764,0.027594293],"genre_scores_gemma":[0.018731562,0.7548316,0.046342503,0.042694025,0.09047178,0.00032442782,0.0009974095,0.00096706924,0.04463976],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99507755,0.0012621175,0.0003996303,0.0006628669,0.0024381678,0.00015973182],"domain_scores_gemma":[0.9739991,0.0060173203,0.00070632325,0.0013268339,0.017199788,0.0007507014],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007972559,0.0006227725,0.0009877858,0.0034981922,0.0011275456,0.0028014185,0.0026748576,0.0026295285,0.0057681007],"category_scores_gemma":[0.03248337,0.00053345633,0.00053078466,0.005272944,0.0030478232,0.00389671,0.0014533645,0.0034841595,0.006713548],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000066960376,0.000018315202,0.00011802263,0.003199359,0.000055006713,0.00004098252,0.000101639154,0.0009685264,0.0004389164,0.037173953,0.5685803,0.38923803],"study_design_scores_gemma":[0.00000636278,0.00002303129,0.00021504419,0.00073031045,0.000016579139,0.000118822914,0.000028094688,0.00034187743,0.00032558612,0.003995603,0.9941776,0.000021095495],"about_ca_topic_score_codex":0.004477183,"about_ca_topic_score_gemma":0.004575862,"teacher_disagreement_score":0.007972559,"about_ca_system_score_codex":0.0028649839,"about_ca_system_score_gemma":0.004823517,"threshold_uncertainty_score":0.042163372},"labels":[],"label_agreement":null},{"id":"W4407827487","doi":"10.1109/tcyb.2025.3537108","title":"Automated Cluster Elimination Guided by High-Density Points","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Cybernetics","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Chengdu Science and Technology Program; Fundamental Research Funds for the Central Universities; Guangxi Key Research and Development Program; National Natural Science Foundation of China","keywords":"Cluster (spacecraft); Computer science; Operating system","score_opus":0.013430609403275243,"score_gpt":0.295411280232498,"score_spread":0.2819806708292228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407827487","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033264685,0.00025911766,0.9630977,0.0001112579,0.000030094905,0.00011641994,0.00009448217,0.0015557006,0.001470596],"genre_scores_gemma":[0.29469046,0.00020502546,0.7020195,0.00012474299,0.00002016601,0.00019512928,0.00064138864,0.00033055368,0.0017729858],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99749714,0.0004378084,0.00013488259,0.000608596,0.0011007037,0.00022083422],"domain_scores_gemma":[0.99490213,0.001854664,0.0004295182,0.00072422886,0.0019249357,0.00016447114],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026490416,0.0013309423,0.0014851488,0.0032616681,0.0020719878,0.0020809497,0.0032727288,0.0015792045,0.0017860874],"category_scores_gemma":[0.008550174,0.00069245556,0.0014047467,0.002207749,0.0013106756,0.0021033029,0.0027395892,0.0015324164,0.0012455271],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005826621,0.00017218819,0.0056716376,0.0004740286,0.00018733708,0.00036325908,0.00093557575,0.43105677,0.024435556,0.017533233,0.006316847,0.51227087],"study_design_scores_gemma":[0.000020333524,0.000025992395,0.00079303706,0.00002242574,0.000020335949,0.00009939028,0.00010109611,0.9796537,0.010990219,0.006942261,0.0013048793,0.000026425017],"about_ca_topic_score_codex":0.011642856,"about_ca_topic_score_gemma":0.011131247,"teacher_disagreement_score":0.011642856,"about_ca_system_score_codex":0.0014843724,"about_ca_system_score_gemma":0.0034589928,"threshold_uncertainty_score":0.023150146},"labels":[],"label_agreement":null},{"id":"W4407937849","doi":"10.1109/tpami.2025.3545573","title":"Gauging-: A Non-Parametric Hierarchical Clustering Algorithm","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Cluster analysis; Artificial intelligence; Hierarchical clustering; Algorithm; Pattern recognition (psychology); Parametric statistics; Data mining; Mathematics; Statistics","score_opus":0.017910348891412187,"score_gpt":0.3089522056968399,"score_spread":0.2910418568054277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407937849","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034905567,0.00009938835,0.9947314,0.000063871994,0.000021533755,0.000072302006,0.000054354005,0.0010878975,0.00037864168],"genre_scores_gemma":[0.0765961,0.000102516424,0.92096764,0.0001385466,0.000042735777,0.00018013104,0.00040736084,0.0003002535,0.0012647734],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983442,0.00043717923,0.0001113336,0.00043951778,0.0005381069,0.00012974006],"domain_scores_gemma":[0.9982199,0.00062726584,0.00018905324,0.00036193847,0.00050605286,0.000095717114],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024394651,0.0010856501,0.0015711032,0.0016819135,0.001082872,0.0015200974,0.004140939,0.0016641144,0.0022992787],"category_scores_gemma":[0.006135969,0.00064369326,0.001182691,0.0018099232,0.0009866472,0.0027237183,0.002876661,0.0018066122,0.0013176674],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027217876,0.00009768626,0.0020223032,0.00018981725,0.00016859158,0.00011191468,0.00044516881,0.19403325,0.0147022735,0.018407144,0.008057534,0.76149213],"study_design_scores_gemma":[0.000029349263,0.00007658231,0.0004545462,0.000014810395,0.000019864896,0.00008494175,0.00007377151,0.9752358,0.0044639194,0.01543978,0.0040670414,0.000039616647],"about_ca_topic_score_codex":0.004305034,"about_ca_topic_score_gemma":0.0057159932,"teacher_disagreement_score":0.004305034,"about_ca_system_score_codex":0.0009443099,"about_ca_system_score_gemma":0.002163521,"threshold_uncertainty_score":0.012901306},"labels":[],"label_agreement":null},{"id":"W4408062361","doi":"10.5220/0013105300003893","title":"Data Clustering Using Mother Tree Optimization","year":2025,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Computer science; Cluster analysis; Tree (set theory); Data mining; Artificial intelligence; Mathematics","score_opus":0.08909702423773201,"score_gpt":0.3795694031184865,"score_spread":0.2904723788807545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408062361","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037560614,0.00011072369,0.9950021,0.00008884706,0.000029082681,0.000032571803,0.000055733646,0.00051746075,0.00040742042],"genre_scores_gemma":[0.105419084,0.00017352674,0.8905812,0.00010105087,0.000051334875,0.00019502954,0.00051644474,0.00036035216,0.0026018838],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985287,0.0005099999,0.00009174782,0.0003164227,0.00043318584,0.000119917735],"domain_scores_gemma":[0.9974802,0.0011448452,0.00014887261,0.00042583482,0.0006987891,0.000101410056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022997262,0.0006862286,0.00192113,0.001706308,0.0012322665,0.0013460135,0.0020159255,0.0014657596,0.003263304],"category_scores_gemma":[0.006610024,0.0008433118,0.0016972149,0.0023232664,0.0005880318,0.0016458781,0.0021913687,0.0017553531,0.0015056509],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031849786,0.0001373912,0.0015242862,0.00015865926,0.0001912061,0.00010491808,0.00018999007,0.5136383,0.0064107897,0.025296051,0.0090792375,0.44295064],"study_design_scores_gemma":[0.0000070159717,0.000014943858,0.00006868913,0.0000036612282,0.000008234276,0.000014882191,0.000011733888,0.9939016,0.00086015405,0.0043922234,0.0007133686,0.000003616155],"about_ca_topic_score_codex":0.005095862,"about_ca_topic_score_gemma":0.0042273724,"teacher_disagreement_score":0.005095862,"about_ca_system_score_codex":0.001327124,"about_ca_system_score_gemma":0.002054112,"threshold_uncertainty_score":0.012162268},"labels":[],"label_agreement":null},{"id":"W4408609303","doi":"10.1109/tsmc.2025.3547350","title":"An Adaptive Federated Fuzzy C-Means Clustering With Nonindependently and Identically Distributed Data","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Systems Man and Cybernetics Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Natural Science Foundation of China","keywords":"Independent and identically distributed random variables; Computer science; Cluster analysis; Fuzzy logic; Fuzzy clustering; Data mining; Artificial intelligence; Mathematics; Statistics; Random variable","score_opus":0.026326408643167022,"score_gpt":0.2855842154885161,"score_spread":0.25925780684534905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408609303","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01755896,0.00044928156,0.98002464,0.00026968966,0.00007422778,0.000056337514,0.00011179685,0.0006963947,0.0007587569],"genre_scores_gemma":[0.41802692,0.00040523,0.57694286,0.000455337,0.00009681942,0.000194656,0.00061654404,0.0001470058,0.0031146277],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981183,0.00041962552,0.00009634741,0.00067291997,0.00052058196,0.00017229888],"domain_scores_gemma":[0.99751675,0.00079466373,0.00020483468,0.0003918276,0.0009691946,0.00012272096],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002784527,0.0013147515,0.0019493836,0.0013981715,0.0015841359,0.0015051257,0.003988393,0.0025571624,0.0011059053],"category_scores_gemma":[0.008616953,0.0006382345,0.0015302058,0.0017879163,0.0011891072,0.0019795378,0.0020011466,0.001982033,0.0004744912],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024260329,0.00009842103,0.0011904659,0.00011427627,0.0001395084,0.00014232263,0.00018063285,0.8479474,0.0031504221,0.0062470697,0.0037230805,0.13682373],"study_design_scores_gemma":[0.0000057504785,0.000011017656,0.00008167745,0.0000052396917,0.0000060444772,0.000018185083,0.000009687848,0.9974584,0.000495979,0.0016477504,0.00025141038,0.000008904214],"about_ca_topic_score_codex":0.032790866,"about_ca_topic_score_gemma":0.022988902,"teacher_disagreement_score":0.032790866,"about_ca_system_score_codex":0.0019365629,"about_ca_system_score_gemma":0.003407391,"threshold_uncertainty_score":0.06520003},"labels":[],"label_agreement":null},{"id":"W4408793869","doi":"10.1109/swc62898.2024.00216","title":"Clustering Algorithms with Balanced Weights for Geographic Data Processing","year":2024,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"Chinese Academy of Sciences","keywords":"Computer science; Cluster analysis; Data mining; Algorithm; Artificial intelligence","score_opus":0.04745828689423992,"score_gpt":0.3390360936300708,"score_spread":0.2915778067358309,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408793869","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01069486,0.00022100356,0.9876016,0.00013316715,0.0000405021,0.00006887169,0.00006854681,0.00044604047,0.0007254632],"genre_scores_gemma":[0.21503286,0.00046221886,0.7806494,0.00012905367,0.000113843416,0.00029454078,0.0007257324,0.0002377643,0.002354608],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99869734,0.0003002025,0.0000777488,0.0002946637,0.0005142546,0.000115724564],"domain_scores_gemma":[0.9983944,0.00039187665,0.00019101403,0.00029510242,0.0006522276,0.000075285956],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015585397,0.0010285112,0.0009454808,0.002394176,0.0013200432,0.0012685958,0.0019182657,0.001234579,0.0012768296],"category_scores_gemma":[0.0054968107,0.00056764565,0.0007788526,0.004078231,0.00059886824,0.0024626076,0.0014859155,0.0013190642,0.0009098129],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024332039,0.00015355014,0.0014768033,0.00012339487,0.00012155889,0.00006489376,0.00021028907,0.68263066,0.010818934,0.02100073,0.004960541,0.2781953],"study_design_scores_gemma":[0.000013966503,0.000024455516,0.00030052997,0.000006165016,0.000008619616,0.000025927091,0.000036266243,0.9855951,0.0018958438,0.010211873,0.0018694033,0.000011787294],"about_ca_topic_score_codex":0.0068845074,"about_ca_topic_score_gemma":0.007886017,"teacher_disagreement_score":0.0068845074,"about_ca_system_score_codex":0.0014646427,"about_ca_system_score_gemma":0.0015935063,"threshold_uncertainty_score":0.013688862},"labels":[],"label_agreement":null},{"id":"W4409363136","doi":"10.1609/aaai.v39i20.35385","title":"ICE-T: Interactions-aware Cross-column Contrastive Embedding for Heterogeneous Tabular Datasets","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Mitacs","keywords":"Column (typography); Computer science; Embedding; Artificial intelligence","score_opus":0.06925681773120876,"score_gpt":0.4013077380625092,"score_spread":0.33205092033130046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409363136","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041725602,0.001199982,0.9378018,0.0005071627,0.00019707554,0.00025921303,0.0037038792,0.012273485,0.002331753],"genre_scores_gemma":[0.3059525,0.0006958682,0.6584182,0.0009092744,0.00022253783,0.00070834346,0.02524421,0.0018155171,0.0060335337],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9983656,0.0005036442,0.000116486655,0.0005405995,0.00033583256,0.00013776745],"domain_scores_gemma":[0.9952206,0.0016716471,0.00040461868,0.0019197019,0.00057393283,0.00020950421],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002448025,0.002301681,0.0016091259,0.0021098775,0.00086255104,0.0026792996,0.003179165,0.0016984976,0.005139333],"category_scores_gemma":[0.012674917,0.0006610592,0.0018308312,0.0031141425,0.0012071293,0.0068489136,0.0038581667,0.003573985,0.0028760978],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010328621,0.0006970067,0.0092467535,0.0007065379,0.00052320375,0.00034661058,0.00057974545,0.20186773,0.019172058,0.03153003,0.06586318,0.66843426],"study_design_scores_gemma":[0.00007796839,0.00019789946,0.0008123037,0.000045106313,0.000049911858,0.00016123164,0.00016842506,0.95280135,0.00556071,0.033392314,0.0066801487,0.000052526724],"about_ca_topic_score_codex":0.003066876,"about_ca_topic_score_gemma":0.00609398,"teacher_disagreement_score":0.005139333,"about_ca_system_score_codex":0.0011743126,"about_ca_system_score_gemma":0.0015606581,"threshold_uncertainty_score":0.017192721},"labels":[],"label_agreement":null},{"id":"W4409885889","doi":"10.1007/s10586-024-04981-8","title":"SPINEX-clustering: similarity-based predictions with explainable neighbors exploration for clustering problems","year":2025,"lang":"en","type":"article","venue":"Cluster Computing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"Clemson University","keywords":"Computer science; Cluster analysis; Similarity (geometry); Data mining; Artificial intelligence; Information retrieval; Image (mathematics)","score_opus":0.03347734993308603,"score_gpt":0.3032434677044574,"score_spread":0.2697661177713714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409885889","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03152567,0.00044275913,0.9635216,0.0004908072,0.00005986396,0.000164699,0.00021176145,0.0017817143,0.0018010455],"genre_scores_gemma":[0.36246505,0.0003117215,0.63262105,0.00030975437,0.00011193873,0.00042544198,0.0011011331,0.00037653357,0.0022773582],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988211,0.00045984334,0.000050756127,0.0002474301,0.00034041717,0.00008046443],"domain_scores_gemma":[0.99678934,0.001997332,0.00023801747,0.00041385638,0.00043998647,0.00012145838],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002193979,0.0013203187,0.0014199222,0.001452127,0.0011358811,0.0014871625,0.002366071,0.0017134487,0.0027514112],"category_scores_gemma":[0.009187346,0.00052063115,0.00097387814,0.0013754135,0.0012251723,0.0022624268,0.002744128,0.0021314197,0.0007902537],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018941659,0.00013682542,0.002662188,0.0001861404,0.000106307496,0.00008883434,0.00025794774,0.82323843,0.0015144579,0.03314367,0.0060281977,0.1324475],"study_design_scores_gemma":[0.000014412129,0.000032557436,0.00010378079,0.00000817487,0.000005153153,0.000016915805,0.000024260253,0.98351824,0.0004650804,0.015201042,0.00060483126,0.0000056141307],"about_ca_topic_score_codex":0.0048649744,"about_ca_topic_score_gemma":0.0057440503,"teacher_disagreement_score":0.0048649744,"about_ca_system_score_codex":0.0012185867,"about_ca_system_score_gemma":0.0022880677,"threshold_uncertainty_score":0.011602998},"labels":[],"label_agreement":null},{"id":"W4410398281","doi":"10.32473/flairs.38.1.138970","title":"Flexible Dirichlet Mixture Model for Multi-modal data Clustering","year":2025,"lang":"en","type":"article","venue":"Proceedings of the ... International Florida