{"meta":{"query_hash":"962ff7d3bf67","filters":{"venue":"Knowledge and Information Systems"},"cohort_total":123,"direct_labels_cover":0,"predictions_cover":123,"exported":123,"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/962ff7d3bf67","api":"https://metacan.xera.ac/api/v1/cohort?venue=Knowledge+and+Information+Systems"},"results":[{"id":"W1599200063","doi":"10.1007/s10115-015-0871-2","title":"Partial materialization for online analytical processing over multi-tagged document collections","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Database Systems and Queries","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":"University of Waterloo; Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Computer science; Centroid; Information retrieval; Medoid; Document clustering; Digital library; Data mining; Artificial intelligence; Cluster analysis","score_opus":0.04078338301597355,"score_gpt":0.3180951839573361,"score_spread":0.27731180094136254,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1599200063","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":[0.031239547,0.0004666232,0.954784,0.00035311055,0.000071345836,0.00044748856,0.0013761518,0.009003701,0.002257974],"genre_scores_gemma":[0.21575274,0.00050695206,0.7720177,0.00018624983,0.00014075254,0.0007977035,0.004886377,0.0008160646,0.004895428],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978696,0.00040304978,0.00035559133,0.0004327748,0.0007509559,0.000187995],"domain_scores_gemma":[0.99222845,0.0021749125,0.00069840776,0.0029066931,0.0017104524,0.00028093957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002305905,0.0010386108,0.0013090145,0.0038835816,0.0016125975,0.004334277,0.0026177135,0.0007460133,0.0068416586],"category_scores_gemma":[0.012569184,0.00069117325,0.0012608279,0.0078025176,0.0013569426,0.0056377854,0.0047327313,0.00088155956,0.0045478106],"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":[0.0012372271,0.0002480588,0.0029594293,0.0005707243,0.000105318155,0.00060482736,0.0016574,0.027705882,0.04573967,0.03069949,0.018357761,0.87011427],"study_design_scores_gemma":[0.00030811617,0.0015100654,0.005626458,0.00017335157,0.00030136414,0.0026976755,0.0037671686,0.6165429,0.17494333,0.09046786,0.103354685,0.00030704378],"about_ca_topic_score_codex":0.002478098,"about_ca_topic_score_gemma":0.0030295064,"teacher_disagreement_score":0.0068416586,"about_ca_system_score_codex":0.0011350607,"about_ca_system_score_gemma":0.0025492616,"threshold_uncertainty_score":0.022887647},"labels":[],"label_agreement":null},{"id":"W1967641500","doi":"10.1007/s101150200002","title":"Correlation-Based Web Document Clustering for Adaptive Web Interface Design","year":2002,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Web Data Mining and Analysis","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":"Simon Fraser University","funders":"","keywords":"Computer science; Web page; Cluster analysis; Web modeling; Data Web; Web mining; Information retrieval; Web navigation; Data mining; World Wide Web; Static web page; Web design; Machine learning","score_opus":0.03196732839296157,"score_gpt":0.24740335138207364,"score_spread":0.21543602298911207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967641500","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":[0.03510398,0.00018535659,0.959173,0.00007716326,0.00002930349,0.0001665244,0.00016710252,0.0041724867,0.00092518993],"genre_scores_gemma":[0.3690826,0.00013955847,0.6275714,0.000057945286,0.000033773536,0.0002467407,0.0006263243,0.00040137375,0.0018403493],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983304,0.0004341303,0.00012923642,0.00035134924,0.0006477023,0.000107197775],"domain_scores_gemma":[0.9943229,0.002099641,0.0004827613,0.0008518746,0.0020633389,0.00017957459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015296384,0.00056980876,0.0009052391,0.0036388298,0.0010242278,0.0014537845,0.0013205411,0.0007561197,0.0027538182],"category_scores_gemma":[0.008805963,0.00046098078,0.00066917605,0.003256671,0.00045964253,0.0015746362,0.0009398226,0.0007335302,0.0011761597],"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":[0.0006445758,0.0005532188,0.010634258,0.00027651724,0.00014260315,0.00013739512,0.00034981483,0.05479311,0.03724574,0.0069786417,0.0075631593,0.8806809],"study_design_scores_gemma":[0.00003964241,0.00009929387,0.003815333,0.000015524103,0.000058695845,0.00016999213,0.00008323353,0.96643996,0.023744151,0.0031812608,0.0023204621,0.000032547414],"about_ca_topic_score_codex":0.00613893,"about_ca_topic_score_gemma":0.011230748,"teacher_disagreement_score":0.00613893,"about_ca_system_score_codex":0.0010678652,"about_ca_system_score_gemma":0.0015632177,"threshold_uncertainty_score":0.012206376},"labels":[],"label_agreement":null},{"id":"W1968727285","doi":"10.1007/s10115-012-0546-1","title":"Out-of-core detection of periodicity from sequence databases","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Algorithms and Data Compression","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 Calgary","funders":"","keywords":"Tree traversal; Computer science; Scalability; Suffix tree; Pruning; Suffix; Out-of-core algorithm; Sequence (biology); Auxiliary memory; Tree (set theory); Core (optical fiber); Data structure; Time complexity; Algorithm; Generalized suffix tree; Series (stratigraphy); Theoretical computer science; Data mining; Database; Mathematics","score_opus":0.06032132968878795,"score_gpt":0.29010241177526475,"score_spread":0.2297810820864768,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1968727285","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":[0.46887526,0.0014413733,0.525161,0.00030549936,0.00017704738,0.00012610768,0.000867567,0.0012380261,0.0018080925],"genre_scores_gemma":[0.8406058,0.0005502999,0.15538588,0.00012453523,0.00016449158,0.00008545324,0.0018812797,0.00011688129,0.0010853515],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99866223,0.00023122116,0.00018304396,0.00025272547,0.0004876338,0.00018310962],"domain_scores_gemma":[0.9943929,0.0021763209,0.00068838295,0.0014352552,0.0010060092,0.00030114464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014602747,0.00050571054,0.001186384,0.0031229153,0.00060277624,0.0010710929,0.00087952666,0.0008059649,0.000913612],"category_scores_gemma":[0.008954189,0.00033573274,0.00039365963,0.0021533761,0.0004600344,0.0015246354,0.0013038141,0.0008072993,0.00059897185],"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":[0.003343778,0.00056335534,0.025974939,0.00044408272,0.00015413445,0.0012993771,0.0004286017,0.024923326,0.12706633,0.01170883,0.0042313254,0.7998619],"study_design_scores_gemma":[0.000070241884,0.00064981,0.023376731,0.00007787119,0.00010446127,0.0026224332,0.0002936713,0.8805722,0.06815063,0.018840127,0.005191656,0.000050235773],"about_ca_topic_score_codex":0.00076661195,"about_ca_topic_score_gemma":0.001078438,"teacher_disagreement_score":0.0031229153,"about_ca_system_score_codex":0.0003033847,"about_ca_system_score_gemma":0.001042755,"threshold_uncertainty_score":0.0077227354},"labels":[],"label_agreement":null},{"id":"W1970296249","doi":"10.1007/s10115-011-0379-3","title":"Aggregate keyword search on large relational databases","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","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":"Simon Fraser University","funders":"","keywords":"Computer