Artificial Intelligence Research Society Conference","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Modal; Cluster analysis; Latent Dirichlet allocation; Mixture model; Computer science; Dirichlet distribution; Data mining; Mathematics; Artificial intelligence; Topic model; Materials science","score_opus":0.3601967099388323,"score_gpt":0.46848168253118033,"score_spread":0.10828497259234804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410398281","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028982447,0.0005626147,0.99539006,0.00021881824,0.000050988117,0.000047948357,0.00013778206,0.00027657312,0.0004170707],"genre_scores_gemma":[0.3053878,0.002094825,0.6831809,0.000577728,0.00037965731,0.00081134104,0.002239205,0.00039358117,0.004934823],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99375325,0.003118668,0.0003027449,0.0012978553,0.001158871,0.0003686486],"domain_scores_gemma":[0.99463296,0.0034031721,0.00035372018,0.0007737197,0.00066447863,0.0001719709],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007832869,0.0015736193,0.0023867665,0.0029119279,0.0015135912,0.0027354755,0.003991283,0.00301669,0.00259586],"category_scores_gemma":[0.015750049,0.0011588986,0.0031762377,0.0037879343,0.0020023412,0.00427793,0.003271738,0.004190778,0.0017251602],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003939426,0.00016428583,0.0033067702,0.00047113164,0.00044142228,0.0002498603,0.0007788814,0.71244234,0.004109185,0.106806286,0.00699818,0.1638378],"study_design_scores_gemma":[0.000010914468,0.000017827246,0.00031201038,0.000024085186,0.000019697014,0.00006788918,0.00004587876,0.94690335,0.00065313175,0.050141133,0.0017656046,0.000038555947],"about_ca_topic_score_codex":0.0069306167,"about_ca_topic_score_gemma":0.00677555,"teacher_disagreement_score":0.007832869,"about_ca_system_score_codex":0.0020621663,"about_ca_system_score_gemma":0.0017565615,"threshold_uncertainty_score":0.04142469},"labels":[],"label_agreement":null},{"id":"W4410920770","doi":"10.1007/s11634-025-00637-6","title":"Mixed-type kernel-based quantification of similarity for clustering","year":2025,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus","funders":"University of British Columbia","keywords":"Kernel (algebra); Similarity (geometry); Cluster analysis; Artificial intelligence; Computer science; Pattern recognition (psychology); Type (biology); Mathematics; Data mining; Combinatorics; Geology","score_opus":0.09468252969729223,"score_gpt":0.3986018150346892,"score_spread":0.303919285337397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410920770","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003950176,0.00019701543,0.9949944,0.000045258566,0.0000267404,0.000023508986,0.000048018603,0.00018237483,0.0005325302],"genre_scores_gemma":[0.24529389,0.00040764015,0.7498999,0.00012217174,0.00012909004,0.00021325203,0.0003725477,0.00031526873,0.0032461635],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99498993,0.0017941638,0.00041683327,0.00083682843,0.0017287189,0.00023346931],"domain_scores_gemma":[0.99213105,0.0024299582,0.00072956784,0.002188446,0.0022030212,0.0003178754],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041791056,0.0008677308,0.0013969677,0.0032772245,0.0008860683,0.0034340057,0.0027566063,0.0017247489,0.0024336723],"category_scores_gemma":[0.016226893,0.00055075885,0.0014414595,0.0037125049,0.0017433023,0.006097429,0.0034153748,0.002599036,0.0013961683],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00061741634,0.00022472035,0.0031179953,0.000748225,0.00037476904,0.00013292013,0.00057818706,0.09166138,0.050273433,0.51747686,0.00507403,0.32971996],"study_design_scores_gemma":[0.000016884942,0.000118852135,0.0013261051,0.000041737378,0.000053755655,0.00033311202,0.00009355048,0.87022537,0.0135590695,0.109232105,0.004924844,0.00007460771],"about_ca_topic_score_codex":0.0010051233,"about_ca_topic_score_gemma":0.0009494374,"teacher_disagreement_score":0.0041791056,"about_ca_system_score_codex":0.0015045925,"about_ca_system_score_gemma":0.0011524975,"threshold_uncertainty_score":0.022101462},"labels":[],"label_agreement":null},{"id":"W4410944754","doi":"10.21203/rs.3.rs-6795375/v1","title":"Introduction of a Method to Choose the Number of Clusters of a Mixed Dataset under Kproto","year":2025,"lang":"en","type":"preprint","venue":"Research Square","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Statistics Canada","funders":"","keywords":"Computer science; Chemistry; Statistics; Mathematics","score_opus":0.0772401653793962,"score_gpt":0.48828246581726664,"score_spread":0.41104230043787043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410944754","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013372195,0.00013892238,0.9965797,0.00021621624,0.00018938648,0.00010240924,0.0001397233,0.000775575,0.0005208774],"genre_scores_gemma":[0.019870214,0.00013052559,0.9765322,0.00021016829,0.00030787181,0.0003654538,0.0003889078,0.00042295252,0.0017716343],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99614656,0.0011868121,0.00032150478,0.0011112483,0.0010123242,0.00022162145],"domain_scores_gemma":[0.99273926,0.0026272866,0.00047011618,0.0015408322,0.002248118,0.00037440576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005042892,0.0011173051,0.0014628912,0.0024672272,0.001325776,0.0029479982,0.0027361873,0.00225809,0.0061058523],"category_scores_gemma":[0.017742323,0.00065398443,0.0014205187,0.001959424,0.0014719841,0.0026856791,0.003988524,0.0032412088,0.003747955],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007597648,0.00034893412,0.0052915993,0.00082341,0.0002796701,0.00027606857,0.00047594955,0.027347583,0.029125916,0.1089891,0.021543633,0.8047383],"study_design_scores_gemma":[0.0001818429,0.00027086298,0.003573377,0.00017823937,0.000121223165,0.00077413203,0.00024601794,0.8153689,0.019695297,0.102585115,0.056805633,0.0001993519],"about_ca_topic_score_codex":0.0018034418,"about_ca_topic_score_gemma":0.0034267223,"teacher_disagreement_score":0.0061058523,"about_ca_system_score_codex":0.0010698707,"about_ca_system_score_gemma":0.0018330675,"threshold_uncertainty_score":0.026669681},"labels":[],"label_agreement":null},{"id":"W4411086763","doi":"10.1109/tpami.2025.3577171","title":"A Clustering Validity Index With Multi-Granularity Fusion for Multiple Fuzzy Clustering Algorithms","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Fundamental Research Funds for the Central Universities; National Natural Science Foundation of China","keywords":"Cluster analysis; Granularity; Computer science; Fuzzy clustering; Artificial intelligence; Data mining; Pattern recognition (psychology); Fuzzy logic; Index (typography); Fusion; Fuzzy set; Algorithm","score_opus":0.04519106119666905,"score_gpt":0.3254033522786221,"score_spread":0.28021229108195306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411086763","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05166133,0.0021030465,0.9402125,0.0003601096,0.00016562286,0.0002351514,0.00043767926,0.0006839306,0.004140623],"genre_scores_gemma":[0.65158635,0.0006267938,0.34516257,0.00011847705,0.00016329529,0.00038273633,0.0011296478,0.00010671475,0.0007234332],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9878864,0.0026047833,0.001248089,0.0018109427,0.005904274,0.0005455431],"domain_scores_gemma":[0.9884941,0.004467313,0.0013144915,0.0013386067,0.0040576104,0.00032784828],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009782375,0.0014080119,0.0019202242,0.010684621,0.0023029023,0.003883565,0.0017824084,0.002130196,0.0013208418],"category_scores_gemma":[0.03244055,0.00042958895,0.00224558,0.008705922,0.0017082618,0.005736066,0.0032450086,0.0015185615,0.00036352943],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00073237426,0.00025613507,0.025904696,0.0011037027,0.0008719205,0.00025058998,0.00097294385,0.32306692,0.013893157,0.05598342,0.0068082158,0.5701559],"study_design_scores_gemma":[0.00003761664,0.00020216712,0.007612134,0.00012571963,0.00016689012,0.00025181277,0.00026201172,0.95701414,0.008026299,0.0231484,0.0030303663,0.00012235844],"about_ca_topic_score_codex":0.0038374036,"about_ca_topic_score_gemma":0.0020643352,"teacher_disagreement_score":0.010684621,"about_ca_system_score_codex":0.0031446994,"about_ca_system_score_gemma":0.0021870253,"threshold_uncertainty_score":0.051734746},"labels":[],"label_agreement":null},{"id":"W4411635074","doi":"10.1016/j.patcog.2025.111953","title":"WANN-DPC: Density peaks finding clustering based on Weighted Adaptive Nearest Neighbors","year":2025,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Science Basic Research Program of Shaanxi Province; China Postdoctoral Science Foundation; Natural Science Foundation of Shaanxi Province; National Natural Science Foundation of China","keywords":"Cluster analysis; Pattern recognition (psychology); Artificial intelligence; k-nearest neighbors algorithm; Computer science; Mathematics; Algorithm","score_opus":0.03834201504974158,"score_gpt":0.2962184310452347,"score_spread":0.25787641599549316,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411635074","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007113457,0.00021643763,0.98827386,0.00006738175,0.000100918325,0.00013188303,0.00026296222,0.0027559556,0.0010771794],"genre_scores_gemma":[0.07268691,0.00020149356,0.9216818,0.000083719504,0.000050004313,0.00027133682,0.0012320603,0.0003643608,0.0034283055],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860233,0.0001593517,0.0000790713,0.00031271664,0.0007319825,0.00011460194],"domain_scores_gemma":[0.998963,0.0001481467,0.000059351973,0.00021824098,0.0005501967,0.00006115185],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008968064,0.0009340263,0.0015561841,0.002626535,0.0014473447,0.0012346779,0.0036228695,0.0012725239,0.004090604],"category_scores_gemma":[0.0029470916,0.00061568513,0.0010853296,0.003946702,0.00057246117,0.0018133947,0.0019552938,0.0013416085,0.001991627],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034918584,0.00021333012,0.0019050905,0.00023362436,0.00018353376,0.00014347023,0.00020472954,0.06995473,0.017154718,0.009747415,0.013863386,0.88604677],"study_design_scores_gemma":[0.000036926915,0.000042272077,0.00086867134,0.000012088605,0.000036077003,0.00018742896,0.00007382787,0.9751287,0.0117790215,0.0052767787,0.0065007578,0.00005739695],"about_ca_topic_score_codex":0.018534603,"about_ca_topic_score_gemma":0.019773478,"teacher_disagreement_score":0.018534603,"about_ca_system_score_codex":0.0011352088,"about_ca_system_score_gemma":0.0020007715,"threshold_uncertainty_score":0.036853433},"labels":[],"label_agreement":null},{"id":"W4411655587","doi":"10.18280/ria.390201","title":"Exploring Swarm Intelligence for Enhanced Clustering in Artificial Intelligence: A PSO-Kmeans Hybrid Approach","year":2025,"lang":"en","type":"article","venue":"Revue d intelligence artificielle","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"k-means clustering; Swarm intelligence; Cluster analysis; Particle swarm optimization; Artificial intelligence; Computer science; Pattern recognition (psychology); Machine learning","score_opus":0.16023196000590909,"score_gpt":0.34934451377614545,"score_spread":0.18911255377023636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411655587","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011739837,0.0009679202,0.9821644,0.00037404848,0.000063169675,0.000057840796,0.000010889719,0.00016624005,0.004455574],"genre_scores_gemma":[0.4447763,0.0015663853,0.5488422,0.00023857421,0.00012291445,0.00021034553,0.00007370919,0.00008880721,0.0040807296],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995701,0.00015301887,0.000025990603,0.00006215308,0.00016171124,0.000027034612],"domain_scores_gemma":[0.99970824,0.00013738673,0.000032541804,0.000023231598,0.0000806405,0.000017901313],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000948268,0.0006842535,0.00078487874,0.001090962,0.00047949055,0.0010076967,0.0010259863,0.000993493,0.00081123406],"category_scores_gemma":[0.0013699908,0.00036464274,0.00074540684,0.0010127418,0.0006477766,0.0011707894,0.0008859421,0.0006270957,0.00022322794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000058933627,0.000102663624,0.0009960636,0.0002442668,0.00016960417,0.00010193437,0.00023345719,0.8361362,0.004194538,0.03887786,0.0017711901,0.117113255],"study_design_scores_gemma":[0.000009860045,0.000029698102,0.00014578088,0.000014886022,0.000013187379,0.000025389345,0.000030120416,0.9915139,0.00063104555,0.005989732,0.0015879871,0.000008406713],"about_ca_topic_score_codex":0.002617808,"about_ca_topic_score_gemma":0.0027323307,"teacher_disagreement_score":0.002617808,"about_ca_system_score_codex":0.0006322402,"about_ca_system_score_gemma":0.00066948443,"threshold_uncertainty_score":0.005205214},"labels":[],"label_agreement":null},{"id":"W4412421561","doi":"10.1016/j.patcog.2025.112128","title":"Gauging- <mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" altimg=\"si3.svg\"> <mml:mi>β</mml:mi> </mml:math> : Border-aware hierarchical clustering based on density and proximity","year":2025,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Computer science; Scalable Vector Graphics; Hierarchical clustering; Mathematics; Artificial intelligence; World Wide Web","score_opus":0.024043183018317774,"score_gpt":0.27805980188601126,"score_spread":0.2540166188676935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412421561","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019362612,0.00012749192,0.97175,0.00028765292,0.000053010524,0.000103813414,0.0004099376,0.0034555981,0.0044498583],"genre_scores_gemma":[0.34758413,0.0002242786,0.64540774,0.00012325087,0.000028289609,0.00016692071,0.0013621082,0.0012093883,0.003893973],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99813265,0.00044729095,0.00011892224,0.00048156048,0.000705195,0.00011448219],"domain_scores_gemma":[0.9961999,0.0012239959,0.00043470573,0.0010481509,0.00097403314,0.000119297394],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002225748,0.0008892099,0.00079001277,0.0018714078,0.0011554835,0.0029545845,0.0021445183,0.0012092642,0.0060029468],"category_scores_gemma":[0.013436458,0.0005855056,0.0007139614,0.0020406942,0.00092911714,0.0034709033,0.0029514087,0.0014683197,0.0049482626],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040045453,0.00028246295,0.0123593025,0.0005035987,0.00013787107,0.00026897955,0.001277807,0.23121522,0.06400798,0.087550044,0.02060074,0.58139557],"study_design_scores_gemma":[0.000032201846,0.00008915422,0.00393712,0.000068190166,0.00003183582,0.00018087191,0.00040379574,0.8471874,0.08601755,0.04409894,0.017862147,0.00009081402],"about_ca_topic_score_codex":0.0049719783,"about_ca_topic_score_gemma":0.006101759,"teacher_disagreement_score":0.0060029468,"about_ca_system_score_codex":0.0012755181,"about_ca_system_score_gemma":0.0014113736,"threshold_uncertainty_score":0.020081818},"labels":[],"label_agreement":null},{"id":"W4413324958","doi":"10.1016/j.neucom.2025.131324","title":"S2F: Border peeling clustering from cluster skeletons to final clusters","year":2025,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Yangzhou University; Ministry of Education of the People's Republic of China; Ministry of Education; Natural Science Foundation of Yangzhou City","keywords":"Cluster (spacecraft); Cluster analysis; Computer science; Artificial intelligence; Computer