science; Aggregate (composite); Information retrieval; Tuple; Set (abstract data type); Keyword density; Relational database; Keyword search; Matching (statistics); Data mining; Table (database); Focus (optics); Database; Mathematics","score_opus":0.06452476657364406,"score_gpt":0.26599489127396836,"score_spread":0.2014701247003243,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1970296249","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":[0.3098049,0.009358137,0.64986145,0.0020152319,0.00028704555,0.00036235398,0.007928794,0.012066835,0.008315222],"genre_scores_gemma":[0.65268636,0.0031362555,0.32792452,0.00035348753,0.0003384192,0.00026368708,0.0075423573,0.0006213183,0.0071336576],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934994,0.0013764652,0.0008330661,0.0007347416,0.0030047223,0.00055163587],"domain_scores_gemma":[0.9855137,0.008071089,0.0007477979,0.0033933034,0.0017052514,0.0005688084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034998413,0.0008962854,0.0038381678,0.00650467,0.001738496,0.0056625092,0.0022932813,0.0013388653,0.004895737],"category_scores_gemma":[0.019153943,0.0011228877,0.0012954368,0.014570538,0.0010241508,0.012638628,0.0045952224,0.0011653068,0.0021864998],"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":[0.003238755,0.0006049239,0.013749818,0.0018559156,0.00072634144,0.0010540867,0.0013731569,0.15339872,0.021497715,0.06445933,0.047334284,0.69070685],"study_design_scores_gemma":[0.00018551611,0.000303434,0.0029263163,0.000089590285,0.00043143536,0.0007381408,0.0010320254,0.807201,0.012337389,0.16375433,0.010931452,0.00006939976],"about_ca_topic_score_codex":0.0067887036,"about_ca_topic_score_gemma":0.013997144,"teacher_disagreement_score":0.0067887036,"about_ca_system_score_codex":0.0015252184,"about_ca_system_score_gemma":0.0027428586,"threshold_uncertainty_score":0.01850915},"labels":[],"label_agreement":null},{"id":"W1971159983","doi":"10.1007/s10115-012-0610-x","title":"Fuzzy set-oriented neural networks based on fuzzy polynomial inference and dynamic genetic optimization","year":2013,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":4,"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":"Mathematical optimization; Fuzzy logic; Computer science; Polynomial; Mathematics; Adaptive neuro fuzzy inference system; Neuro-fuzzy; Algorithm; Fuzzy control system; Artificial intelligence","score_opus":0.007798783949087225,"score_gpt":0.22675202680502993,"score_spread":0.2189532428559427,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1971159983","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":[0.009897048,0.00036173058,0.98638004,0.00013595371,0.00006427047,0.000026402544,0.000023238184,0.00013243248,0.0029788096],"genre_scores_gemma":[0.6379192,0.00073379517,0.35676557,0.00010332786,0.000106272346,0.00016583248,0.00009718555,0.000062936604,0.004045922],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99972934,0.0000794734,0.000014286185,0.00005283447,0.000097803655,0.000026244563],"domain_scores_gemma":[0.9994184,0.0003148129,0.000058950827,0.000034126875,0.00015373279,0.000020059591],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007482593,0.0005039724,0.0010439482,0.0006174254,0.0005895737,0.0010259594,0.0013504943,0.0011205625,0.0020991105],"category_scores_gemma":[0.0029065707,0.0003329199,0.00062661956,0.001102346,0.00068288646,0.0012666242,0.00060143165,0.0011041746,0.00039857827],"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":[0.00005085978,0.000027912669,0.00014259653,0.000056239227,0.000031271113,0.00002589365,0.000045851113,0.91611993,0.0013711368,0.029270127,0.0005777501,0.05228034],"study_design_scores_gemma":[0.0000031184848,0.00000446906,0.000019084808,0.000002966125,0.0000036843437,0.0000036847653,0.0000015547485,0.9960437,0.00017928371,0.003617097,0.00011859065,0.0000027112399],"about_ca_topic_score_codex":0.009594932,"about_ca_topic_score_gemma":0.008683268,"teacher_disagreement_score":0.009594932,"about_ca_system_score_codex":0.0011479739,"about_ca_system_score_gemma":0.0009389539,"threshold_uncertainty_score":0.019078135},"labels":[],"label_agreement":null},{"id":"W1971324060","doi":"10.1007/pl00011676","title":"Parallel and Sequential Algorithms for Data Mining Using Inductive Logic","year":2001,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada); Queen's University","funders":"","keywords":"Computer science; Inductive logic programming; Algorithm; Data mining; Artificial intelligence","score_opus":0.09966565189006675,"score_gpt":0.33770732626556443,"score_spread":0.23804167437549767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1971324060","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":[0.0029701754,0.00037514145,0.9938308,0.000234148,0.00005884591,0.00010815413,0.00008015616,0.00076835154,0.0015741646],"genre_scores_gemma":[0.0631979,0.00051974965,0.93123806,0.00022146056,0.00023812264,0.00037417584,0.00051303685,0.00018789218,0.0035096938],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99534875,0.0014019227,0.00043842226,0.0009614877,0.0016092723,0.00024011578],"domain_scores_gemma":[0.98508644,0.01046018,0.0004107839,0.002330694,0.0014974966,0.00021442458],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006658421,0.0012576891,0.001628465,0.0032928446,0.0018954824,0.002954239,0.0033026955,0.0011573546,0.007881403],"category_scores_gemma":[0.019797703,0.0011188462,0.0022038994,0.004046678,0.0022990087,0.006034729,0.0032039327,0.0030435536,0.0025066074],"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":[0.0005970395,0.0003777491,0.0025917355,0.0005831955,0.00026995843,0.00017378587,0.0003038859,0.07629567,0.0034756945,0.16132808,0.008006992,0.7459961],"study_design_scores_gemma":[0.00013565234,0.00009630648,0.00040276072,0.00005178773,0.00012516954,0.0002285492,0.00007631966,0.439181,0.004065546,0.54617023,0.009438873,0.000027774935],"about_ca_topic_score_codex":0.0029898384,"about_ca_topic_score_gemma":0.0053814505,"teacher_disagreement_score":0.007881403,"about_ca_system_score_codex":0.0016074192,"about_ca_system_score_gemma":0.0033089044,"threshold_uncertainty_score":0.03521353},"labels":[],"label_agreement":null},{"id":"W1973249903","doi":"10.1007/s10115-012-0585-7","title":"Models of distributed data clustering in peer-to-peer environments","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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":"University of Waterloo; D2L (Canada)","funders":"","keywords":"Cluster analysis; Computer science; Automatic summarization; Scalability; Data mining; Distributed computing; Process (computing); Data stream clustering; Node (physics); Correlation clustering; Set (abstract data type); CURE data clustering algorithm; Machine learning; Artificial