network","score_opus":0.022156098549891222,"score_gpt":0.3391739224309043,"score_spread":0.31701782388101307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413324958","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010551357,0.00015954553,0.958199,0.00022562353,0.0002041025,0.00029776682,0.0015622874,0.024704875,0.004095386],"genre_scores_gemma":[0.049856216,0.000078526886,0.93392116,0.00015361731,0.00005423874,0.00023259473,0.0042381235,0.0036744964,0.0077910596],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989956,0.000111574875,0.000047872567,0.00028998835,0.00039688984,0.0001580837],"domain_scores_gemma":[0.99846846,0.00026409648,0.00006889894,0.00048510474,0.00060617627,0.000107221604],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015594288,0.0023082693,0.0014933058,0.003006003,0.002055348,0.0027491343,0.0035077147,0.0027895642,0.028990226],"category_scores_gemma":[0.0053085005,0.0010233124,0.00202993,0.0023495103,0.000966775,0.0019040982,0.0037198986,0.0022096594,0.011054516],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008699754,0.00020791734,0.0015545617,0.0003785378,0.00020360749,0.00021357364,0.00048427592,0.06766176,0.018518258,0.010678794,0.08909682,0.8101319],"study_design_scores_gemma":[0.00022424274,0.00017246002,0.0010959812,0.00007848401,0.00005751204,0.00025751765,0.00032385928,0.90115404,0.024744822,0.033133253,0.03867938,0.000078509096],"about_ca_topic_score_codex":0.014901436,"about_ca_topic_score_gemma":0.02754239,"teacher_disagreement_score":0.028990226,"about_ca_system_score_codex":0.0011771349,"about_ca_system_score_gemma":0.0022999034,"threshold_uncertainty_score":0.096982},"labels":[],"label_agreement":null},{"id":"W4413359139","doi":"10.1109/access.2025.3600986","title":"Deep Clustering With Self-Supervision Using Pairwise Similarities","year":2025,"lang":"en","type":"article","venue":"IEEE Access","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Cluster analysis; Pairwise comparison; Artificial intelligence","score_opus":0.03610949039550173,"score_gpt":0.3456854684973579,"score_spread":0.30957597810185616,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413359139","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01855212,0.0001458541,0.97914356,0.0001252936,0.000016952214,0.000043574582,0.00009984371,0.0010135211,0.0008593693],"genre_scores_gemma":[0.5147702,0.00022709937,0.47909978,0.00021971726,0.000056412453,0.00017792296,0.0012087827,0.0003073295,0.003932782],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983181,0.00035343622,0.00009601004,0.00064197334,0.0004594027,0.00013118687],"domain_scores_gemma":[0.99787235,0.00045019915,0.00032487052,0.0006486831,0.00057290436,0.00013096334],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014507083,0.0013423382,0.0015620601,0.0015671566,0.00083771517,0.0014161533,0.002828273,0.0016257315,0.0018725038],"category_scores_gemma":[0.0041369256,0.00084158994,0.0013982554,0.001507857,0.0014970852,0.0032711052,0.0029852365,0.0019685416,0.0010658412],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019657836,0.00017364715,0.0031192033,0.00018257392,0.00019496301,0.00008397171,0.0002829665,0.6361379,0.010415346,0.02204645,0.0046809805,0.32248542],"study_design_scores_gemma":[0.000006553859,0.000027984324,0.0002619598,0.0000073911883,0.000006999345,0.00002521032,0.000019322419,0.98796684,0.0022529794,0.008888889,0.00052478147,0.0000110953815],"about_ca_topic_score_codex":0.0066369553,"about_ca_topic_score_gemma":0.010480913,"teacher_disagreement_score":0.0066369553,"about_ca_system_score_codex":0.0018100291,"about_ca_system_score_gemma":0.0019567097,"threshold_uncertainty_score":0.013196647},"labels":[],"label_agreement":null},{"id":"W4414183863","doi":"10.3390/app151810052","title":"AutoProPos: An Extension of Prototype Scattering and Positive Sampling Clustering for an Unknown Number of Clusters","year":2025,"lang":"en","type":"article","venue":"Applied Sciences","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Chicoutimi","funders":"Fonds de recherche du Québec – Nature et technologies","keywords":"Silhouette; Cluster analysis; Pattern recognition (psychology); Subspace topology; Extension (predicate logic); Autoencoder; Image (mathematics); Parametric statistics; Binary number","score_opus":0.047046133104338805,"score_gpt":0.3798647348832161,"score_spread":0.3328186017788773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414183863","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013072122,0.0002739662,0.9741969,0.00016830507,0.00007331915,0.0001244957,0.00024014279,0.009567519,0.0022832293],"genre_scores_gemma":[0.23175436,0.00030388436,0.7542729,0.00052148255,0.00013609434,0.00037840346,0.0020531053,0.0026088033,0.007971029],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985655,0.0002934594,0.00006149349,0.0005453016,0.0004148414,0.0001194332],"domain_scores_gemma":[0.99772745,0.0005970289,0.00014257668,0.00092743203,0.00048205844,0.00012339123],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019805743,0.0017944153,0.0014408188,0.0015072612,0.0012421139,0.0017151306,0.0048899376,0.0020717187,0.004157509],"category_scores_gemma":[0.006632336,0.0011883163,0.0015262818,0.0011517857,0.0016456501,0.0028217204,0.0040123453,0.0027853996,0.002700855],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005136142,0.00028577287,0.0036871056,0.0003808448,0.00028191094,0.00015939793,0.00040470785,0.30950388,0.016131438,0.027620729,0.02338222,0.6176484],"study_design_scores_gemma":[0.000025799764,0.00006176701,0.0002854593,0.000021307644,0.000014569793,0.000094887095,0.000036659174,0.97998923,0.005105332,0.010673191,0.0036707665,0.000021124213],"about_ca_topic_score_codex":0.007549721,"about_ca_topic_score_gemma":0.014874811,"teacher_disagreement_score":0.007549721,"about_ca_system_score_codex":0.0013446432,"about_ca_system_score_gemma":0.0023921782,"threshold_uncertainty_score":0.015011549},"labels":[],"label_agreement":null},{"id":"W4414633048","doi":"10.48550/arxiv.2504.15781","title":"Interacting Immediate Neighbour Interpolation for Geoscientific Data","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Resources Canada","keywords":"Interpolation (computer graphics); Curvature; Multivariate interpolation; Range (aeronautics); Trilinear interpolation; Grid; Function (biology); Nearest-neighbor interpolation","score_opus":0.15263742288790713,"score_gpt":0.40201059623415186,"score_spread":0.24937317334624473,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414633048","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011119307,0.000111658286,0.98708445,0.000046802976,0.00002175694,0.00003285397,0.0001587461,0.0006561295,0.00076833554],"genre_scores_gemma":[0.24890034,0.00019656902,0.7479414,0.000047071127,0.00002366921,0.00015005296,0.0011115151,0.00032776268,0.0013016876],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860483,0.00044198034,0.000069452784,0.00025288068,0.0005632069,0.00006778144],"domain_scores_gemma":[0.9972886,0.0012630499,0.00022940882,0.0006144461,0.0005462527,0.0000582607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002432087,0.0005813379,0.0007170051,0.0018414223,0.0006843585,0.0010250859,0.0017691719,0.00095841417,0.00174839],"category_scores_gemma":[0.010152757,0.0003999191,0.0011832183,0.0026079053,0.00074184657,0.0011013814,0.0013407794,0.0013295347,0.00086862926],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001468911,0.000077496734,0.006315359,0.0002994744,0.00012482333,0.00012067938,0.00036756517,0.755459,0.008182188,0.04131839,0.0023571826,0.18523097],"study_design_scores_gemma":[0.000005073518,0.00001766527,0.0009647588,0.000014191493,0.000005530441,0.000039870283,0.000036914484,0.9800419,0.0024969657,0.013201568,0.0031545963,0.00002094852],"about_ca_topic_score_codex":0.007588839,"about_ca_topic_score_gemma":0.010954694,"teacher_disagreement_score":0.007588839,"about_ca_system_score_codex":0.00075187936,"about_ca_system_score_gemma":0.0014080355,"threshold_uncertainty_score":0.015089333},"labels":[],"label_agreement":null},{"id":"W4415092883","doi":"10.1007/978-3-032-08649-5_4","title":"Pairwise Clustering on Numerical Datasets by Translation","year":2025,"lang":"en","type":"book-chapter","venue":"Communications in computer and information science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor","funders":"","keywords":"Pairwise comparison; Cluster analysis; Partition (number theory); Categorical variable; Parameterized complexity; Constrained clustering; Set (abstract data type); Computation; Similarity (geometry)","score_opus":0.05057951670700949,"score_gpt":0.3428955448297427,"score_spread":0.2923160281227332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415092883","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033156394,0.00078210887,0.98406684,0.00061608426,0.0005091556,0.00012613223,0.001808647,0.0027530813,0.0060223583],"genre_scores_gemma":[0.03614468,0.0010556678,0.9437294,0.00030190582,0.00043625402,0.00056906417,0.006246355,0.0013439351,0.010172738],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99749386,0.0008859435,0.00022061268,0.00066856405,0.0006292611,0.0001017539],"domain_scores_gemma":[0.9961928,0.0013178724,0.00013586569,0.0016420672,0.0006490205,0.000062369756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024909547,0.00090967765,0.0014936499,0.0023134407,0.0011950727,0.0024523525,0.0022328917,0.0009923606,0.013971526],"category_scores_gemma":[0.012036302,0.0006668793,0.0021764175,0.0077102073,0.0012910609,0.0029288367,0.0034703673,0.0019524577,0.012337791],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001911908,0.00006871771,0.0006332073,0.0009428131,0.00014219643,0.00014249074,0.00048190574,0.026076619,0.0063573713,0.11136372,0.0832832,0.7703165],"study_design_scores_gemma":[0.00008095559,0.00012605201,0.001818388,0.00023675562,0.00009663904,0.0006639019,0.0004959405,0.4279785,0.01173095,0.43409967,0.12258007,0.00009223965],"about_ca_topic_score_codex":0.0020468477,"about_ca_topic_score_gemma":0.0027685186,"teacher_disagreement_score":0.013971526,"about_ca_system_score_codex":0.0011302361,"about_ca_system_score_gemma":0.001425996,"threshold_uncertainty_score":0.04673946},"labels":[],"label_agreement":null},{"id":"W4415826823","doi":"10.1016/j.neucom.2025.131886","title":"Evidential clustering with view-weight learning for proximity data","year":2025,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Agence Nationale de la Recherche","keywords":"Cluster analysis; Interpretability; Categorical variable; Constraint (computer-aided design); Conceptual clustering; Matching (statistics); Constrained clustering; Relational database","score_opus":0.04372034419290236,"score_gpt":0.342046180839285,"score_spread":0.2983258366463826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415826823","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038921626,0.000111107714,0.9955389,0.00006779072,0.00001804737,0.000015207515,0.00003727759,0.00009173527,0.000227762],"genre_scores_gemma":[0.3425517,0.0003888818,0.6536779,0.00014360853,0.00016120338,0.000154415,0.00072641426,0.00012711028,0.0020688458],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971615,0.0010469089,0.00019002838,0.00066925236,0.00079914957,0.00013311615],"domain_scores_gemma":[0.9957202,0.0018364877,0.00035301945,0.0012233587,0.0007145546,0.00015236631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035089264,0.00080736546,0.0014772866,0.0016530511,0.00081102,0.00183092,0.0031902303,0.0019418104,0.0018655155],"category_scores_gemma":[0.014352307,0.0006705815,0.0017325338,0.0021022353,0.0014033426,0.003588617,0.003500813,0.0030799243,0.00082121097],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029825553,0.00028651705,0.0019435288,0.00034715032,0.00032861388,0.000120500845,0.0004264847,0.37657335,0.006918703,0.16142392,0.0044819643,0.44685102],"study_design_scores_gemma":[0.0000092887785,0.000043383956,0.00018892332,0.000013307921,0.000020326303,0.000036233032,0.000017056076,0.93775254,0.0008868784,0.060219593,0.0007990489,0.000013451905],"about_ca_topic_score_codex":0.0020454933,"about_ca_topic_score_gemma":0.0026651071,"teacher_disagreement_score":0.0035089264,"about_ca_system_score_codex":0.0008762865,"about_ca_system_score_gemma":0.0012391076,"threshold_uncertainty_score":0.01855719},"labels":[],"label_agreement":null},{"id":"W4416221854","doi":"10.1016/j.compind.2025.104410","title":"Integrating static and dynamic hierarchical clustering and its application to retail segmentation","year":2025,"lang":"en","type":"article","venue":"Computers in Industry","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Ministerio de Ciencia e Innovación","keywords":"Cluster analysis; Profiling (computer programming); Hierarchical clustering; Variety (cybernetics); Big data; Key (lock); Identification (biology); Market segmentation; Dynamic data","score_opus":0.016353627886842876,"score_gpt":0.32588103996101325,"score_spread":0.3095274120741704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416221854","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.049468476,0.00061009405,0.94319695,0.00039132877,0.000058187296,0.00014164098,0.00047108013,0.0012215207,0.004440688],"genre_scores_gemma":[0.5300956,0.0004848127,0.4646967,0.00013954945,0.00013248489,0.0001359792,0.0016612043,0.00023117951,0.0024224587],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979601,0.0008008375,0.000112291855,0.00048540367,0.00047369514,0.00016765554],"domain_scores_gemma":[0.99781346,0.0008176323,0.00021386285,0.0004469923,0.0006007206,0.00010734375],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023361202,0.0006713207,0.00070934044,0.0041181836,0.0010101936,0.0017813346,0.0010507656,0.00096634816,0.0010877252],"category_scores_gemma":[0.0059241685,0.0003567058,0.0010695027,0.0064202566,0.00057379896,0.0015167212,0.001752314,0.00078983395,0.00059381913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002097057,0.0002786837,0.028164517,0.00027780884,0.00040680743,0.00036480278,0.001418885,0.35585645,0.009457813,0.038352933,0.0061043617,0.5591073],"study_design_scores_gemma":[0.000007238236,0.000052508916,0.008415196,0.000031617667,0.00006283937,0.00011776328,0.0005494256,0.9603095,0.0024477576,0.020031951,0.007924185,0.0000499544],"about_ca_topic_score_codex":0.020229729,"about_ca_topic_score_gemma":0.018786673,"teacher_disagreement_score":0.020229729,"about_ca_system_score_codex":0.0012116861,"about_ca_system_score_gemma":0.001492725,"threshold_uncertainty_score":0.040223956},"labels":[],"label_agreement":null},{"id":"W4416455941","doi":"10.7717/peerj-cs.3309","title":"The Silhouette coefficient and the Davies-Bouldin index are more informative than Dunn index, Calinski-Harabasz index, Shannon entropy, and Gap statistic for unsupervised clustering internal evaluation of two convex clusters","year":2025,"lang":"en","type":"article","venue":"PeerJ Computer Science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Silhouette; Cluster analysis; Rand index; Pattern recognition (psychology); Ground truth; Consensus clustering; Euclidean distance; Correlation clustering","score_opus":0.0262903710357365,"score_gpt":0.34893061353911187,"score_spread":0.32264024250337536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416455941","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45530197,0.0077891457,0.5112124,0.0014012073,0.0005390292,0.00040690036,0.0020960213,0.0016983318,0.01955501],"genre_scores_gemma":[0.8990781,0.0009042195,0.09663117,0.00013189665,0.00011957555,0.00015679224,0.0015619512,0.00027623016,0.001140012],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99169207,0.0023092385,0.00083380664,0.001382007,0.0034096304,0.00037326396],"domain_scores_gemma":[0.9526408,0.02879458,0.005990224,0.0033594486,0.0080739455,0.0011410128],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0076809786,0.0012201901,0.0015754144,0.008020844,0.001386555,0.0044364077,0.0010079142,0.0013450589,0.0017265165],"category_scores_gemma":[0.054434024,0.0003239923,0.0008849786,0.0070056263,0.0027146821,0.0052574473,0.0018035965,0.0014225967,0.00063099625],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015481671,0.0002883534,0.24188119,0.0020719946,0.001237614,0.00055764645,0.0028864546,0.14276756,0.014405892,0.08735598,0.020781117,0.484218],"study_design_scores_gemma":[0.00011559243,0.0008270955,0.1760358,0.0005710403,0.00044840662,0.0016314473,0.003499918,0.6446015,0.035894938,0.112657554,0.022952937,0.0007637604],"about_ca_topic_score_codex":0.0043619825,"about_ca_topic_score_gemma":0.0045047496,"teacher_disagreement_score":0.008020844,"about_ca_system_score_codex":0.002239652,"about_ca_system_score_gemma":0.002162755,"threshold_uncertainty_score":0.0406214},"labels":[],"label_agreement":null},{"id":"W4416959460","doi":"10.1016/j.patcog.2025.112829","title":"Fast online ℓ0 elastic net subspace clustering via a novel dictionary update strategy","year":2025,"lang":"en","type":"article","venue":"Pattern Recognition","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Cluster analysis; Subspace