intelligence; Database; Engineering","score_opus":0.04619821659333672,"score_gpt":0.2805203146211613,"score_spread":0.23432209802782455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1973249903","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":[0.04077503,0.0011011845,0.9467171,0.0024782182,0.00017599875,0.00021099999,0.00036014372,0.00043766727,0.0077437046],"genre_scores_gemma":[0.79888046,0.00274831,0.176283,0.0004790736,0.00050459115,0.0007756779,0.00075149455,0.00024072776,0.019336723],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9951513,0.0020759262,0.00029720605,0.00103437,0.001012831,0.00042832174],"domain_scores_gemma":[0.98586607,0.009124485,0.0011723804,0.0015482573,0.0015458459,0.00074297184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056922375,0.0010651153,0.0026839734,0.0021260837,0.0025314835,0.0055704876,0.007922711,0.0043453495,0.00475095],"category_scores_gemma":[0.023942003,0.0013313605,0.0013347492,0.0039623925,0.003243704,0.008168394,0.0035086698,0.002215834,0.0014140084],"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":[0.00010423092,0.00008460171,0.0009936631,0.0001938781,0.000088393266,0.00026399083,0.00060769106,0.6990098,0.000530084,0.28388074,0.0028201095,0.01142274],"study_design_scores_gemma":[0.000034219902,0.000026929736,0.00015441286,0.000015909816,0.000030009476,0.000083780054,0.00009799179,0.87758243,0.00015059055,0.1203392,0.0014642058,0.000020335037],"about_ca_topic_score_codex":0.008759757,"about_ca_topic_score_gemma":0.0069611156,"teacher_disagreement_score":0.008759757,"about_ca_system_score_codex":0.003104671,"about_ca_system_score_gemma":0.001997167,"threshold_uncertainty_score":0.030103803},"labels":[],"label_agreement":null},{"id":"W1974256760","doi":"10.1007/s10115-009-0237-8","title":"A general measure of similarity for categorical sequences","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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":"Université de Sherbrooke","funders":"","keywords":"Categorical variable; Similarity (geometry); Computer science; Data mining; Matching (statistics); Similarity measure; Pattern recognition (psychology); Artificial intelligence; Range (aeronautics); Domain (mathematical analysis); Measure (data warehouse); Process (computing); Sequence (biology); Noise (video); Mathematics; Machine learning; Statistics; Image (mathematics)","score_opus":0.024416134772725646,"score_gpt":0.2734250690558867,"score_spread":0.24900893428316107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1974256760","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":[0.045206767,0.0031216904,0.9387853,0.0005631897,0.00040096662,0.00036489175,0.002099721,0.000668155,0.008789389],"genre_scores_gemma":[0.38833877,0.0022537503,0.59616053,0.00058709865,0.0007454078,0.000923077,0.0051826397,0.0001972318,0.005611462],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9895984,0.0018850401,0.0017623209,0.0025479877,0.0037802926,0.00042593863],"domain_scores_gemma":[0.98873156,0.004154519,0.0012577876,0.0026546153,0.0026630107,0.0005384806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047430717,0.0007437362,0.001874827,0.011698879,0.0013157311,0.0051156194,0.0025528811,0.0024731243,0.0038988981],"category_scores_gemma":[0.029129056,0.00035306226,0.0015616792,0.013081081,0.0024794303,0.009679774,0.0032459064,0.0019484857,0.0016329491],"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":[0.00053340464,0.00041827344,0.02348525,0.0016824107,0.00075393694,0.0005963148,0.001421844,0.012402196,0.03142103,0.4322059,0.010525889,0.48455352],"study_design_scores_gemma":[0.00007429229,0.0010050668,0.021629004,0.0005353868,0.0003416663,0.005960668,0.0011411818,0.12685825,0.009745094,0.780867,0.051603366,0.00023901316],"about_ca_topic_score_codex":0.00073313154,"about_ca_topic_score_gemma":0.00069553894,"teacher_disagreement_score":0.011698879,"about_ca_system_score_codex":0.0013687211,"about_ca_system_score_gemma":0.0012609899,"threshold_uncertainty_score":0.025084019},"labels":[],"label_agreement":null},{"id":"W1976274084","doi":"10.1007/s10115-007-0120-4","title":"A framework for simulating real-time multi-agent systems","year":2008,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Multi-Agent Systems and Negotiation","field":"Computer Science","cited_by":12,"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; Multi-agent system; Context (archaeology); Architecture; Distributed computing; Autonomous agent; Real-time computing; Artificial intelligence","score_opus":0.045866521007289794,"score_gpt":0.2892687766946496,"score_spread":0.24340225568735982,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1976274084","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":[0.0042096437,0.00018952676,0.99079376,0.00018260015,0.00006237907,0.00009351562,0.0001272335,0.0009627823,0.0033785107],"genre_scores_gemma":[0.26982003,0.00078816764,0.722161,0.0001375249,0.000101764104,0.0007328403,0.00041780042,0.0003643503,0.005476428],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99883264,0.00045402948,0.00011449617,0.00016358877,0.00030781407,0.0001273451],"domain_scores_gemma":[0.9980578,0.0010171288,0.00015712735,0.00035563513,0.00022976383,0.00018253725],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021263438,0.0011922744,0.0016408984,0.0011124134,0.0015264148,0.0033283336,0.0047540898,0.0028869677,0.006552006],"category_scores_gemma":[0.0055068503,0.001051113,0.0020807923,0.0013862527,0.0021012153,0.003412106,0.0024113539,0.0026240903,0.0010722982],"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":[0.000049858107,0.0000463546,0.00024168777,0.00008374807,0.000060116097,0.00013592007,0.00016359043,0.7622071,0.0010625519,0.22650278,0.0008203544,0.008625972],"study_design_scores_gemma":[0.000035878787,0.000017952238,0.000030021592,0.000012836997,0.000023380215,0.00002969337,0.000019952133,0.93933034,0.00037946162,0.055054646,0.0050508063,0.000015030213],"about_ca_topic_score_codex":0.012163465,"about_ca_topic_score_gemma":0.007037026,"teacher_disagreement_score":0.012163465,"about_ca_system_score_codex":0.0015857657,"about_ca_system_score_gemma":0.0022899508,"threshold_uncertainty_score":0.0241853},"labels":[],"label_agreement":null},{"id":"W1976355580","doi":"10.1007/s10115-011-0439-8","title":"Finding best evidence for evidence-based best practice recommendations in health care: the initial decision support system design","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Quality and Management","field":"Decision Sciences","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":"Best practice; Medical prescription; Clinical decision support system; Decision support system; Harm; Health care; Quality (philosophy); eHealth; Evidence-based practice; Medicine; Risk analysis (engineering); Computer science; Data mining; Nursing; Psychology; Alternative medicine","score_opus":0.539744659034669,"score_gpt":0.4976609918738763,"score_spread":0.04208366716079265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1976355580","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":[0.088109374,0.085846074,0.55686074,0.11464869,0.0044103116,0.13050453,0.0055312077,0.0012531367,0.01283595],"genre_scores_gemma":[0.116424814,0.005325227,0.843739,0.004323019,0.00037052284,0.028878618,0.0005901824,0.00005975192,0.00028892697],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.60416955,0.29728827,0.06020134,0.00910348,0.027743315,0.0014939447],"domain_scores_gemma":[0.2345123,0.70249665,0.017880669,0.012322744,0.02752743,0.0052602217],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.33023068,0.0038454267,0.013390765,0.020123579,0.004487221,0.02192273,0.0069785276,0.011780768,0.011327439],"category_scores_gemma":[0.69716316,0.00533757,0.009681214,0.009313162,0.0059784623,0.018105129,0.011655856,0.011781845,0.0013698804],"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":[0.016824756,0.003141283,0.018615624,0.13584907,0.01799012,0.00065180176,0.010554362,0.011384175,0.0020081229,0.05017519,0.010645437,0.7221601],"study_design_scores_gemma":[0.041944567,0.015705451,0.015860189,0.18688777,0.062956646,0.0010509902,0.0082261255,0.116269715,0.011989528,0.47708184,0.059985913,0.0020413029],"about_ca_topic_score_codex":0.0017420165,"about_ca_topic_score_gemma":0.0051016402,"teacher_disagreement_score":0.6697693,"about_ca_system_score_codex":0.013332491,"about_ca_system_score_gemma":0.039695367,"threshold_uncertainty_score":0.8259448},"labels":[],"label_agreement":null},{"id":"W1977437156","doi":"10.1007/s10115-010-0347-3","title":"Class