topology; Block (permutation group theory); Piecewise; Diagonal; Data stream clustering; Adaptability; Norm (philosophy); Convergence (economics); Elastic net regularization","score_opus":0.03269478410830945,"score_gpt":0.3017830425501344,"score_spread":0.26908825844182493,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416959460","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005319925,0.00012791085,0.99217176,0.00009461825,0.00009655493,0.000042067153,0.00006450201,0.00081467646,0.0012680559],"genre_scores_gemma":[0.18385966,0.0002627508,0.80439305,0.00022799164,0.00018811382,0.00017509022,0.0007608344,0.0002707576,0.00986181],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99928504,0.00009388559,0.00003989612,0.00014811973,0.0003602962,0.00007266584],"domain_scores_gemma":[0.99924695,0.00018408078,0.000049621893,0.00019070604,0.00025826602,0.000070383074],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045684515,0.0009068737,0.0013235663,0.0007676559,0.0007334264,0.0010261975,0.0018668016,0.0009317828,0.0057554203],"category_scores_gemma":[0.0016776673,0.0005258266,0.00049261464,0.0012982282,0.0005559964,0.0017303927,0.0022662717,0.0012485684,0.0029647595],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005389017,0.00024518237,0.0006828774,0.00013081719,0.000089562054,0.00013517807,0.00008205144,0.12156088,0.02135389,0.01647848,0.013701068,0.8250011],"study_design_scores_gemma":[0.000023107694,0.000044361674,0.00016461343,0.0000037648133,0.000011005948,0.000090742215,0.000022376338,0.9895056,0.0040221806,0.0039440426,0.0021539275,0.000014247769],"about_ca_topic_score_codex":0.004562329,"about_ca_topic_score_gemma":0.008736431,"teacher_disagreement_score":0.0057554203,"about_ca_system_score_codex":0.0003681863,"about_ca_system_score_gemma":0.0009845537,"threshold_uncertainty_score":0.01925379},"labels":[],"label_agreement":null},{"id":"W4416963394","doi":"10.1109/cisp-bmei68103.2025.11259320","title":"Clustering Through Classifying: A Novel Neural-Based Approach","year":2025,"lang":"","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Lethbridge","funders":"","keywords":"Cluster analysis; Focus (optics); Correlation clustering; CURE data clustering algorithm; Data stream clustering; Conceptual clustering; Canopy clustering algorithm; Encoding (memory); Object (grammar)","score_opus":0.08079583600671128,"score_gpt":0.3511631127687199,"score_spread":0.27036727676200867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416963394","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0047075977,0.0002999883,0.98809123,0.00044096602,0.00008309625,0.000076875585,0.00015246181,0.0013246491,0.0048231687],"genre_scores_gemma":[0.13956065,0.0005600604,0.8469554,0.0007189863,0.00019426254,0.00016683654,0.00074793806,0.0003764739,0.010719417],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99827194,0.00022404657,0.00010469622,0.00066216895,0.0005952006,0.00014184353],"domain_scores_gemma":[0.99881494,0.00026969053,0.00012640996,0.0002935089,0.00042458673,0.00007082367],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001191478,0.0010032194,0.0012012859,0.002377797,0.0011447034,0.0028877023,0.0048802155,0.0025514816,0.003360931],"category_scores_gemma":[0.0031000564,0.00073349284,0.0015666493,0.0023559649,0.0015354311,0.0029284132,0.0020994733,0.0019580063,0.0016142082],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015823843,0.00019071532,0.0024684682,0.00023264266,0.00019652228,0.00017193597,0.00044974533,0.1677712,0.01310303,0.09466866,0.012746066,0.70784277],"study_design_scores_gemma":[0.000010301102,0.000036386908,0.0004617063,0.00003895775,0.000042167987,0.00011525476,0.00006577723,0.93261427,0.0054290737,0.051271588,0.009881078,0.000033405984],"about_ca_topic_score_codex":0.011083825,"about_ca_topic_score_gemma":0.013109076,"teacher_disagreement_score":0.011083825,"about_ca_system_score_codex":0.0020861032,"about_ca_system_score_gemma":0.0017845333,"threshold_uncertainty_score":0.022038579},"labels":[],"label_agreement":null},{"id":"W49490112","doi":"","title":"On the Complexity of Minimum Sum-of-Squares Clustering","year":2007,"lang":"fr","type":"article","venue":"PolyPublie (École Polytechnique de Montréal)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Conceptual clustering; Mathematics; Confusion; Explained sum of squares; Computer science; Artificial intelligence; Combinatorics; Correlation clustering; Statistics; Psychology; CURE data clustering algorithm","score_opus":0.05104021855297638,"score_gpt":0.3008957281111915,"score_spread":0.2498555095582151,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W49490112","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.35149702,0.0068365424,0.57424134,0.013414687,0.000820605,0.00047577702,0.0051776064,0.0035136156,0.04402278],"genre_scores_gemma":[0.65596586,0.0018450585,0.3189556,0.0008358428,0.00057792326,0.0004220426,0.0053247716,0.0011532594,0.014919635],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9952744,0.0018353555,0.00029034418,0.00074580143,0.0013323723,0.00052162865],"domain_scores_gemma":[0.9448694,0.0472533,0.0013366595,0.0030630452,0.002506165,0.00097139407],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035784948,0.0012973672,0.0027288417,0.0016629321,0.0016292002,0.0036782695,0.0039536115,0.00277154,0.010482372],"category_scores_gemma":[0.043800447,0.0009831763,0.0014496088,0.0041068178,0.0017789274,0.0066887154,0.0038592054,0.0025610707,0.0021285485],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021723623,0.0003885634,0.0060403226,0.0010590869,0.00032543868,0.000327184,0.0006993394,0.68501204,0.0045430874,0.06815041,0.03965567,0.19162656],"study_design_scores_gemma":[0.00012481849,0.000077287135,0.0013379005,0.000037105132,0.00004804702,0.0001343169,0.00019062607,0.9081623,0.0009091574,0.08664514,0.002302377,0.000030860938],"about_ca_topic_score_codex":0.013019841,"about_ca_topic_score_gemma":0.016953198,"teacher_disagreement_score":0.013019841,"about_ca_system_score_codex":0.0038563432,"about_ca_system_score_gemma":0.003603562,"threshold_uncertainty_score":0.03506702},"labels":[],"label_agreement":null},{"id":"W565269975","doi":"10.1609/aaai.v29i1.9594","title":"SoF: Soft-Cluster Matrix Factorization for Probabilistic Clustering","year":2015,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Baidu","keywords":"Cluster analysis; Non-negative matrix factorization; Probabilistic logic; Algorithm; Computer science; Mathematics; Fuzzy clustering; Correlation clustering; Data mining; Pattern recognition (psychology); Matrix decomposition; Artificial intelligence","score_opus":0.1442253030763614,"score_gpt":0.3621481781069662,"score_spread":0.2179228750306048,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W565269975","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00070852187,0.0000683214,0.9983575,0.00005921347,0.000023617957,0.000033522425,0.00008991529,0.0004125085,0.00024681786],"genre_scores_gemma":[0.053879235,0.00024055163,0.9424523,0.00021856483,0.00010843192,0.00036073948,0.0011840648,0.00029825137,0.0012579121],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967765,0.001170573,0.00015733471,0.0006763947,0.0010395736,0.00017960173],"domain_scores_gemma":[0.99706453,0.0011485048,0.00033426582,0.0005476585,0.0007598475,0.00014518328],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003477444,0.00209221,0.0018207437,0.002562144,0.0013014178,0.0016745677,0.002951646,0.0017797489,0.004284437],"category_scores_gemma":[0.008973165,0.0009629889,0.0022226814,0.0027505504,0.0015724525,0.002619883,0.0027841905,0.0025135847,0.0026651733],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020108659,0.00010584368,0.0012150608,0.00044078915,0.00030579083,0.00018325483,0.00032269483,0.54149354,0.006256235,0.083708934,0.02038877,0.345378],"study_design_scores_gemma":[0.000014149282,0.00002264346,0.00014424272,0.000017660359,0.000009774151,0.000047608817,0.00002343816,0.9545846,0.0009415334,0.041058782,0.0031145008,0.000020974518],"about_ca_topic_score_codex":0.0056738616,"about_ca_topic_score_gemma":0.0075143036,"teacher_disagreement_score":0.0056738616,"about_ca_system_score_codex":0.0013910996,"about_ca_system_score_gemma":0.0024439297,"threshold_uncertainty_score":0.018390656},"labels":[],"label_agreement":null},{"id":"W632121358","doi":"10.1007/s10115-015-0851-6","title":"On strategies for building effective ensembles of relative clustering validity criteria","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Measure (data warehouse); Data mining; Set (abstract data type); Partition (number theory); Complementarity (molecular biology); Consensus clustering; Stability (learning theory); Variety (cybernetics); Machine learning; Artificial intelligence; Mathematics; Fuzzy clustering; CURE data clustering algorithm","score_opus":0.06181771647573652,"score_gpt":0.357017140401257,"score_spread":0.29519942392552045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W632121358","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040025027,0.0001425275,0.99473107,0.00010976373,0.000016482809,0.00007049635,0.000033775785,0.00013815844,0.00075523875],"genre_scores_gemma":[0.112838894,0.00020469495,0.88465095,0.00017968082,0.000084528845,0.0004177625,0.00034897157,0.00020809556,0.0010665203],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9912582,0.0046175965,0.00067880226,0.00094923336,0.0020858424,0.00041027417],"domain_scores_gemma":[0.96212643,0.027617354,0.0011439591,0.0030884512,0.0054950663,0.0005287752],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013291887,0.0015763268,0.0024271323,0.0039020905,0.0017494073,0.0035848098,0.004255763,0.0028441173,0.0039589503],"category_scores_gemma":[0.06355019,0.0011780496,0.0017051072,0.0035774035,0.002044486,0.0063332026,0.00576961,0.0034598657,0.0010578184],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020762696,0.0003177333,0.0027277207,0.00033124583,0.00032166837,0.000119526136,0.0007637503,0.36985558,0.0037626699,0.18691298,0.0057558524,0.42892373],"study_design_scores_gemma":[0.000025157191,0.00007503813,0.0003319236,0.000070786526,0.000049815324,0.00005339753,0.00012991777,0.8874778,0.0018084095,0.10841921,0.0015296381,0.000028869556],"about_ca_topic_score_codex":0.00435276,"about_ca_topic_score_gemma":0.006689926,"teacher_disagreement_score":0.013291887,"about_ca_system_score_codex":0.0017261411,"about_ca_system_score_gemma":0.0026406266,"threshold_uncertainty_score":0.070295036},"labels":[],"label_agreement":null},{"id":"W65834064","doi":"","title":"Clustering in the Presence of Background Noise","year":2014,"lang":"en","type":"article","venue":"International Conference on Machine Learning","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Robustness (evolution); Computer science; Centroid; CURE data clustering algorithm; Data mining; Cluster (spacecraft); Correlation clustering; Noise (video); Algorithm; Constrained clustering; Canopy clustering algorithm; Artificial intelligence","score_opus":0.05514190194778961,"score_gpt":0.34961984386203676,"score_spread":0.29447794191424714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W65834064","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018792897,0.00027436414,0.9791661,0.00029377258,0.000050688457,0.00002780159,0.00007411163,0.00026058874,0.0010596545],"genre_scores_gemma":[0.5211476,0.0008205532,0.47075495,0.00043264992,0.0003009391,0.00014406645,0.00059877656,0.00035284556,0.0054475446],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99610704,0.0011081606,0.00014360445,0.0010804221,0.001238547,0.0003223372],"domain_scores_gemma":[0.9908144,0.0048779747,0.000959158,0.0016034192,0.0014527536,0.0002922715],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041297437,0.0010315375,0.0016371578,0.0020031768,0.0012721749,0.0024050327,0.002741913,0.0032901887,0.00085459923],"category_scores_gemma":[0.018792924,0.0006234635,0.0010511286,0.0022393398,0.0025426452,0.0028188091,0.0025577536,0.0021147584,0.0008906404],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032242175,0.00006073327,0.0017260485,0.00027430672,0.00015815994,0.00044042972,0.00040156007,0.8401367,0.022064447,0.08893355,0.0022311604,0.043250516],"study_design_scores_gemma":[0.000010099025,0.00003542747,0.00048097572,0.000013439829,0.000017777124,0.000083516716,0.000039623985,0.9554898,0.006364023,0.03571974,0.0017247326,0.000020870399],"about_ca_topic_score_codex":0.002632922,"about_ca_topic_score_gemma":0.001687512,"teacher_disagreement_score":0.0041297437,"about_ca_system_score_codex":0.0014481249,"about_ca_system_score_gemma":0.000981695,"threshold_uncertainty_score":0.021840394},"labels":[],"label_agreement":null},{"id":"W6920510158","doi":"10.60692/pgp62-5kc97","title":"A unified view of density-based methods for semi-supervised clustering and classification","year":2019,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Core (optical fiber); Bridge (graph theory); Fraction (chemistry); Labeled data; Conceptual clustering; Correlation clustering; Data stream clustering; Pattern recognition (psychology)","score_opus":0.07282835786029025,"score_gpt":0.31763390026555316,"score_spread":0.24480554240526292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920510158","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007251303,0.00017212977,0.9985251,0.00013884302,0.000017691598,0.000021706897,0.000016071615,0.0000646653,0.0003186637],"genre_scores_gemma":[0.14271286,0.0009736633,0.85340166,0.00027982917,0.00039334604,0.00041201874,0.00026304147,0.00014625698,0.001417319],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934063,0.0033333602,0.00032428777,0.00088210683,0.0018821465,0.00017185614],"domain_scores_gemma":[0.98863333,0.005990139,0.0006875924,0.0015649954,0.0028107339,0.0003132499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007969012,0.0011950054,0.0019257491,0.0039969054,0.0013400186,0.0036147458,0.003747024,0.0022078855,0.0014818734],"category_scores_gemma":[0.017721053,0.00096410053,0.0016745533,0.0030493508,0.0037054725,0.0047421106,0.0039198417,0.0035346549,0.0009008702],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008634466,0.00011964744,0.001549593,0.00040005596,0.00021339962,0.00014012102,0.0006047669,0.32892314,0.0036912484,0.48076022,0.004540144,0.17897132],"study_design_scores_gemma":[0.0000068233535,0.000029782526,0.00022486981,0.00003385608,0.000017288596,0.00005688246,0.000047262347,0.89539486,0.0009722727,0.10071784,0.0024734796,0.00002483594],"about_ca_topic_score_codex":0.0032888143,"about_ca_topic_score_gemma":0.0028232303,"teacher_disagreement_score":0.007969012,"about_ca_system_score_codex":0.0021562325,"about_ca_system_score_gemma":0.0017992585,"threshold_uncertainty_score":0.042144656},"labels":[],"label_agreement":null},{"id":"W6920973824","doi":"10.6084/m9.figshare.25807416","title":"Additional file 1 of An inversion-based clustering approach for complex clusters","year":2024,"lang":"en","type":"article","venue":"Open MIND","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Table (database); Data file; File format; Code (set theory); Cluster (spacecraft)","score_opus":0.0896934857428779,"score_gpt":0.3483930517080126,"score_spread":0.2586995659651347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920973824","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00048118265,0.0000671872,0.0064912383,0.0001318911,0.00010427177,0.0001757214,0.9859703,0.004426034,0.002152172],"genre_scores_gemma":[0.0108926585,0.00022736005,0.040655382,0.00060315325,0.00015362223,0.002679728,0.9233555,0.010146535,0.01128602],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989386,0.00018673913,0.00012173548,0.00034940656,0.00028336595,0.00012009001],"domain_scores_gemma":[0.9841086,0.011047343,0.00047890152,0.0016729117,0.0023009821,0.0003911951],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0020507425,0.0018422866,0.001526715,0.0022444595,0.0013550036,0.0029170618,0.0026847136,0.0013625473,0.805477],"category_scores_gemma":[0.027343757,0.0008699968,0.0011508919,0.0035966686,0.0004232595,0.0022047197,0.0015573804,0.001587646,0.25024837],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040825095,0.00010244588,0.001280529,0.0023354096,0.00007813377,0.00007626175,0.00007586569,0.0013189957,0.0007649725,0.0016472673,0.9727461,0.019165855],"study_design_scores_gemma":[0.0029176355,0.0003602644,0.01149347,0.0016815944,0.00029943616,0.00080320874,0.0004173256,0.014113413,0.0063434145,0.03925518,0.922029,0.00028606917],"about_ca_topic_score_codex":0.0045294743,"about_ca_topic_score_gemma":0.00927606,"teacher_disagreement_score":0.805477,"about_ca_system_score_codex":0.0011510113,"about_ca_system_score_gemma":0.0021206734,"threshold_uncertainty_score":0.27746332},"labels":[],"label_agreement":null},{"id":"W6924105539","doi":"10.15468/dl.pr7yrs","title":"Occurrence Download","year":2023,"lang":"en","type":"dataset","venue":"Global Biodiversity Information Facility","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Download; Matching (statistics); Range (aeronautics); Set (abstract data type); Confidentiality","score_opus":0.025895767468059317,"score_gpt":0.2681033199922915,"score_spread":0.24220755252423215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6924105539","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00008903441,0.000039025483,0.000070915405,0.000049536968,0.000016390391,0.000008595937,0.9983125,0.0006063715,0.0008076368],"genre_scores_gemma":[0.00019529786,0.000030328873,0.00027444848,0.00004308711,0.000003268017,0.000044589742,0.99890256,0.00011843269,0.0003879377],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99900466,0.00013020879,0.00013034645,0.00033677934,0.00023572257,0.00016226116],"domain_scores_gemma":[0.99800175,0.00049068435,0.00019059323,0.000572797,0.0005047712,0.00023936246],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009236685,0.0020984185,0.0015156557,0.0046087014,0.0010775559,0.0022705994,0.003049358,0.0022580577,0.09513448],"category_scores_gemma":[0.0049766796,0.0008463798,0.0011996969,0.008541177,0.0004644961,0.0021155714,0.0024347028,0.0021395558,0.1618119],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031927622,0.000016739037,0.000445739,0.00048184273,0.000014500527,0.000019024463,0.000025972995,0.00015910494,0.00015570056,0.00041184193,0.9966007,0.0016368838],"study_design_scores_gemma":[0.00008329494,0.000010171156,0.0021466408,0.00017447391,0.000013728867,0.000056476063,0.00009571566,0.00026250776,0.00027139165,0.0008325511,0.99603385,0.00001915239],"about_ca_topic_score_codex":0.019488623,"about_ca_topic_score_gemma":0.037315313,"teacher_disagreement_score":0.9048655,"about_ca_system_score_codex":0.0017031854,"about_ca_system_score_gemma":0.0022291457,"threshold_uncertainty_score":0.31825656},"labels":[],"label_agreement":null},{"id":"W6925415964","doi":"10.18712/nsd-nsd2584-2-v3","title":"Travel and Holiday Survey 2017, 2nd quarter, Person File + Wood Firing + Health & Vaccination","year":2022,"lang":"en","type":"dataset","venue":"NSD – Norsk senter for forskningsdata","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quarter (Canadian coin); The Internet; Norwegian; Vaccination; Work (physics); National Health Interview Survey","score_opus":0.06840023896232364,"score_gpt":0.3445974789706527,"score_spread":0.276197240008329,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925415964","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00035381355,0.000027398197,0.000031438678,0.00004848119,0.00002673701,0.0000136579865,0.999086,0.00005686461,0.00035562794],"genre_scores_gemma":[0.0006230833,0.000032087635,0.00014928692,0.000027336842,0.0000093291865,0.000082349616,0.9983158,0.000021354008,0.00073943066],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99895847,0.00019713615,0.00017350314,0.0002510598,0.00023966984,0.00018006268],"domain_scores_gemma":[0.998054,0.0003449491,0.000271442,0.00032623002,0.0007871149,0.0002162903],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013838019,0.0015989747,0.0011533787,0.0021749465,0.00067621947,0.0016386377,0.0018643565,0.0011852534,0.038460407],"category_scores_gemma":[0.005324232,0.0006733422,0.00092219247,0.004547564,0.00031581213,0.00092725287,0.0012472633,0.0014167839,0.047400653],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007661282,0.00003594824,0.0022309928,0.0003747915,0.00002364893,0.000020451653,0.000029041226,0.00016617603,0.00005803248,0.00020970812,0.9953934,0.0013811312],"study_design_scores_gemma":[0.00068332313,0.00006811019,0.03602494,0.0005330035,0.000055065073,0.00011188562,0.0005965409,0.0007704052,0.0003972514,0.0005912594,0.9601186,0.000049618084],"about_ca_topic_score_codex":0.06178732,"about_ca_topic_score_gemma":0.08735298,"teacher_disagreement_score":0.06178732,"about_ca_system_score_codex":0.001586203,"about_ca_system_score_gemma":0.0023369708,"threshold_uncertainty_score":0.12866288},"labels":[],"label_agreement":null},{"id":"W6929161022","doi":"10.4224/40000433","title":"Guide des règles de l'art sur les coupe-feu et les pare-feu et leur effet sur la transmission acoustique","year":2007,"lang":"en","type":"report","venue":"NPARC","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Terminology; Context (archaeology); Jurisdiction; Best practice; Fire safety; Fire protection","score_opus":0.05981230376779776,"score_gpt":0.3696977348816851,"score_spread":0.30988543111388733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6929161022","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002280313,0.03805627,0.21664701,0.020355372,0.006892151,0.0010174337,0.004888113,0.008360699,0.7015027],"genre_scores_gemma":[0.012174222,0.029771572,0.15189533,0.006058804,0.00048197203,0.0007762135,0.0036116063,0.003490818,0.7917394],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9934783,0.0007224485,0.0003142904,0.00039048624,0.0047645927,0.00032982175],"domain_scores_gemma":[0.9908029,0.0014821101,0.0002048225,0.0007648357,0.0065078274,0.0002375926],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050662966,0.001448705,0.0008617571,0.0029145526,0.003160422,0.0058472184,0.0036835175,0.0053087343,0.08766848],"category_scores_gemma":[0.009616398,0.0014492266,0.0009976237,0.0023164516,0.0031897998,0.005108642,0.002378013,0.004703921,0.07351955],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033727927,0.000047483416,0.00026621448,0.00074172154,0.0000074740988,0.00020405455,0.001224363,0.0019708716,0.0027465338,0.07066334,0.64757353,0.2745207],"study_design_scores_gemma":[0.0000020567234,0.0000072378366,0.000092048176,0.00019263689,0.000001992671,0.00009718609,0.00010563878,0.00013410159,0.00030470206,0.0012730451,0.99777776,0.000011616205],"about_ca_topic_score_codex":0.15899383,"about_ca_topic_score_gemma":0.19546472,"teacher_disagreement_score":0.15899383,"about_ca_system_score_codex":0.0063103894,"about_ca_system_score_gemma":0.018295126,"threshold_uncertainty_score":0.31613678},"labels":[],"label_agreement":null},{"id":"W6929807060","doi":"10.5061/dryad.bk3j9kd6f","title":"Data from: Integrating over uncertainty in spatial scale of response within multispecies occupancy models yields more accurate assessments of community composition","year":2019,"lang":"en","type":"dataset","venue":"Open MIND","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Scale (ratio); Bayesian probability; Occupancy; Range (aeronautics); Spatial ecology; Selection (genetic algorithm)","score_opus":0.22259440158859317,"score_gpt":0.46460979305338657,"score_spread":0.2420153914647934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6929807060","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08659858,0.0004831364,0.8800264,0.0025129302,0.00024139004,0.00024631285,0.01580664,0.0035466498,0.010537906],"genre_scores_gemma":[0.7123011,0.00040841458,0.26091236,0.0010980783,0.00013032742,0.0006748508,0.018293925,0.0009392644,0.0052416897],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99760497,0.00096416817,0.00016881115,0.00060342293,0.00052607915,0.00013260062],"domain_scores_gemma":[0.98962224,0.005202399,0.0010386645,0.0029077388,0.00096713775,0.00026188645],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005861261,0.0007529021,0.0009405037,0.0007151023,0.00056427385,0.0020735215,0.0025997309,0.0014648916,0.009833737],"category_scores_gemma":[0.028161129,0.00072649965,0.0019490655,0.0009845861,0.0010032557,0.0037019956,0.0025346954,0.0024559118,0.0023147208],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009087187,0.0003108652,0.13568816,0.0011729144,0.00067508116,0.0003863489,0.0010210166,0.47509614,0.007900101,0.18821074,0.025599374,0.1630305],"study_design_scores_gemma":[0.00011639013,0.00016666457,0.020992696,0.00019529599,0.000095201176,0.00024697924,0.00020905775,0.78134245,0.0055168048,0.15510936,0.035871893,0.00013722967],"about_ca_topic_score_codex":0.011363862,"about_ca_topic_score_gemma":0.016555365,"teacher_disagreement_score":0.011363862,"about_ca_system_score_codex":0.0012279664,"about_ca_system_score_gemma":0.0014655439,"threshold_uncertainty_score":0.032897055},"labels":[],"label_agreement":null},{"id":"W6929870454","doi":"10.5061/dryad.rfj6q579r","title":"Source pool diversity and proximity shape the compositional uniqueness of insular mammal assemblages worldwide","year":2021,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Concordia University; Université de Sherbrooke","funders":"","keywords":"Ectotherm; Population; Variation (astronomy); Range (aeronautics); Natural (archaeology); Lacertidae","score_opus":0.031747393981363045,"score_gpt":0.2536503977295346,"score_spread":0.22190300374817157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6929870454","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9994655,0.00006133146,0.00018489228,0.000008730017,3.4864723e-7,8.677221e-7,0.00004464645,0.000004034423,0.00022959319],"genre_scores_gemma":[0.9996793,0.00002553202,0.00015531959,0.000005643451,0.0000010508687,0.0000013625826,0.00007904501,0.0000028517886,0.00004986091],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99972934,0.00009074649,0.000018219325,0.00010716857,0.000024078287,0.000030430494],"domain_scores_gemma":[0.9983341,0.00069738063,0.0005089684,0.00020330226,0.00011596337,0.00014032453],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00049529556,0.0002271839,0.0003034715,0.0015103013,0.00037979908,0.0009235286,0.00032803966,0.00028789358,0.0022815275],"category_scores_gemma":[0.0024787395,0.00025018438,0.00040365008,0.0007803415,0.0006779288,0.0006141037,0.0011028853,0.0002765874,0.00023440564],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000075333955,0.000012733815,0.9876556,0.000025644265,0.0002769441,0.000073710704,0.0005290515,0.00068266934,0.0058073173,0.00015775135,0.00005139765,0.0046518985],"study_design_scores_gemma":[0.000001177911,0.0000068778827,0.99839705,0.0000037807997,0.000026902411,0.000043930402,0.00022424909,0.0010237911,0.00010574881,0.00012095738,0.000043077383,0.000002534903],"about_ca_topic_score_codex":0.0043773465,"about_ca_topic_score_gemma":0.013532779,"teacher_disagreement_score":0.0043773465,"about_ca_system_score_codex":0.00017209306,"about_ca_system_score_gemma":0.00011493776,"threshold_uncertainty_score":0.008703709},"labels":[],"label_agreement":null},{"id":"W6930383710","doi":"10.5281/zenodo.12342515","title":"bankamp katalog 2017 pdf","year":2024,"lang":"en","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subpoena; Paraphernalia; Pretext; Fusible alloy; Limiting","score_opus":0.033752732861100324,"score_gpt":0.2736382651706521,"score_spread":0.23988553230955179,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6930383710","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00023755233,0.0005191213,0.00051807927,0.0011387103,0.002361797,0.00011529051,0.003446858,0.0030274275,0.98863506],"genre_scores_gemma":[0.0008882636,0.0003627959,0.00019991677,0.0004770438,0.00025731514,0.00004251338,0.0014223179,0.0011852616,0.9951645],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990903,0.00006350464,0.00004729559,0.00010308125,0.00058086345,0.00011495016],"domain_scores_gemma":[0.9968451,0.00033402877,0.00009485419,0.00034215229,0.0018442187,0.0005396741],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0007161256,0.0011798363,0.0009655943,0.0025066182,0.002637799,0.009910053,0.0021351855,0.0027959477,0.9274435],"category_scores_gemma":[0.0043670754,0.00092661957,0.0007938736,0.0023278445,0.0010157422,0.008715182,0.00353948,0.0033059446,0.91188866],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002356003,0.00001700059,0.000023567489,0.00007484492,9.638188e-7,0.000037515343,0.000018897139,0.000024690391,0.00013921357,0.0016924965,0.97653437,0.02141297],"study_design_scores_gemma":[0.00000674158,0.000008108077,0.0000842841,0.00004631894,0.0000011009904,0.000054498352,0.00004362193,0.000028957496,0.00018097866,0.00058299885,0.9989562,0.00000615556],"about_ca_topic_score_codex":0.0046526007,"about_ca_topic_score_gemma":0.00961682,"teacher_disagreement_score":0.072556496,"about_ca_system_score_codex":0.0021743309,"about_ca_system_score_gemma":0.0027607242,"threshold_uncertainty_score":0.103492975},"labels":[],"label_agreement":null},{"id":"W6931096469","doi":"10.5281/zenodo.4699907","title":"+27733404752 REVENGE SPELL, INSTANT DEATH SPELLS IN AUSTRALIA, CANADA, MALTA,AUSTRIA, BAHAMAS, BAHRAIN, BARBADOS, BELGIUM, BHUTAN, BRUNEI.","year":2021,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Spell; MAGIC (telescope); Instant; Harm; Wife","score_opus":0.04873091082236333,"score_gpt":0.2688419079113559,"score_spread":0.2201109970889926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931096469","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011537378,0.00039669196,0.00027127037,0.002376925,0.001268451,0.00014335995,0.0022532623,0.001541451,0.99059486],"genre_scores_gemma":[0.0010118696,0.0001231188,0.000086213804,0.0003339523,0.000051698673,0.000017130731,0.00046120462,0.00025360147,0.9976611],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99941885,0.0000525639,0.000021122683,0.000116608775,0.00020263287,0.00018829494],"domain_scores_gemma":[0.99710125,0.00015391163,0.000093768,0.000120204306,0.0006491426,0.0018817075],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00056907086,0.0010232189,0.0006608422,0.000819467,0.0034161392,0.00389678,0.0008657315,0.0018989053,0.94478244],"category_scores_gemma":[0.0017168524,0.0004603276,0.0005202782,0.0009713596,0.0005306537,0.0018522578,0.0032576772,0.0014328641,0.86539465],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003079922,0.000024855937,0.00024383143,0.000050260005,0.0000011510172,0.000059079284,0.000085347056,0.000020137555,0.00036243728,0.0003343058,0.96979666,0.028991194],"study_design_scores_gemma":[0.000009328505,0.00002045855,0.0011526557,0.000034548128,0.0000014880908,0.000069122514,0.000319328,0.00003055504,0.000089817295,0.000060654613,0.99820673,0.0000051910947],"about_ca_topic_score_codex":0.018162321,"about_ca_topic_score_gemma":0.05854801,"teacher_disagreement_score":0.055217564,"about_ca_system_score_codex":0.001952793,"about_ca_system_score_gemma":0.0029165498,"threshold_uncertainty_score":0.07876116},"labels":[],"label_agreement":null},{"id":"W6931585265","doi":"10.5281/zenodo.7629517","title":"Fig. 78. Mesoscutum pubescence. A in Revision of the Nearctic species of the Lasioglossum (Dialictus) gemmatum species complex (Hymenoptera: Halictidae)","year":2023,"lang":"en","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Nearctic ecozone; Dorsum; Species complex; Scale (ratio)","score_opus":0.04716349607864643,"score_gpt":0.260761916863621,"score_spread":0.21359842078497457,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931585265","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16126248,0.03438506,0.03523339,0.0034560794,0.006151904,0.0024402405,0.042942278,0.006747877,0.70738065],"genre_scores_gemma":[0.7860374,0.008541777,0.036911953,0.0007304769,0.001248811,0.00044005,0.021870391,0.0006682571,0.14355095],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.999841,0.000028684053,0.000011287075,0.00003912678,0.00005927455,0.000020658259],"domain_scores_gemma":[0.9997546,0.00003734812,0.00007116701,0.000035689656,0.00006122647,0.00003993082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019871515,0.00074751314,0.00025859007,0.002753978,0.0010454934,0.0005970798,0.0005985187,0.00050057704,0.045351382],"category_scores_gemma":[0.00054350356,0.0002939364,0.00038631115,0.0013296847,0.0010072639,0.0006813493,0.00083847065,0.000766685,0.011896583],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00074229273,0.00008039492,0.025450079,0.0019310955,0.00010389441,0.0031054094,0.0011563248,0.001053803,0.042299166,0.01198498,0.15444386,0.75764865],"study_design_scores_gemma":[0.00005575233,0.00011815789,0.15163001,0.0004002551,0.000063091924,0.007208688,0.0006192536,0.00063388894,0.0024998868,0.0013966978,0.83534384,0.000030441137],"about_ca_topic_score_codex":0.0063926894,"about_ca_topic_score_gemma":0.012980017,"teacher_disagreement_score":0.045351382,"about_ca_system_score_codex":0.00049945933,"about_ca_system_score_gemma":0.00041706266,"threshold_uncertainty_score":0.15171552},"labels":[],"label_agreement":null},{"id":"W6931742362","doi":"10.5281/zenodo.5708940","title":"Automated