separation through variance: a new application of outlier detection","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Anomaly Detection Techniques and Applications","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":"University of Alberta","funders":"","keywords":"Mahalanobis distance; Linear subspace; Outlier; Subspace topology; Anomaly detection; Pattern recognition (psychology); Computer science; Artificial intelligence; Ranking (information retrieval); Variance (accounting); Class (philosophy); Data mining; Mathematics; Algorithm","score_opus":0.008671805795515073,"score_gpt":0.26656406260817367,"score_spread":0.2578922568126586,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1977437156","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":[0.004708734,0.00017936366,0.99402857,0.000109996494,0.0000560613,0.000014709144,0.000034736964,0.0004702724,0.00039752017],"genre_scores_gemma":[0.24223998,0.00056952203,0.7540846,0.00016761875,0.0003792851,0.000075381125,0.0002923663,0.00042789395,0.0017633945],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9952657,0.0009418246,0.00027719527,0.0010575865,0.0022530775,0.00020469644],"domain_scores_gemma":[0.99363977,0.003260902,0.0007336725,0.000965668,0.0011930529,0.00020698603],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031147166,0.0010372145,0.0025657243,0.0039772345,0.000766555,0.0030997174,0.0026977935,0.0014979993,0.0011392859],"category_scores_gemma":[0.012719702,0.0005242459,0.0014711322,0.0047005,0.0016742256,0.0033174765,0.0028537551,0.0028271421,0.0005531117],"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":[0.00039299758,0.0002612394,0.010375918,0.00032923598,0.0005827174,0.00036982808,0.00053552666,0.058870398,0.024925314,0.06846079,0.0052601304,0.8296359],"study_design_scores_gemma":[0.000036917274,0.00013030757,0.0029047697,0.000032486612,0.00014254278,0.0007005958,0.00013272392,0.894026,0.011065549,0.0846141,0.0061259903,0.00008803331],"about_ca_topic_score_codex":0.0015561567,"about_ca_topic_score_gemma":0.0012117801,"teacher_disagreement_score":0.0039772345,"about_ca_system_score_codex":0.00056541234,"about_ca_system_score_gemma":0.0010345096,"threshold_uncertainty_score":0.0164724},"labels":[],"label_agreement":null},{"id":"W1981324119","doi":"10.1007/s10115-010-0292-1","title":"Architecturing large integrated complex information systems: an application to healthcare","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Electronic Health Records Systems","field":"Health Professions","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":"Université Laval","funders":"","keywords":"Computer science; Knowledge management; Enterprise architecture; Information system; Schema (genetic algorithms); Key (lock); Enterprise information system; Multitude; Data science; Enterprise integration; Health care; Architecture; Process management; Enterprise software; Business; Engineering; Computer security","score_opus":0.03302324811347194,"score_gpt":0.38800239655700186,"score_spread":0.3549791484435299,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981324119","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":[0.20403466,0.0010473034,0.7777877,0.0024261468,0.00012739103,0.00053291454,0.00022352087,0.004023068,0.009797285],"genre_scores_gemma":[0.27127007,0.0006417045,0.72460777,0.00012850808,0.000041094176,0.000091348964,0.00018132779,0.00017575226,0.0028624712],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99905485,0.00035092552,0.0000784618,0.00014692941,0.00029315107,0.00007570285],"domain_scores_gemma":[0.99634176,0.0021591089,0.00020396746,0.00061078795,0.00040919404,0.00027523623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017482684,0.00060678023,0.0004038471,0.00065836764,0.0015818683,0.003069135,0.00083646807,0.0014583444,0.0040066573],"category_scores_gemma":[0.0057016565,0.0005807528,0.0005118616,0.0014255613,0.0010137748,0.0020697096,0.0018948745,0.0009980472,0.00074600626],"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":[0.0004039096,0.00063797104,0.012847189,0.000864596,0.00009440077,0.0016917444,0.005166403,0.1535731,0.047500398,0.042798486,0.009691349,0.7247305],"study_design_scores_gemma":[0.00017969761,0.00040286937,0.005921331,0.00019361488,0.0001998935,0.0015807134,0.0038576,0.816568,0.03736756,0.07252799,0.061105438,0.00009525692],"about_ca_topic_score_codex":0.00593515,"about_ca_topic_score_gemma":0.0085820975,"teacher_disagreement_score":0.00593515,"about_ca_system_score_codex":0.0009163639,"about_ca_system_score_gemma":0.0020503425,"threshold_uncertainty_score":0.0134035945},"labels":[],"label_agreement":null},{"id":"W1981664281","doi":"10.1007/s10115-012-0593-7","title":"A multi-phase correlation search framework for mining non-taxonomic relations from unstructured text","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Canarie","keywords":"Computer science; Ontology; Taxonomy (biology); Taxonomic rank; Domain (mathematical analysis); Information retrieval; Sentence; Information extraction; Relationship extraction; Lift (data mining); Natural language processing; Artificial intelligence; Data science; Data mining; Mathematics; Ecology; Biology","score_opus":0.03944275429594141,"score_gpt":0.3070370740464318,"score_spread":0.2675943197504904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981664281","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":[0.009819525,0.00072693324,0.9836135,0.00023366373,0.000041611962,0.00032124692,0.0012552807,0.0031636918,0.0008246634],"genre_scores_gemma":[0.06261061,0.0003266489,0.9312177,0.00019297021,0.00006404371,0.0003810389,0.0036597932,0.00019492528,0.0013523207],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971107,0.0006346594,0.00035501894,0.00056578417,0.001127974,0.00020572585],"domain_scores_gemma":[0.9951218,0.0027132486,0.0003934574,0.0005991612,0.0009991045,0.00017327031],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003514435,0.0010085165,0.0020628122,0.00855108,0.0019128959,0.002805946,0.0024632835,0.0018050384,0.0030803548],"category_scores_gemma":[0.009757639,0.0007581919,0.0021808562,0.012005512,0.000897604,0.0035077077,0.0027048795,0.0014423309,0.0017563986],"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":[0.0009322442,0.0011424982,0.009630884,0.0012781677,0.00067051494,0.0010352241,0.00088280835,0.041592672,0.028432235,0.06503427,0.028374746,0.8209938],"study_design_scores_gemma":[0.00016731487,0.00030458864,0.004201883,0.00013277818,0.00039283893,0.0010056695,0.00037020832,0.8956404,0.011939,0.06779982,0.017927958,0.0001175883],"about_ca_topic_score_codex":0.010149661,"about_ca_topic_score_gemma":0.02068359,"teacher_disagreement_score":0.010149661,"about_ca_system_score_codex":0.0009666276,"about_ca_system_score_gemma":0.0053622047,"threshold_uncertainty_score":0.020181179},"labels":[],"label_agreement":null},{"id":"W1981679236","doi":"10.1007/s10115-004-0150-0","title":"Multiknowledge