Risk of Bias Tool for Prevalence Studies","year":2021,"lang":"en","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Calgary; McGill University Health Centre; University of Waterloo; University of Toronto","funders":"","keywords":"Checklist; Risk assessment; MEDLINE; Data collection; Risk factor; Seroprevalence","score_opus":0.07633642572904348,"score_gpt":0.3219613033725742,"score_spread":0.24562487764353075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931742362","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013306684,0.0031289128,0.7801833,0.0031476351,0.0013360657,0.09768459,0.057687983,0.032792613,0.010732237],"genre_scores_gemma":[0.022293836,0.0005256143,0.8536594,0.0010292731,0.00025021838,0.1058169,0.010175802,0.002536737,0.0037122152],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.75159216,0.13230573,0.09211283,0.0046567204,0.017664967,0.0016675696],"domain_scores_gemma":[0.2693264,0.5756999,0.042495497,0.034779012,0.075007714,0.0026914761],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.21286242,0.0036054354,0.004497592,0.017703291,0.0019350252,0.0044719134,0.0048734676,0.003426299,0.065007806],"category_scores_gemma":[0.48470145,0.0038325656,0.00812828,0.0060601635,0.0013064842,0.003777055,0.0068409266,0.004124685,0.008256598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0073868693,0.0011733588,0.021615678,0.05982557,0.0072956868,0.0009958431,0.0044163037,0.0057741296,0.0047849524,0.020389965,0.28826404,0.57807755],"study_design_scores_gemma":[0.013820907,0.002331063,0.055383734,0.045886014,0.009937715,0.0034207855,0.0018460371,0.05282084,0.026163368,0.0657612,0.72097266,0.0016556389],"about_ca_topic_score_codex":0.0015331256,"about_ca_topic_score_gemma":0.0031308627,"teacher_disagreement_score":0.78713757,"about_ca_system_score_codex":0.0033293085,"about_ca_system_score_gemma":0.011247662,"threshold_uncertainty_score":0.9706808},"labels":[],"label_agreement":null},{"id":"W6931817005","doi":"10.5683/sp3/xy4jjv","title":"Environics Focus Ontario 1996-1","year":2009,"lang":"en","type":"dataset","venue":"Borealis","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Innovative Research Group (Canada)","funders":"","keywords":"Focus (optics); Public opinion; Field (mathematics); Focus group; Government (linguistics)","score_opus":0.019977066730199833,"score_gpt":0.27927511177958914,"score_spread":0.2592980450493893,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931817005","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00070513994,0.000060317434,0.000025903884,0.000066775945,0.00001647899,0.000015027948,0.99775535,0.00007534181,0.0012797242],"genre_scores_gemma":[0.0014026063,0.00005807925,0.00009409651,0.000022293914,0.000006214066,0.000038972194,0.99552035,0.000018062326,0.002839246],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989526,0.000058702874,0.00008214918,0.00022284637,0.00040815503,0.00027556354],"domain_scores_gemma":[0.9970305,0.0001170874,0.0002708734,0.00029779624,0.0019548507,0.00032888245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005997519,0.0019435207,0.0011581684,0.0045541395,0.0019868067,0.0020417676,0.0029206865,0.0010033126,0.024150124],"category_scores_gemma":[0.003515594,0.0007213744,0.0008270946,0.01267802,0.0006123722,0.00084165647,0.001298922,0.0010368861,0.024947166],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008487358,0.000015720154,0.004101084,0.00020535872,0.000020586609,0.000036385154,0.000052618303,0.00017522514,0.00007729633,0.00027845302,0.9930861,0.0018663688],"study_design_scores_gemma":[0.0001720937,0.000020009506,0.06651235,0.00019104178,0.000034350134,0.00006928656,0.00044593328,0.00051521277,0.00038566097,0.00027936278,0.9313352,0.000039442202],"about_ca_topic_score_codex":0.9304718,"about_ca_topic_score_gemma":0.9762782,"teacher_disagreement_score":0.06952822,"about_ca_system_score_codex":0.01594312,"about_ca_system_score_gemma":0.017108154,"threshold_uncertainty_score":0.13987535},"labels":[],"label_agreement":null},{"id":"W6958204250","doi":"10.6084/m9.figshare.12441566.v1","title":"Additional file 5 of Developing a framework to inform scale-up success for population health interventions: a critical interpretive synthesis of the literature","year":2020,"lang":"en","type":"article","venue":"Figshare","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Ottawa; University of Calgary; Government of New Brunswick; University of New Brunswick","funders":"","keywords":"Population; Population health; Visualization; Qualitative research; Public health; Health professionals; MEDLINE","score_opus":0.04939400675366261,"score_gpt":0.3808540122971685,"score_spread":0.33146000554350585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958204250","genre_codex":"dataset","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006257464,0.00013025534,0.0030434604,0.001230397,0.00014758148,0.0018615108,0.98861045,0.0005762047,0.0037743237],"genre_scores_gemma":[0.053103376,0.0014808597,0.1642548,0.003365524,0.00057236745,0.07270465,0.6702925,0.0030986986,0.031127216],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9953484,0.0020501853,0.0008596258,0.0005405422,0.00088916894,0.00031199123],"domain_scores_gemma":[0.7621903,0.21068507,0.0050873356,0.004020545,0.017071437,0.0009453329],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.016848996,0.0015800407,0.0016608095,0.009260188,0.0017889377,0.0033521482,0.0033968592,0.0014579924,0.79586256],"category_scores_gemma":[0.16681394,0.00067841966,0.002156407,0.011161026,0.0006388382,0.004115132,0.0024862553,0.0016982518,0.042844698],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048931607,0.00007730115,0.0019107815,0.027878264,0.00016305271,0.00009339286,0.0015226335,0.00097984,0.00010739636,0.00727172,0.9327414,0.026764765],"study_design_scores_gemma":[0.0040688324,0.00031014573,0.017811542,0.03825951,0.0010541873,0.00024657082,0.007776215,0.0042275283,0.00092354213,0.055401497,0.86961347,0.00030702775],"about_ca_topic_score_codex":0.013996452,"about_ca_topic_score_gemma":0.025575314,"teacher_disagreement_score":0.79586256,"about_ca_system_score_codex":0.003187096,"about_ca_system_score_gemma":0.008663904,"threshold_uncertainty_score":0.29117727},"labels":[],"label_agreement":null},{"id":"W6977015603","doi":"10.6084/m9.figshare.25807416.v1","title":"Additional file 1 of An inversion-based clustering approach for complex clusters","year":2024,"lang":"en","type":"article","venue":"Figshare","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Cluster analysis; Table (database); Data file; File format; Code (set theory); Cluster (spacecraft)","score_opus":0.08145911007452508,"score_gpt":0.3101200265644326,"score_spread":0.22866091648990752,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977015603","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000431408,0.00007105976,0.006791352,0.0001392099,0.00011023809,0.00016007564,0.9845841,0.00548154,0.0022310573],"genre_scores_gemma":[0.010231092,0.00023370262,0.03878288,0.00059256086,0.00015632415,0.0023596815,0.92456716,0.012054982,0.011021568],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988418,0.00020541548,0.00012672938,0.00038493317,0.000310996,0.00013009646],"domain_scores_gemma":[0.98303014,0.011808127,0.0004752845,0.001908314,0.0023653253,0.0004128044],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0021966195,0.0019469552,0.001622556,0.0023553898,0.0014047233,0.0031872098,0.0028614004,0.0014184417,0.8162694],"category_scores_gemma":[0.028464042,0.0009786447,0.001264378,0.0036994833,0.00044652514,0.002389776,0.0016495043,0.0016742352,0.28768244],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036976708,0.00008424242,0.001090037,0.0020878129,0.00007965802,0.00006615266,0.00006604789,0.0012935787,0.0006975832,0.001623172,0.9754566,0.017085273],"study_design_scores_gemma":[0.002816949,0.00034197595,0.010395205,0.0015306206,0.00030087482,0.00074051105,0.0003532654,0.014149793,0.006273435,0.0399644,0.9228471,0.00028579208],"about_ca_topic_score_codex":0.0040365206,"about_ca_topic_score_gemma":0.0083718505,"teacher_disagreement_score":0.8162694,"about_ca_system_score_codex":0.001192606,"about_ca_system_score_gemma":0.002159481,"threshold_uncertainty_score":0.26206934},"labels":[],"label_agreement":null},{"id":"W6978058864","doi":"10.6084/m9.figshare.24892825.v1","title":"Additional file 1 of Geospatial analysis and participant characteristics associated with colorectal cancer screening participation in Alberta, Canada: a population-based cross-sectional study","year":2023,"lang":"en","type":"article","venue":"Figshare","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Calgary; Alberta Health Services","funders":"","keywords":"Colorectal cancer screening; Geospatial analysis; Cancer screening; Colorectal cancer; Data collection","score_opus":0.049989912967286894,"score_gpt":0.32920261522279964,"score_spread":0.27921270225551276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6978058864","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006329772,0.000048135906,0.00023511877,0.00008004115,0.000025531455,0.00025108733,0.9913898,0.000054289172,0.0015863458],"genre_scores_gemma":[0.12563026,0.00044106948,0.00567796,0.00043916013,0.00006770951,0.0037004398,0.8435383,0.00017780959,0.02032725],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9993498,0.00006260023,0.000088773355,0.00015438808,0.00017194924,0.00017261367],"domain_scores_gemma":[0.99332774,0.002023122,0.00069321267,0.00050801726,0.0028782,0.0005697217],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00090914045,0.00074016233,0.0010707078,0.0030818835,0.00267368,0.0013124169,0.002110942,0.00077035883,0.37786072],"category_scores_gemma":[0.011904753,0.00061852235,0.0009012372,0.008644395,0.00041406657,0.0006969772,0.0007709295,0.00065817474,0.0125098],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057267566,0.0003266649,0.14069413,0.0019471501,0.00023174407,0.00024683334,0.0006465637,0.0010432587,0.00018142053,0.0013352449,0.8352245,0.017549798],"study_design_scores_gemma":[0.0018020185,0.00020243428,0.80750597,0.0026096753,0.0004710059,0.00078451436,0.0037289297,0.002717716,0.00035565716,0.0021983646,0.17743813,0.00018557115],"about_ca_topic_score_codex":0.9152729,"about_ca_topic_score_gemma":0.93112487,"teacher_disagreement_score":0.37786072,"about_ca_system_score_codex":0.007026206,"about_ca_system_score_gemma":0.018497951,"threshold_uncertainty_score":0.8874062},"labels":[],"label_agreement":null},{"id":"W6986648496","doi":"","title":"¿Qué forma de educación lingüística quieren los padres inmigrantes? Una investigación sobre los deseos educativos de los miembros de la comunidad árabe de Ontario","year":2010,"lang":"en","type":"article","venue":"Dialnet (Universidad de la Rioja)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Arabic; Heritage language; Value (mathematics); Second language; English language; English as a second language","score_opus":0.010879139164801465,"score_gpt":0.2992022145775078,"score_spread":0.2883230754127063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6986648496","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9846347,0.0010951605,0.000063062056,0.004386123,0.000031124968,0.00001589212,0.0000744631,0.0000036928,0.009695781],"genre_scores_gemma":[0.99222636,0.0017593107,0.00013378872,0.00034225325,0.000010621146,0.000014859332,0.00003674172,0.0000023880186,0.0054736757],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.99936396,0.00013319598,0.000020979292,0.000050751754,0.00016380993,0.00026731126],"domain_scores_gemma":[0.9983668,0.0002525269,0.00034582004,0.00004596392,0.0004570692,0.00053193205],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00094287645,0.00014842719,0.00021712833,0.00055597123,0.0037521399,0.0018163673,0.000607587,0.0004385875,0.0025860511],"category_scores_gemma":[0.0028185155,0.00017985556,0.00016454059,0.0009899914,0.0019273766,0.0009890227,0.0010702172,0.0007249515,0.00021585988],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015914225,0.00019942634,0.46237934,0.00037081834,0.00002192506,0.0013695619,0.43404537,0.000059804883,0.0016914993,0.0029872304,0.005172983,0.09154288],"study_design_scores_gemma":[0.000013752725,0.000088361885,0.46172968,0.00034403626,0.00003857404,0.0003795224,0.5018612,0.000068299596,0.00028803485,0.00021594773,0.03495047,0.00002219134],"about_ca_topic_score_codex":0.8876056,"about_ca_topic_score_gemma":0.9648939,"teacher_disagreement_score":0.11239439,"about_ca_system_score_codex":0.014875545,"about_ca_system_score_gemma":0.021960024,"threshold_uncertainty_score":0.22611254},"labels":[],"label_agreement":null},{"id":"W6999392703","doi":"","title":"Clustering Oligarchies","year":2013,"lang":"en","type":"article","venue":"CaltechAUTHORS (California Institute of Technology)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Robustness (evolution); Initialization; Set (abstract data type); Data set; Data point","score_opus":0.015206979535115283,"score_gpt":0.2641469856410328,"score_spread":0.2489400061059175,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6999392703","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41260037,0.00064993167,0.5688147,0.0015326194,0.000107135944,0.00019343985,0.00029377974,0.00042717296,0.015380791],"genre_scores_gemma":[0.9498584,0.00020113811,0.044964124,0.00021593319,0.000043508626,0.00012553323,0.000293582,0.000112910275,0.0041848905],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99473095,0.0015175755,0.0002559329,0.0014225412,0.0014766746,0.0005963645],"domain_scores_gemma":[0.97316223,0.010551256,0.005014963,0.0068411566,0.0031159113,0.0013144844],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010243404,0.0006389484,0.0011640338,0.0019219285,0.0026868926,0.003088169,0.002063576,0.0014385278,0.0026640592],"category_scores_gemma":[0.052460868,0.00047727185,0.00070568826,0.0017676557,0.0042877747,0.0035429818,0.0053321905,0.002086554,0.0006375141],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007378246,0.00015464971,0.02456374,0.0002896545,0.00043310132,0.0003807408,0.0022526341,0.49965727,0.014115364,0.35176617,0.005399566,0.10024929],"study_design_scores_gemma":[0.00007661523,0.0003095847,0.009870821,0.0001097023,0.00008366415,0.00024297007,0.000928592,0.77330816,0.013145747,0.19100578,0.010817012,0.0001012878],"about_ca_topic_score_codex":0.0038528333,"about_ca_topic_score_gemma":0.003645805,"teacher_disagreement_score":0.010243404,"about_ca_system_score_codex":0.003025061,"about_ca_system_score_gemma":0.0018837014,"threshold_uncertainty_score":0.054172933},"labels":[],"label_agreement":null},{"id":"W7017732879","doi":"","title":"Canada – Manitoba Province – Winnipeg – Main Street Looking South from City Hall – Canadian Dog Train and Remains of Old Fort Garry","year":2013,"lang":"en","type":"article","venue":"Digital USD (University of San Diego)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Work (physics); Sinkhole; Settlement (finance)","score_opus":0.010670402446127408,"score_gpt":0.18015374946761437,"score_spread":0.16948334702148696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7017732879","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03770698,0.0046872445,0.0041511124,0.0062744194,0.0012349316,0.00067599746,0.12509713,0.0021888453,0.81798327],"genre_scores_gemma":[0.083415546,0.0027644068,0.0064807786,0.0009547623,0.00006636699,0.00011423439,0.02771136,0.0008626825,0.87762994],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996673,0.0000136437175,0.000008432516,0.00006616476,0.00011488422,0.00012959151],"domain_scores_gemma":[0.9988249,0.000032495784,0.000046949666,0.000040450108,0.00066732714,0.00038789],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002544437,0.00067675166,0.00036760446,0.0021171737,0.0068044374,0.0025387977,0.00089993543,0.00058278,0.22082648],"category_scores_gemma":[0.0010916626,0.0004494204,0.00031480961,0.0052572577,0.00077100145,0.00081938773,0.0013492139,0.000614279,0.053338937],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014242686,0.000029952422,0.015076409,0.00033803255,0.000019254378,0.0004466259,0.0011653745,0.00031722727,0.0011484001,0.005884703,0.84249485,0.13293688],"study_design_scores_gemma":[0.000011229097,0.000013015735,0.057791214,0.00015249908,0.000009858101,0.00021531447,0.0021615669,0.00012463181,0.00059513823,0.0003911696,0.93851316,0.000021234468],"about_ca_topic_score_codex":0.96887344,"about_ca_topic_score_gemma":0.9938261,"teacher_disagreement_score":0.22082648,"about_ca_system_score_codex":0.011480128,"about_ca_system_score_gemma":0.0408256,"threshold_uncertainty_score":0.7387382},"labels":[],"label_agreement":null},{"id":"W7023682103","doi":"","title":"Ottawa Senators HOT TAKES Heading Into The 2023-24 NHL Regular Season","year":2023,"lang":"en","type":"other","venue":"Bulletin of Miscellaneous Information (Royal Gardens Kew)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Heading (navigation); Favourite; Taste; White (mutation); Code (set theory); Begging","score_opus":0.008605565159378567,"score_gpt":0.2199428136018655,"score_spread":0.2113372484424869,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7023682103","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043648384,0.0014575161,0.00036837478,0.012215766,0.011390479,0.00046033805,0.031746663,0.0013073041,0.93668866],"genre_scores_gemma":[0.0073285364,0.00042920356,0.00019414049,0.0016878868,0.00030419623,0.00011338638,0.003922666,0.00040381323,0.9856162],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990221,0.00005821716,0.000025964615,0.000113753995,0.000343506,0.0004365009],"domain_scores_gemma":[0.9971968,0.000047895046,0.00006215955,0.0001106281,0.00086210883,0.0017204193],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0007651366,0.0011220232,0.0005478494,0.0008348738,0.0114398515,0.007834017,0.0015633636,0.0028510576,0.52278495],"category_scores_gemma":[0.0016550518,0.00075341883,0.0006357966,0.001051955,0.0012151957,0.0022832789,0.004687078,0.0031068942,0.29217052],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003326613,0.000013313719,0.00036273524,0.00004028966,0.0000025601762,0.00005181476,0.00012967119,0.000011162009,0.00013049069,0.0004671149,0.99469984,0.0040576584],"study_design_scores_gemma":[0.0000067854958,0.000012959705,0.0022507785,0.00004348397,0.0000027528697,0.000015832245,0.0009132181,0.000012386172,0.00008469249,0.000045942546,0.99659866,0.000012560749],"about_ca_topic_score_codex":0.6452336,"about_ca_topic_score_gemma":0.92842203,"teacher_disagreement_score":0.52278495,"about_ca_system_score_codex":0.008185916,"about_ca_system_score_gemma":0.017122352,"threshold_uncertainty_score":0.7137114},"labels":[],"label_agreement":null},{"id":"W7024212856","doi":"","title":"The Relevance of International Law Standards to Religious Leaders","year":2023,"lang":"en","type":"other","venue":"White Rose Research Online (University of Leeds, The University of Sheffield, University of York)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Economic and Social Research Council","keywords":"Relevance (law); Accountability; Human rights; International law; Reflexivity; Legitimacy; Norm (philosophy); International human rights law; Plea; Normative; Section (typography)","score_opus":0.03779702065530324,"score_gpt":0.30623318522100945,"score_spread":0.2684361645657062,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7024212856","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030690582,0.011998554,0.017256213,0.24155116,0.0032443572,0.00017703345,0.00006096882,0.00009176154,0.6949293],"genre_scores_gemma":[0.8455781,0.011262739,0.016661713,0.05468378,0.0030300766,0.00040805043,0.00011444663,0.00012935941,0.068131745],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9891026,0.0057427245,0.00056420144,0.0006487088,0.002796415,0.0011454104],"domain_scores_gemma":[0.9816636,0.012445222,0.0014471156,0.0014828622,0.0020593076,0.0009019318],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011396394,0.0002443665,0.00048117316,0.0009598004,0.0056562466,0.012323459,0.0016674061,0.004758988,0.0051781135],"category_scores_gemma":[0.022886056,0.0003550065,0.0004136214,0.0014544257,0.020548603,0.007972352,0.004890116,0.011545692,0.0007500041],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000026683388,0.000017495331,0.00035609098,0.000032659253,0.0000013211466,0.000039578797,0.00784595,0.00007057951,0.000046587968,0.97824144,0.004591164,0.00875456],"study_design_scores_gemma":[0.000010428942,0.00006428594,0.0025163947,0.0012565989,0.000007998184,0.00018364498,0.023950053,0.00051464397,0.00031763222,0.5248693,0.4462723,0.000036680936],"about_ca_topic_score_codex":0.0030100644,"about_ca_topic_score_gemma":0.0054778224,"teacher_disagreement_score":0.012323459,"about_ca_system_score_codex":0.0049196105,"about_ca_system_score_gemma":0.007715572,"threshold_uncertainty_score":0.060270607},"labels":[],"label_agreement":null},{"id":"W7027101180","doi":"","title":"Big Data Clustering: Models and Applications","year":2023,"lang":"en","type":"dissertation","venue":"MacSphere (McMaster University)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Cluster analysis; Heuristic; Categorization; Big data; Identification (biology); Conceptual clustering; Similarity (geometry)","score_opus":0.07643698189621714,"score_gpt":0.2879257419409178,"score_spread":0.21148876004470066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7027101180","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043313997,0.008882922,0.97169185,0.0049466584,0.00037063102,0.0002573353,0.0008895714,0.0007133276,0.0079163145],"genre_scores_gemma":[0.23291396,0.021431923,0.7277683,0.0014588243,0.0021103171,0.0017631116,0.0030751175,0.00039266775,0.009085883],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9953446,0.0020554808,0.00030899164,0.0009459185,0.0011652518,0.00017975722],"domain_scores_gemma":[0.9861385,0.009593642,0.00093571836,0.001524066,0.0015169433,0.00029111537],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0073347082,0.0016343374,0.0019677156,0.0046142917,0.0016408672,0.0060827006,0.004087827,0.003652181,0.0039316462],"category_scores_gemma":[0.021906458,0.0012738883,0.002376051,0.008384294,0.0030728593,0.0074777026,0.0038882315,0.0033256917,0.0016332064],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044086013,0.000087023254,0.0031130111,0.0006337331,0.00020329621,0.0001486991,0.00049159466,0.20905806,0.00020685639,0.6730668,0.015065135,0.09788172],"study_design_scores_gemma":[0.00001010544,0.000016427186,0.0005099151,0.0001257505,0.000021577813,0.00008730518,0.000137703,0.5030043,0.0001448389,0.4804855,0.015427945,0.000028589011],"about_ca_topic_score_codex":0.006829762,"about_ca_topic_score_gemma":0.0047189626,"teacher_disagreement_score":0.0073347082,"about_ca_system_score_codex":0.0038001356,"about_ca_system_score_gemma":0.0020912604,"threshold_uncertainty_score":0.038790107},"labels":[],"label_agreement":null},{"id":"W7027250496","doi":"","title":"A clustering algorithm to find groups with homogeneous preferences","year":2003,"lang":"en","type":"article","venue":"Consultation of the Doctoral Thesis Database (TESEO) (Ministerio de Educación, Cultura y Deporte)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Homogeneous; Cluster analysis; Information system; Field (mathematics); Feature (linguistics)","score_opus":0.03423239114366127,"score_gpt":0.2914982224660955,"score_spread":0.25726583132243425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7027250496","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007343202,0.00029464677,0.98896515,0.00013289056,0.000079904814,0.00042405346,0.00038736698,0.0013171369,0.0010556014],"genre_scores_gemma":[0.017354654,0.00009881219,0.97910565,0.00006880187,0.000036827096,0.00032323904,0.0010692774,0.00017527619,0.0017674898],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963897,0.0006868704,0.00034235363,0.0011620324,0.0011727486,0.000246326],"domain_scores_gemma":[0.99528027,0.0016865623,0.0002742111,0.00077107386,0.0017923972,0.00019556303],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003282479,0.0019883711,0.0028049906,0.008999227,0.0038144214,0.0023621921,0.0045536486,0.003267151,0.00554782],"category_scores_gemma":[0.009667002,0.0014059726,0.0033750692,0.010381742,0.0013725576,0.0028789889,0.0029683402,0.0018442008,0.0044940617],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00055470766,0.00046279604,0.003039243,0.00037948875,0.0005416985,0.0001262678,0.00074170873,0.051597875,0.0113390125,0.014207384,0.019105842,0.897904],"study_design_scores_gemma":[0.00047838432,0.0003950135,0.003333141,0.00012697694,0.00043483655,0.00069442927,0.00065545994,0.91140974,0.013402508,0.04769262,0.021232493,0.00014436094],"about_ca_topic_score_codex":0.011207135,"about_ca_topic_score_gemma":0.013611262,"teacher_disagreement_score":0.011207135,"about_ca_system_score_codex":0.0014754597,"about_ca_system_score_gemma":0.0029266637,"threshold_uncertainty_score":0.022283792},"labels":[],"label_agreement":null},{"id":"W7033311453","doi":"","title":"Predicting the performance of tubular daylighting devices","year":2013,"lang":"en","type":"article","venue":"NPARC","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Daylighting; Daylight; Illuminance; Roof; Solar energy; Efficient energy use; Skylight; Glazing","score_opus":0.013944177839396353,"score_gpt":0.25477907981208553,"score_spread":0.24083490197268917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7033311453","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9796382,0.00020490128,0.016647035,0.00007658267,0.00006249768,0.000036043646,0.0007637065,0.00045668386,0.0021142126],"genre_scores_gemma":[0.994739,0.000064615466,0.0032526427,0.000010782943,0.00000834372,0.000011438869,0.00096471887,0.000020931508,0.00092756865],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997638,0.000026496327,0.000009234242,0.00008430374,0.000058633275,0.000057620084],"domain_scores_gemma":[0.99943835,0.00022501044,0.00004845622,0.000050811326,0.00017676016,0.00006053395],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003577762,0.00069850695,0.00048556132,0.00058438163,0.00028584586,0.00058712927,0.0006224624,0.0008379003,0.0012782558],"category_scores_gemma":[0.001634097,0.00017877347,0.00030167002,0.0006842359,0.00021203412,0.00055951276,0.00031386383,0.00034190892,0.0011933123],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009710512,0.00042769327,0.08862142,0.00014558165,0.000091923124,0.00022248675,0.00012168339,0.7784144,0.031903632,0.00068537676,0.0060469145,0.09234779],"study_design_scores_gemma":[0.000011615229,0.00024864243,0.029265847,0.000007063918,0.000015583428,0.00003734981,0.0000772463,0.96123505,0.008081235,0.00030826704,0.0006917018,0.000020467978],"about_ca_topic_score_codex":0.0067009204,"about_ca_topic_score_gemma":0.0064843637,"teacher_disagreement_score":0.0067009204,"about_ca_system_score_codex":0.0005305558,"about_ca_system_score_gemma":0.00024648441,"threshold_uncertainty_score":0.0133238435},"labels":[],"label_agreement":null},{"id":"W7033425217","doi":"","title":"Quelle politique pour répondre aux discriminations à l'université ?","year":2024,"lang":"fr","type":"article","venue":"Digital Access to Libraries (Université catholique de Louvain (UCL), l'Université de Namur (UNamur) and the Université Saint-Louis (USL-B))","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Social power; Limiting","score_opus":0.019022440958892668,"score_gpt":0.24884224754802545,"score_spread":0.22981980658913279,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7033425217","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04787395,0.004459553,0.064240806,0.13502997,0.0008877465,0.00006935995,0.00017784025,0.0001333243,0.7471275],"genre_scores_gemma":[0.91836315,0.00211436,0.008641097,0.009243133,0.00066793885,0.00019234803,0.00009420728,0.00013087371,0.0605529],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9887671,0.00553684,0.000314433,0.0016254669,0.0022863024,0.0014698404],"domain_scores_gemma":[0.99352807,0.002676718,0.0007787822,0.0011327389,0.0012449902,0.0006386658],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006927191,0.00045338945,0.0005792798,0.0020098428,0.007925425,0.012525295,0.0012079213,0.0040243105,0.020561656],"category_scores_gemma":[0.017264688,0.00036688393,0.0006545067,0.003158594,0.01812612,0.01471839,0.0051531657,0.0058077723,0.0019099496],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004508291,0.0000066907437,0.00037528793,0.00001532865,0.0000042458364,0.000012705461,0.00084838166,0.000105345694,0.000028835966,0.99189055,0.001583209,0.0051248837],"study_design_scores_gemma":[0.000014418722,0.000018864115,0.0019247626,0.00024968584,0.000014949364,0.00011280827,0.0033971833,0.0007383146,0.00026565412,0.8364969,0.15674241,0.000024172343],"about_ca_topic_score_codex":0.01165518,"about_ca_topic_score_gemma":0.012983781,"teacher_disagreement_score":0.020561656,"about_ca_system_score_codex":0.0098782275,"about_ca_system_score_gemma":0.0058935066,"threshold_uncertainty_score":0.0716719},"labels":[],"label_agreement":null},{"id":"W7038497428","doi":"","title":"Hybrid Fall Rye in Canada","year":2022,"lang":"en","type":"other","venue":"University Library (University of Saskatchewan)","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"","score_opus":0.006230477434639576,"score_gpt":0.16334215867167887,"score_spread":0.15711168123703928,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7038497428","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.050585795,0.0028677366,0.0040618693,0.017316505,0.002966312,0.000825089,0.017630601,0.0049870587,0.89875907],"genre_scores_gemma":[0.05285049,0.0005955097,0.002642023,0.0010100009,0.0000790281,0.00007100252,0.003431358,0.00045540876,0.93886524],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99743396,0.00009482635,0.00004024827,0.00029422503,0.0013047263,0.00083206606],"domain_scores_gemma":[0.9949773,0.00013615945,0.000057148172,0.00024807066,0.003209869,0.0013714501],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014538518,0.0005006873,0.0005535489,0.0022531506,0.008701998,0.0060509224,0.0019319823,0.0012218452,0.13716811],"category_scores_gemma":[0.002232879,0.0004422265,0.0005740068,0.0026145496,0.0015706957,0.00097533874,0.0025583666,0.0013543752,0.028724313],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035351346,0.00019586235,0.007537784,0.0001772264,0.000038779657,0.0004455371,0.001172019,0.000993578,0.0018984209,0.023208557,0.8035471,0.16043168],"study_design_scores_gemma":[0.000045688463,0.00005066158,0.015279139,0.00007132419,0.000011969201,0.000060922157,0.001567097,0.00076730066,0.00063795573,0.001041382,0.98043764,0.000028856051],"about_ca_topic_score_codex":0.9695404,"about_ca_topic_score_gemma":0.98760766,"teacher_disagreement_score":0.13716811,"about_ca_system_score_codex":0.03179707,"about_ca_system_score_gemma":0.13724293,"threshold_uncertainty_score":0.45887303},"labels":[],"label_agreement":null},{"id":"W70833783","doi":"","title":"Central clustering of categorical data with automated feature weighting","year":2013,"lang":"en","type":"article","venue":"International Joint Conference on Artificial Intelligence","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Categorical variable; Cluster analysis; Computer science; Weighting; Artificial intelligence; Data mining; Kernel (algebra); Pattern recognition (psychology); Feature (linguistics); Clustering high-dimensional data; Machine