for decision making","year":2004,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Rough Sets and Fuzzy Logic","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":"Queen's University","funders":"","keywords":"Reduct; Rough set; Computer science; Data mining; Artificial intelligence; Decision system; Naive Bayes classifier; Machine learning; Classifier (UML); Decision table; Mathematics; Support vector machine","score_opus":0.01815329828864976,"score_gpt":0.2729560177640058,"score_spread":0.25480271947535604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981679236","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":[0.012644543,0.062495362,0.75285625,0.019281283,0.0020242394,0.000063792206,0.00033597686,0.00029507204,0.15000345],"genre_scores_gemma":[0.69587827,0.029246122,0.22450407,0.002345065,0.0031036593,0.00022801239,0.00044364916,0.00013844442,0.044112682],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9984471,0.0006389344,0.0001096843,0.00026595686,0.00046012862,0.00007817723],"domain_scores_gemma":[0.9967615,0.0022878048,0.00017421716,0.0004407364,0.00023792958,0.00009791485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002127454,0.0008582024,0.0015030467,0.002256387,0.0009997969,0.004874372,0.0011589343,0.0028839193,0.010189078],"category_scores_gemma":[0.0053971647,0.00043801032,0.0009819714,0.0024027776,0.0036620782,0.00931182,0.0022190697,0.0031480077,0.0017816086],"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":[0.000015133274,0.000014692004,0.000044817894,0.00014521468,0.00003327056,0.00006362894,0.00008805733,0.0026679626,0.000097460725,0.9655845,0.0023542712,0.028890854],"study_design_scores_gemma":[0.0000020165328,0.000002272999,0.000018536972,0.000029079656,0.0000064170345,0.00001523731,0.000011614574,0.0027864603,0.00004176172,0.99344796,0.0036354184,0.0000032268324],"about_ca_topic_score_codex":0.0011483206,"about_ca_topic_score_gemma":0.00086287456,"teacher_disagreement_score":0.010189078,"about_ca_system_score_codex":0.001781545,"about_ca_system_score_gemma":0.0008480193,"threshold_uncertainty_score":0.03408587},"labels":[],"label_agreement":null},{"id":"W1982481381","doi":"10.1007/pl00011667","title":"ActiveCBR: An Agent System That Integrates Case-Based Reasoning and Active Databases","year":2001,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"AI-based Problem Solving and Planning","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":"Simon Fraser University","funders":"","keywords":"Active database; Computer science; Case-based reasoning; Property (philosophy); Artificial intelligence; Reasoning system; Feature (linguistics); Database; Data mining; Event (particle physics); Expert system; Machine learning","score_opus":0.034879124724015026,"score_gpt":0.26716649458248526,"score_spread":0.23228736985847023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1982481381","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":[0.0090493355,0.0006323795,0.9290668,0.00066972454,0.00024255771,0.0007365783,0.0014419173,0.040750522,0.017410202],"genre_scores_gemma":[0.08987054,0.00069240574,0.89500254,0.00045060043,0.0001264048,0.000522193,0.00261786,0.0013321515,0.009385427],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985422,0.00034654987,0.00017363837,0.00025107988,0.0006247784,0.00006166417],"domain_scores_gemma":[0.99641657,0.0016885432,0.00027139962,0.0009108873,0.00042641078,0.00028613483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029995064,0.0009170312,0.0012361318,0.0029796753,0.0008518391,0.0047417665,0.0052657705,0.0020062574,0.014066636],"category_scores_gemma":[0.008110265,0.0012675951,0.0009784764,0.0015194505,0.00092023384,0.005488611,0.0032077986,0.0017887233,0.0048326124],"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":[0.0013440738,0.0016422658,0.0033897655,0.001400177,0.00045974157,0.0010926081,0.0012723785,0.023502672,0.014124413,0.10885958,0.068047605,0.77486473],"study_design_scores_gemma":[0.00077647704,0.00027157378,0.0011746848,0.0003418582,0.00073800463,0.0018759966,0.00026223136,0.5116697,0.028955393,0.1361356,0.31755865,0.00023984523],"about_ca_topic_score_codex":0.002520579,"about_ca_topic_score_gemma":0.0035258585,"teacher_disagreement_score":0.014066636,"about_ca_system_score_codex":0.0005850206,"about_ca_system_score_gemma":0.0015176444,"threshold_uncertainty_score":0.04705757},"labels":[],"label_agreement":null},{"id":"W1988047293","doi":"10.1007/s10115-013-0617-y","title":"Analyzing topics and authors in chat logs for crime investigation","year":2013,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Fonds Québécois de la Recherche sur la Nature et les Technologies","keywords":"Computer science; Exploit; Latent Dirichlet allocation; Online chat; Process (computing); The Internet; Order (exchange); Volume (thermodynamics); Channel (broadcasting); Topic model; Data science; World Wide Web; Information retrieval; Computer security","score_opus":0.026709074949374544,"score_gpt":0.2508178429459265,"score_spread":0.22410876799655197,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1988047293","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":[0.96769696,0.00082205096,0.021146316,0.0004883923,0.000097707205,0.00015521313,0.006482151,0.00092259096,0.002188618],"genre_scores_gemma":[0.9831352,0.00027924287,0.00941956,0.00003730364,0.000111577,0.00014878472,0.0055008344,0.00007692893,0.0012905842],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99664754,0.0015421531,0.00030119545,0.000579159,0.0006717679,0.00025814897],"domain_scores_gemma":[0.9528159,0.037120104,0.0033767126,0.0021410005,0.0031849442,0.0013613895],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033713803,0.00057065417,0.0006846928,0.00853906,0.0012344786,0.0022894735,0.0006488718,0.0010902622,0.0013745897],"category_scores_gemma":[0.025254125,0.0003115477,0.0007438794,0.0053070453,0.00043864144,0.0028348956,0.0010244919,0.0012138931,0.0012179287],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014259266,0.0011314289,0.78736377,0.00078687683,0.00047625555,0.00069402345,0.014805056,0.007936661,0.013096523,0.003106773,0.013949834,0.15522689],"study_design_scores_gemma":[0.00006701159,0.000642915,0.6484546,0.00025844324,0.00063135993,0.0017453057,0.015323787,0.29591417,0.012470772,0.008259748,0.016058283,0.00017356238],"about_ca_topic_score_codex":0.004581365,"about_ca_topic_score_gemma":0.008501317,"teacher_disagreement_score":0.00853906,"about_ca_system_score_codex":0.0007007926,"about_ca_system_score_gemma":0.0010414936,"threshold_uncertainty_score":0.017829716},"labels":[],"label_agreement":null},{"id":"W1991278297","doi":"10.1007/pl00011645","title":"Intentions in the Coordinated Generation of Graphics and Text from Tabular Data","year":2000,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":33,"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":"Natural Sciences and Engineering Research Council of Canada","keywords":"Graphics; Computer science; Wizard; Chart; Set (abstract data type); Table (database); Declaration; ASCII; Computer graphics (images); Information retrieval; Focus (optics); Plotter; Programming language; Data mining; World Wide Web","score_opus":0.05029146975346122,"score_gpt":0.2904555233044631,"score_spread":0.2401640535510019,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991278297","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":[0.08037866,0.00017387341,0.89287794,0.0011502265,0.00015523119,0.00029500548,0.00028688074,0.004877598,0.019804548],"genre_scores_gemma":[0.4967621,0.00020984588,0.49072486,0.0003865707,0.00004742288,0.00038351648,0.00086567,0.0021967227,0.00842331],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99380046,0.0030811154,0.00034910432,0.0008311779,0.0015341244,0.00040404708],"domain_scores_gemma":[0.98223025,0.010239744,0.0012799748,0.0025584495,0.0029436736,0.00074799306],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0068835295,0.00060613465,0.00052390125,0.0014798245,0.0013708834,0.007866529,0.0017772357,0.0019771163,0.007870173],"category_scores_gemma":[0.041820012,0.0013362144,0.0009829369,0.0015791053,0.002781905,0.0054711374,0.0046095042,0.0020992954,0.003291975],"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":[0.0025663127,0.0005049019,0.01872144,0.00086850737,0.00015534456,0.0011606843,0.038365938,0.02243179,0.06619407,0.39263967,0.02057676,0.43581453],"study_design_scores_gemma":[0.00037818437,0.0005618103,0.009250482,0.00038521993,0.00028699473,0.00047289548,0.014655186,0.51115555,0.07576867,0.33439437,0.052360937,0.00032972774],"about_ca_topic_score_codex":0.0055534393,"about_ca_topic_score_gemma":0.004345655,"teacher_disagreement_score":0.007870173,"about_ca_system_score_codex":0.0006529426,"about_ca_system_score_gemma":0.0018405004,"threshold_uncertainty_score":0.036404014},"labels":[],"label_agreement":null},{"id":"W1991959600","doi":"10.1007/s10115-008-0145-3","title":"Resolution-based outlier factor: detecting the top-n most outlying data points in engineering data","year":2008,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Anomaly Detection Techniques and Applications","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 Alberta","funders":"","keywords":"Outlier; Anomaly detection; Data mining; Computer science; Identification (biology); Local outlier factor; Rank (graph theory); Nonparametric statistics; Domain (mathematical analysis); Artificial intelligence; Mathematics; Statistics","score_opus":0.05467301865057345,"score_gpt":0.27022438014433875,"score_spread":0.2155513614937653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991959600","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":[0.17880774,0.00079688703,0.8163911,0.00030268394,0.000119519835,0.00015997609,0.0005144753,0.0022260686,0.0006815177],"genre_scores_gemma":[0.55517185,0.00028481174,0.4421778,0.000076379445,0.00011777805,0.000071229806,0.0013898924,0.00013315096,0.0005769672],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970118,0.00033141146,0.00027554535,0.00066008046,0.0014110983,0.00031010868],"domain_scores_gemma":[0.99532133,0.0017746709,0.00088811555,0.00056169694,0.0012326773,0.00022158065],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028867566,0.0011273442,0.0019967488,0.005049569,0.0010038768,0.0014584513,0.0022453205,0.0015496291,0.00072611443],"category_scores_gemma":[0.015083483,0.0003945634,0.0011281327,0.0041678906,0.00086502347,0.0019793164,0.0013331319,0.0013415228,0.00064596377],"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":[0.0014451351,0.0006498828,0.076681964,0.0004957546,0.0004980443,0.0009299132,0.00058036577,0.076391704,0.050550185,0.0040931753,0.007899131,0.7797847],"study_design_scores_gemma":[0.00007340296,0.00025969636,0.01967377,0.000036263296,0.00019852888,0.00113597,0.00030882793,0.9387363,0.031053115,0.005598107,0.0028541496,0.00007188108],"about_ca_topic_score_codex":0.0038116921,"about_ca_topic_score_gemma":0.0034512188,"teacher_disagreement_score":0.005049569,"about_ca_system_score_codex":0.00037168915,"about_ca_system_score_gemma":0.0011030681,"threshold_uncertainty_score":0.015266836},"labels":[],"label_agreement":null},{"id":"W2004295584","doi":"10.1007/s10115-008-0187-6","title":"The analysis and management of non-canonical requirement specifications through a belief integration game","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software Engineering 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 New Brunswick","funders":"","keywords":"Computer science; Viewpoints; Set (abstract data type); Analogy; Outcome (game theory); Theoretical computer science; Mathematics; Programming language; Mathematical economics","score_opus":0.029788926324930975,"score_gpt":0.29060839348361633,"score_spread":0.26081946715868537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004295584","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":[0.22430095,0.00007782927,0.76565534,0.0014215837,0.000021181051,0.0002615402,0.000073421834,0.00023454532,0.007953606],"genre_scores_gemma":[0.90457416,0.000049502978,0.093454234,0.000097444194,0.000014282265,0.00019855461,0.00006668956,0.00003682958,0.0015083352],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9913573,0.0051434836,0.00029535836,0.0006475792,0.0016348871,0.0009214281],"domain_scores_gemma":[0.93206185,0.059495714,0.0025234134,0.0022929043,0.0023197345,0.0013063777],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011062417,0.00082390016,0.0011503828,0.0009806326,0.0010308061,0.003613183,0.0026991912,0.0022679316,0.0030634021],"category_scores_gemma":[0.052659664,0.0009803039,0.0012003495,0.0008931826,0.0028925582,0.0062580234,0.00298561,0.003069863,0.00024357563],"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":[0.00094525353,0.0006934424,0.004187309,0.0002114531,0.00017279627,0.00061573135,0.0025677995,0.3663638,0.0054324814,0.57020986,0.001604236,0.046995807],"study_design_scores_gemma":[0.00007085908,0.000094876734,0.00028218297,0.000015569802,0.000030154568,0.000028352892,0.00017868895,0.8833995,0.00067769113,0.114797145,0.0004030083,0.000021914391],"about_ca_topic_score_codex":0.0072438554,"about_ca_topic_score_gemma":0.006178808,"teacher_disagreement_score":0.011062417,"about_ca_system_score_codex":0.0025221389,"about_ca_system_score_gemma":0.0037616647,"threshold_uncertainty_score":0.058504343},"labels":[],"label_agreement":null},{"id":"W2007800125","doi":"10.1007/s10115-014-0801-8","title":"Greedy column subset selection for large-scale data sets","year":2014,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Face and Expression Recognition","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":"University of Waterloo","funders":"","keywords":"Computer science; Column (typography); Selection (genetic algorithm); Greedy algorithm; Algorithm; Representation (politics); Matrix (chemical analysis); Big data; Data mining; Artificial intelligence; Frame (networking)","score_opus":0.02411245776584549,"score_gpt":0.2674305953687463,"score_spread":0.2433181376029008,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2007800125","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":[0.0808576,0.0021509107,0.90261775,0.0013024415,0.0002833099,0.0005031752,0.004837399,0.005789918,0.0016575218],"genre_scores_gemma":[0.3958808,0.0010549991,0.5665901,0.0008516598,0.0005967169,0.0010840751,0.028078448,0.0007914052,0.0050718975],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9984737,0.00052997866,0.00014495624,0.0003221947,0.00033842897,0.00019069118],"domain_scores_gemma":[0.9951853,0.0031156563,0.00018120202,0.00079551624,0.000499669,0.00022266559],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002045721,0.001984775,0.003167829,0.0026762646,0.0014283281,0.0019950543,0.0025383146,0.0014076756,0.004025536],"category_scores_gemma":[0.007130517,0.000928579,0.001990488,0.0038001572,0.0008914231,0.0017953662,0.0015932142,0.0016728888,0.0020937347],"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":[0.0014396462,0.00088574336,0.007488468,0.0007926346,0.00062313234,0.00054044224,0.00025984508,0.1773816,0.019912066,0.0040727686,0.05562621,0.73097754],"study_design_scores_gemma":[0.00016201472,0.00022839873,0.0019358188,0.000039039885,0.00015809211,0.00025021782,0.00024000739,0.9735112,0.006154989,0.013845702,0.0034401133,0.000034278488],"about_ca_topic_score_codex":0.0047838097,"about_ca_topic_score_gemma":0.010496132,"teacher_disagreement_score":0.0047838097,"about_ca_system_score_codex":0.0005566605,"about_ca_system_score_gemma":0.0024025445,"threshold_uncertainty_score":0.013466775},"labels":[],"label_agreement":null},{"id":"W2009423697","doi":"10.1007/s10115-013-0658-2","title":"Email mining: tasks, common techniques, and tools","year":2013,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Spam and Phishing Detection","field":"Computer Science","cited_by":73,"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; Categorization; World Wide Web; Data science; Visualization; Information retrieval; Data mining; Artificial intelligence","score_opus":0.015393168495726351,"score_gpt":0.22877223111903855,"score_spread":0.2133790626233122,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2009423697","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":[0.036558647,0.019234058,0.9237629,0.0049868966,0.0003188959,0.00079938676,0.0034493369,0.0051233307,0.005766587],"genre_scores_gemma":[0.17560709,0.014343234,0.7970798,0.00082355615,0.0009404365,0.0007357559,0.00601661,0.000609273,0.0038442276],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9870652,0.004455486,0.0018522117,0.0019369811,0.004121812,0.00056843477],"domain_scores_gemma":[0.97836375,0.012121004,0.001603125,0.0041753463,0.002947624,0.00078927015],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009927617,0.0023244892,0.002752721,0.014544052,0.0024901126,0.007769799,0.0036197228,0.0038494477,0.0017794042],"category_scores_gemma":[0.024344359,0.0011621359,0.0019580147,0.016276201,0.0017643102,0.010044206,0.004155178,0.0029004356,0.0025875228],"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":[0.00025063887,0.0005257511,0.012620452,0.0030899222,0.000250043,0.00019447236,0.0010495686,0.0040899743,0.0052542854,0.026115686,0.021674184,0.924885],"study_design_scores_gemma":[0.00013496041,0.00053370954,0.029436467,0.0026338864,0.00065974327,0.005985161,0.0047917925,0.24842225,0.04568275,0.47311547,0.18819574,0.00040815191],"about_ca_topic_score_codex":0.0011998226,"about_ca_topic_score_gemma":0.0016020988,"teacher_disagreement_score":0.014544052,"about_ca_system_score_codex":0.0011556827,"about_ca_system_score_gemma":0.0025046458,"threshold_uncertainty_score":0.05250287},"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":"W2014564114","doi":"10.1007/pl00011661","title":"An Artificial Network Simulating Cause-to-Effect Reasoning: Cancellation Interactions and Numerical Studies","year":2001,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Bayesian Modeling and Causal Inference","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":"Computer science; Probabilistic logic; Artificial neural network; Class (philosophy); Causal reasoning; Task (project management); Artificial intelligence; Variety (cybernetics); Causal model; Machine learning; Stability (learning theory); Mathematics; Cognition","score_opus":0.044046718684036715,"score_gpt":0.3379794475689115,"score_spread":0.2939327288848748,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2014564114","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":[0.541466,0.00042460806,0.44184,0.0018505316,0.00016077673,0.00011928683,0.0003316011,0.00035513754,0.013452077],"genre_scores_gemma":[0.93828833,0.00017415089,0.058905296,0.00010908649,0.00003198917,0.00009921636,0.0001307244,0.000046053818,0.0022152017],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990101,0.0006014711,0.00004088441,0.00016947379,0.00011507484,0.00006301193],"domain_scores_gemma":[0.9457414,0.05020622,0.0009770832,0.0011500431,0.0013329708,0.00059227634],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046567894,0.00041521087,0.00094243995,0.0015007469,0.0012012904,0.0017225491,0.0021020202,0.0023584205,0.0071939137],"category_scores_gemma":[0.045469936,0.000577633,0.0007127302,0.0015382372,0.002033153,0.004207257,0.001103302,0.0015354136,0.00028146405],"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":[0.00038058244,0.00015296083,0.0024269847,0.00009662421,0.00006624786,0.00013926179,0.0002792796,0.9168601,0.0006329984,0.068956554,0.00072362984,0.009284748],"study_design_scores_gemma":[0.000022104789,0.000014009461,0.00008930389,0.0000043174177,0.000013824515,0.0000108429595,0.000015984559,0.9813907,0.00011682409,0.018192926,0.00012339285,0.0000057706147],"about_ca_topic_score_codex":0.013027926,"about_ca_topic_score_gemma":0.011170097,"teacher_disagreement_score":0.013027926,"about_ca_system_score_codex":0.0018910015,"about_ca_system_score_gemma":0.0011137198,"threshold_uncertainty_score":0.025904238},"labels":[],"label_agreement":null},{"id":"W2016772176","doi":"10.1007/s10115-009-0279-y","title":"A knowledge encapsulation approach to ontology modularization","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Semantic Web and Ontologies","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 New Brunswick","funders":"","keywords":"Computer science; Ontology; Ontology components; Modular design; Modular programming; Encapsulation (networking); Process ontology; Upper ontology; Modularity (biology); Ontology-based data integration; Web Ontology Language; Formalism (music); Software engineering; Information retrieval; Suggested Upper Merged Ontology; Domain knowledge; Semantic Web; Programming language","score_opus":0.01550687693980566,"score_gpt":0.24548345380183273,"score_spread":0.22997657686202708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2016772176","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":[0.0031595528,0.00018632413,0.99075377,0.00078305986,0.000053355325,0.000049623144,0.000051642142,0.00039060987,0.0045721414],"genre_scores_gemma":[0.11818923,0.0006551289,0.8730823,0.00053741934,0.00020235538,0.00024893176,0.00044155674,0.00041074387,0.0062324433],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9971564,0.0010073582,0.00036408496,0.0004143823,0.0008292595,0.00022848631],"domain_scores_gemma":[0.9932111,0.0023108334,0.0003485373,0.002852504,0.0010706605,0.00020636016],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058851177,0.0005514007,0.00077273656,0.0023366816,0.0015676476,0.004504972,0.0025059304,0.0014787532,0.0028713702],"category_scores_gemma":[0.01115851,0.00097650243,0.0026906084,0.0024725385,0.0039464505,0.013560197,0.004677565,0.0038096847,0.0010757579],"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":[0.000019738343,0.00004062028,0.00023327034,0.00006898116,0.00003423815,0.00008753888,0.0008401469,0.0028609242,0.0013682792,0.95169413,0.0022881262,0.040464036],"study_design_scores_gemma":[0.000017560047,0.000019510055,0.00024544727,0.000118935975,0.00011347608,0.00019084416,0.00021327026,0.035172455,0.0036710107,0.92724144,0.03295921,0.000036701636],"about_ca_topic_score_codex":0.0027333328,"about_ca_topic_score_gemma":0.0026939847,"teacher_disagreement_score":0.0058851177,"about_ca_system_score_codex":0.0012353362,"about_ca_system_score_gemma":0.0020381634,"threshold_uncertainty_score":0.031123877},"labels":[],"label_agreement":null},{"id":"W2017217509","doi":"10.1007/s10115-011-0397-1","title":"Scalable