learning; Mathematics","score_opus":0.12553371645470476,"score_gpt":0.35394447805044793,"score_spread":0.22841076159574317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W70833783","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008407294,0.000078734716,0.99095494,0.000028098479,0.000014380192,0.00003105683,0.00002185071,0.0003554534,0.00010825529],"genre_scores_gemma":[0.1754855,0.000099424986,0.8231546,0.000047426653,0.000037833248,0.00014732324,0.00031467827,0.0001507962,0.00056230894],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958813,0.0013127421,0.0003728524,0.0010755209,0.0011028593,0.0002546083],"domain_scores_gemma":[0.99282444,0.0020487767,0.0007751234,0.0015522166,0.0025859436,0.00021360195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052353586,0.00069042074,0.0017461069,0.0029871515,0.0012399205,0.001645401,0.0026034575,0.0011392995,0.00072693493],"category_scores_gemma":[0.0148013355,0.0006166576,0.0012988565,0.0027857996,0.001401549,0.0025616358,0.0028833654,0.0015456171,0.0007105593],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004938879,0.00024602495,0.007271534,0.00022586317,0.00022688671,0.00013148275,0.0008277399,0.11880648,0.023282127,0.032352813,0.003906422,0.81222874],"study_design_scores_gemma":[0.000033515636,0.000088425644,0.0018200938,0.000022066075,0.000035727466,0.00017513071,0.00013899557,0.9422653,0.012625783,0.040202994,0.0025187517,0.00007325795],"about_ca_topic_score_codex":0.0025438755,"about_ca_topic_score_gemma":0.002387199,"teacher_disagreement_score":0.0052353586,"about_ca_system_score_codex":0.0009766532,"about_ca_system_score_gemma":0.0017102186,"threshold_uncertainty_score":0.02768755},"labels":[],"label_agreement":null},{"id":"W7095231604","doi":"","title":"Algorithm for Minimum Sum-of-Squares Clustering","year":2008,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cluster analysis; Context (archaeology); Fuzzy clustering; Sequence (biology); Field (mathematics); Fuzzy logic","score_opus":0.046744747967012674,"score_gpt":0.3169603332495066,"score_spread":0.27021558528249395,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7095231604","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026141068,0.00014644096,0.9915235,0.00030648964,0.00010584163,0.00021283922,0.0002930659,0.0026772215,0.0021205829],"genre_scores_gemma":[0.024251305,0.000082525505,0.9689761,0.00013005712,0.0000529684,0.00046984566,0.0010698508,0.0004938728,0.0044735735],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978236,0.0005631039,0.00015516204,0.00047881182,0.0007595244,0.00021988121],"domain_scores_gemma":[0.99678,0.0010423771,0.00022284019,0.0006059441,0.0011892401,0.00015978215],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020041333,0.0018184884,0.0021205966,0.0030449023,0.0020359473,0.0020676749,0.0041497694,0.003176348,0.020413134],"category_scores_gemma":[0.008417406,0.0010733656,0.0019623963,0.0033153412,0.0010011571,0.0025873282,0.003462869,0.002779952,0.014343062],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005213331,0.00021650559,0.00089311594,0.0003868544,0.00022438043,0.00012143149,0.00031257846,0.21503043,0.006892188,0.025458183,0.040002618,0.7099404],"study_design_scores_gemma":[0.00024377152,0.00013009668,0.0005682085,0.000037055557,0.00005530147,0.00023075179,0.00016534644,0.9406399,0.0056360043,0.034589183,0.017650694,0.000053620166],"about_ca_topic_score_codex":0.00847155,"about_ca_topic_score_gemma":0.015300513,"teacher_disagreement_score":0.020413134,"about_ca_system_score_codex":0.001768006,"about_ca_system_score_gemma":0.0053178524,"threshold_uncertainty_score":0.06828874},"labels":[],"label_agreement":null},{"id":"W7133093713","doi":"","title":"Optimal and Greedy Algorithms for Clustering with Applications to Data Science","year":2023,"lang":"","type":"dissertation","venue":"TSpace","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cluster analysis; Constrained clustering; Correlation clustering; Scalability; Class (philosophy); Heuristic; Bounding overwatch; CURE data clustering algorithm; Clustering high-dimensional data","score_opus":0.12314620837366497,"score_gpt":0.4590791525163828,"score_spread":0.33593294414271785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7133093713","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011106571,0.00072251965,0.994783,0.00052368524,0.00007461323,0.0000510142,0.00006688529,0.00026028746,0.0024073059],"genre_scores_gemma":[0.038240727,0.00169537,0.95570207,0.00033363476,0.00021959508,0.0003366118,0.00034435565,0.00031523805,0.0028123131],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970655,0.0011609314,0.00016028355,0.0006463343,0.0007834157,0.00018348283],"domain_scores_gemma":[0.9935191,0.0042289356,0.00037695144,0.00084492855,0.00084717356,0.00018288917],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004055087,0.0020091601,0.0018181716,0.002029874,0.0015809068,0.002583912,0.0025464587,0.0021927247,0.004324626],"category_scores_gemma":[0.014675074,0.0010508738,0.0017995953,0.004532229,0.002789458,0.003636192,0.0033157398,0.0045958133,0.002167296],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008751658,0.00009362031,0.0005542875,0.00046789314,0.00009992506,0.00006078897,0.00020573627,0.4064524,0.0012328812,0.4076006,0.014686689,0.16845761],"study_design_scores_gemma":[0.00003030896,0.000029825862,0.00010976105,0.00005805028,0.000015468231,0.0000439925,0.000049212304,0.61733663,0.0007814887,0.3720173,0.0095066,0.000021250815],"about_ca_topic_score_codex":0.004100488,"about_ca_topic_score_gemma":0.0053645833,"teacher_disagreement_score":0.004324626,"about_ca_system_score_codex":0.0033464339,"about_ca_system_score_gemma":0.0043556876,"threshold_uncertainty_score":0.02428013},"labels":[],"label_agreement":null},{"id":"W7160045919","doi":"10.20894/ijcoa.101.014.001.001","title":"A Multiview Point Perspective on Clustering Algorithms for Similarity Evaluation","year":2025,"lang":"","type":"article","venue":"International Journal of Computing Algorithm","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Cluster analysis; Viewpoints; Similarity (geometry); Correlation clustering; Fuzzy clustering; Set (abstract data type); Perspective (graphical); Pattern recognition (psychology); Single-linkage clustering","score_opus":0.056779147601151984,"score_gpt":0.43804825507439044,"score_spread":0.38126910747323844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7160045919","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007005039,0.00096116133,0.9971739,0.000104943945,0.000063442334,0.000027434571,0.000027369526,0.00010407603,0.00083720125],"genre_scores_gemma":[0.062176187,0.0025618135,0.93251663,0.00015841081,0.00043314233,0.00017758469,0.00020254096,0.00015701281,0.0016167699],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956312,0.0015568995,0.00027921432,0.00087467046,0.0015012254,0.00015685118],"domain_scores_gemma":[0.9963283,0.0016850446,0.00025098905,0.00047248328,0.001148956,0.00011423328],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003915123,0.0015447484,0.0019649942,0.0048633125,0.00087231066,0.0035348886,0.003207768,0.0021297776,0.0029470131],"category_scores_gemma":[0.009668654,0.0007227575,0.0021635867,0.00695669,0.0019318884,0.0040966556,0.0022424166,0.003214719,0.0016185559],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019312774,0.000090963804,0.0016297111,0.0007372521,0.0003009588,0.00018811191,0.00062520924,0.26333085,0.009859544,0.27495784,0.007079214,0.44100726],"study_design_scores_gemma":[0.000023255923,0.00020687316,0.00080072536,0.00017068737,0.000068270296,0.0003825548,0.00021765014,0.82324266,0.005464734,0.14820589,0.021123078,0.00009357134],"about_ca_topic_score_codex":0.004363552,"about_ca_topic_score_gemma":0.0020906853,"teacher_disagreement_score":0.0048633125,"about_ca_system_score_codex":0.0020241896,"about_ca_system_score_gemma":0.0009159414,"threshold_uncertainty_score":0.020705342},"labels":[],"label_agreement":null},{"id":"W7160145418","doi":"10.20894/ijcoa.101.014.001.21985","title":"A Multiview Point Perspective on Clustering Algorithms for Similarity Evaluation","year":2025,"lang":"","type":"article","venue":"International Journal of Computing Algorithm","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Cluster analysis; Viewpoints; Similarity (geometry); Correlation clustering; Fuzzy clustering; Set (abstract data type); Perspective (graphical); Pattern recognition (psychology); Single-linkage clustering","score_opus":0.056779147601151984,"score_gpt":0.43804825507439044,"score_spread":0.38126910747323844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7160145418","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007005039,0.00096116133,0.9971739,0.000104943945,0.000063442334,0.000027434571,0.000027369526,0.00010407603,0.00083720125],"genre_scores_gemma":[0.062176187,0.0025618135,0.93251663,0.00015841081,0.00043314233,0.00017758469,0.00020254096,0.00015701281,0.0016167699],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956312,0.0015568995,0.00027921432,0.00087467046,0.0015012254,0.00015685118],"domain_scores_gemma":[0.9963283,0.0016850446,0.00025098905,0.00047248328,0.001148956,0.00011423328],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003915123,0.0015447484,0.0019649942,0.0048633125,0.00087231066,0.0035348886,0.003207768,0.0021297776,0.0029470131],"category_scores_gemma":[0.009668654,0.0007227575,0.0021635867,0.00695669,0.0019318884,0.0040966556,0.0022424166,0.003214719,0.0016185559],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019312774,0.000090963804,0.0016297111,0.0007372521,0.0003009588,0.00018811191,0.00062520924,0.26333085,0.009859544,0.27495784,0.007079214,0.44100726],"study_design_scores_gemma":[0.000023255923,0.00020687316,0.00080072536,0.00017068737,0.000068270296,0.0003825548,0.00021765014,0.82324266,0.005464734,0.14820589,0.021123078,0.00009357134],"about_ca_topic_score_codex":0.004363552,"about_ca_topic_score_gemma":0.0020906853,"teacher_disagreement_score":0.0048633125,"about_ca_system_score_codex":0.0020241896,"about_ca_system_score_gemma":0.0009159414,"threshold_uncertainty_score":0.020705342},"labels":[],"label_agreement":null},{"id":"W74666794","doi":"10.1007/978-3-642-22185-9_31","title":"User Centric Homogeneity-Based Clustering Approach for Intelligence Computation","year":2011,"lang":"en","type":"book-chapter","venue":"Communications in computer and information science","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Cluster analysis; Homogeneity (statistics); Computer science; Computation; Data mining; Hierarchical clustering; Software; Information retrieval; Artificial intelligence; Machine learning; Algorithm","score_opus":0.0925614050617813,"score_gpt":0.3398276076177767,"score_spread":0.2472662025559954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W74666794","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042381953,0.00063134386,0.9901815,0.00014822594,0.00006852718,0.000075118616,0.00016581119,0.00050058274,0.003990593],"genre_scores_gemma":[0.22784011,0.0010820987,0.7548665,0.00018633832,0.00022243067,0.0002934089,0.00091419317,0.00027495666,0.0143199535],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99881494,0.00036454867,0.00005228304,0.00023177845,0.00044186556,0.00009460376],"domain_scores_gemma":[0.9992354,0.00021529145,0.00004091306,0.00018339948,0.00028352856,0.000041570664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010883145,0.0006665753,0.0011712757,0.0017260219,0.0010975813,0.0017050568,0.002746347,0.0008634379,0.00373509],"category_scores_gemma":[0.0019435274,0.00036226408,0.0011087744,0.003388577,0.0006424084,0.0017839251,0.0014540179,0.0011630766,0.0016666935],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032516939,0.00023247054,0.0017811739,0.00036511204,0.00035698432,0.00014767569,0.0005761245,0.16176476,0.010980688,0.32635328,0.029993648,0.4671229],"study_design_scores_gemma":[0.000015180996,0.00008241962,0.0011964561,0.000024494564,0.00006615754,0.00017761072,0.00016540888,0.8751455,0.0043797307,0.10294565,0.015764685,0.00003671463],"about_ca_topic_score_codex":0.003409075,"about_ca_topic_score_gemma":0.006307083,"teacher_disagreement_score":0.00373509,"about_ca_system_score_codex":0.0013864104,"about_ca_system_score_gemma":0.001166729,"threshold_uncertainty_score":0.01249516},"labels":[],"label_agreement":null},{"id":"W75896675","doi":"","title":"Characterization of Linkage-based Clustering","year":2010,"lang":"en","type":"article","venue":"","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Computer science; Correlation clustering; Conceptual clustering; CURE data clustering algorithm; Canopy clustering algorithm; Artificial intelligence; Data mining; Fuzzy clustering; Constrained clustering; Data stream clustering; Machine learning","score_opus":0.014460533485080477,"score_gpt":0.2774060828660243,"score_spread":0.26294554938094383,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W75896675","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03606799,0.00047437786,0.9533435,0.00073237333,0.000058595695,0.00014900522,0.00041291775,0.00041459675,0.008346611],"genre_scores_gemma":[0.63857657,0.0013303575,0.3494251,0.00072592916,0.00039172053,0.00085138413,0.0023260468,0.0005158083,0.00585708],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99157435,0.002284944,0.0006138394,0.0015376187,0.0035547332,0.00043442263],"domain_scores_gemma":[0.9772489,0.00788405,0.0036955217,0.004613285,0.0055190525,0.0010391154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0060170987,0.0007674178,0.0011369799,0.0045827394,0.0021584528,0.004899841,0.002892685,0.0025220923,0.0035879523],"category_scores_gemma":[0.031209188,0.0005548841,0.0014388402,0.0038408958,0.0033567364,0.0077210614,0.0041049146,0.002236297,0.0013175268],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010828739,0.000068212554,0.004663803,0.00024576398,0.00009212151,0.0001533308,0.00063030195,0.039961774,0.0046222913,0.90598404,0.0026929453,0.040776987],"study_design_scores_gemma":[0.000025398855,0.00014679103,0.0020563053,0.00007413673,0.000037780537,0.000547151,0.0002843996,0.22586426,0.0037354913,0.7540729,0.013094656,0.000060645558],"about_ca_topic_score_codex":0.00048682583,"about_ca_topic_score_gemma":0.00032195242,"teacher_disagreement_score":0.0060170987,"about_ca_system_score_codex":0.0018792454,"about_ca_system_score_gemma":0.0012036929,"threshold_uncertainty_score":0.031821847},"labels":[],"label_agreement":null},{"id":"W958955349","doi":"10.1017/cbo9781139031998.014","title":"Statistics of large-scale structure","year":2012,"lang":"en","type":"book-chapter","venue":"Cambridge University Press eBooks","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Scale (ratio); Statistics; Geography; Mathematics; Cartography","score_opus":0.018915480158652145,"score_gpt":0.23660011701962425,"score_spread":0.2176846368609721,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W958955349","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7073193,0.009555588,0.23125862,0.0028654994,0.00023549331,0.000090724614,0.009250977,0.0015633388,0.037860457],"genre_scores_gemma":[0.98420113,0.0017120782,0.00630818,0.00009069426,0.00017912874,0.000057729438,0.004263442,0.00014196707,0.0030455806],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99909735,0.00016415777,0.000048243153,0.0003098601,0.00028416663,0.000096190604],"domain_scores_gemma":[0.9884615,0.0064062467,0.0016907967,0.0018410926,0.001011915,0.00058841653],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015027641,0.0002413492,0.00061024324,0.003103157,0.0005611572,0.002353039,0.00081105705,0.0006116095,0.0050875857],"category_scores_gemma":[0.018340005,0.0003110808,0.0002856941,0.003511617,0.001815522,0.0033004517,0.0011218116,0.0008167953,0.0011625045],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001712899,0.00008580989,0.11603001,0.0004968664,0.00025280807,0.00037942154,0.001028462,0.081758864,0.003787423,0.66266024,0.027622035,0.105726756],"study_design_scores_gemma":[0.000021083393,0.00006724345,0.10390349,0.000072285904,0.000029506824,0.0005270095,0.0003131256,0.09098039,0.001070135,0.7916932,0.011242139,0.000080374484],"about_ca_topic_score_codex":0.00171936,"about_ca_topic_score_gemma":0.0015444306,"teacher_disagreement_score":0.0050875857,"about_ca_system_score_codex":0.000970705,"about_ca_system_score_gemma":0.00030959112,"threshold_uncertainty_score":0.01701969},"labels":[],"label_agreement":null}]}