clustering methods for the name disambiguation problem","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":23,"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":"University of Illinois at Urbana-Champaign; University of British Columbia; Advanced Digital Sciences Center; Pennsylvania State University; National Science Foundation","keywords":"Cluster analysis; Computer science; Scalability; Artificial intelligence; Natural language processing; Entity linking; Data mining; Database","score_opus":0.22575786155557567,"score_gpt":0.43212848195334813,"score_spread":0.20637062039777246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017217509","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":[0.009259845,0.0013362265,0.9852647,0.00061902404,0.00016001651,0.00009745931,0.00052251184,0.0016257556,0.0011143294],"genre_scores_gemma":[0.1552874,0.0009935036,0.83585846,0.00030721942,0.0005553812,0.0003063303,0.0030197152,0.00057162123,0.003100267],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99588865,0.0014173252,0.00032061944,0.0009394631,0.0011545283,0.00027932352],"domain_scores_gemma":[0.9902293,0.0051353113,0.000702271,0.0022480988,0.0013636671,0.0003213358],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044222646,0.0015482252,0.0029234984,0.0036842613,0.0020274103,0.0023698176,0.004323401,0.0022421256,0.003866598],"category_scores_gemma":[0.016044196,0.0010878262,0.0017050966,0.0064259195,0.0011911194,0.005085086,0.0042158505,0.0027134987,0.0014225547],"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":[0.0004587182,0.00037639972,0.0018704915,0.00048200856,0.00040749917,0.00012433493,0.00025798054,0.33779973,0.0045161443,0.043678764,0.02886019,0.5811678],"study_design_scores_gemma":[0.0000804997,0.00003216534,0.000441965,0.000020211266,0.000042631742,0.000047490797,0.000054595406,0.9029911,0.0008989739,0.092939384,0.0024284187,0.000022560978],"about_ca_topic_score_codex":0.010633742,"about_ca_topic_score_gemma":0.01596277,"teacher_disagreement_score":0.010633742,"about_ca_system_score_codex":0.0017282113,"about_ca_system_score_gemma":0.0035145045,"threshold_uncertainty_score":0.023387432},"labels":[],"label_agreement":null},{"id":"W2017760966","doi":"10.1007/s10115-009-0214-2","title":"Fuzzy clustering-based discretization for gene expression classification","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","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":"Artificial intelligence; Classifier (UML); Computer science; Data mining; Fuzzy logic; Machine learning; Cluster analysis; Fuzzy classification; Pattern recognition (psychology); Association rule learning; Margin classifier; Support vector machine; Fuzzy rule; Fuzzy set; Feature vector; Fuzzy clustering","score_opus":0.016106872941808476,"score_gpt":0.26921761253277804,"score_spread":0.2531107395909696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017760966","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":[0.016333776,0.00032795625,0.9815709,0.00007851292,0.00003842515,0.00004028143,0.00023571911,0.0003241039,0.0010503546],"genre_scores_gemma":[0.33632705,0.00034658238,0.6601575,0.000078923586,0.000045351488,0.00018676215,0.00095066876,0.00006365228,0.001843527],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993494,0.00016800074,0.00006220155,0.00011363098,0.00025296357,0.00005383573],"domain_scores_gemma":[0.9989666,0.00056519086,0.00004628855,0.00013080983,0.00026106744,0.000030054482],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010045164,0.00031069107,0.0011072689,0.0010393665,0.0005827962,0.00097165385,0.0010241645,0.00059947075,0.0013141125],"category_scores_gemma":[0.0024378768,0.00026346353,0.0008008861,0.0018688589,0.0004349424,0.0006884551,0.00044597386,0.00086043245,0.00037760462],"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":[0.00034610156,0.0001257129,0.0013925915,0.00021566515,0.00013572273,0.00008971366,0.00022880541,0.48248607,0.015706487,0.024445837,0.0033729079,0.47145435],"study_design_scores_gemma":[0.000009240508,0.000017299339,0.00042470417,0.000012148295,0.000014545403,0.000027048605,0.000019764077,0.9871191,0.0017922276,0.009736509,0.00081700203,0.000010441252],"about_ca_topic_score_codex":0.010410951,"about_ca_topic_score_gemma":0.008158467,"teacher_disagreement_score":0.010410951,"about_ca_system_score_codex":0.0011840257,"about_ca_system_score_gemma":0.00086013036,"threshold_uncertainty_score":0.020700753},"labels":[],"label_agreement":null},{"id":"W2019619214","doi":"10.1007/s10115-012-0551-4","title":"A new approach to radial basis function-based polynomial neural networks: analysis and design","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Neural Networks and Applications","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; Particle swarm optimization; Radial basis function; Polynomial; Computer science; Artificial neural network; Perceptron; Radial basis function network; Algorithm; Mathematical optimization; Artificial intelligence; Mathematics","score_opus":0.020905094039850285,"score_gpt":0.23030253814077742,"score_spread":0.20939744410092714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2019619214","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":[0.0003964149,0.00032530748,0.9985707,0.00007059059,0.000036054593,0.000010284401,0.00000816319,0.00003749871,0.0005450001],"genre_scores_gemma":[0.1547897,0.0034750053,0.83021224,0.0003331481,0.00055206113,0.00030619663,0.00014402432,0.00017065952,0.010017045],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989814,0.00035475442,0.000052078205,0.00016981593,0.00039708035,0.000044862416],"domain_scores_gemma":[0.99888295,0.0004203078,0.000073839015,0.00008716765,0.00049821736,0.000037488302],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018267618,0.0011082935,0.001170495,0.0007614949,0.00039034456,0.001341197,0.0021080321,0.0016772059,0.0022544242],"category_scores_gemma":[0.0045174137,0.0005596757,0.0009039208,0.0011234076,0.0008266636,0.0019468562,0.0011860743,0.0021968517,0.0011906315],"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":[0.00007994013,0.00009244097,0.0003886204,0.00035804068,0.00015194324,0.00009635101,0.00009498386,0.6218399,0.005836134,0.16031595,0.005056097,0.20568962],"study_design_scores_gemma":[0.0000035374585,0.0000143620155,0.000031850246,0.0000063996113,0.00001014987,0.000019622668,0.0000030267124,0.98915416,0.00034095914,0.008957814,0.001452392,0.0000056420927],"about_ca_topic_score_codex":0.0030727056,"about_ca_topic_score_gemma":0.0033863941,"teacher_disagreement_score":0.0030727056,"about_ca_system_score_codex":0.0008607121,"about_ca_system_score_gemma":0.0010230386,"threshold_uncertainty_score":0.009660959},"labels":[],"label_agreement":null},{"id":"W2019942346","doi":"10.1007/s10115-012-0485-x","title":"Impact of query intent and search context on clickthrough behavior in sponsored search","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Consumer Market Behavior and Pricing","field":"Business, Management and Accounting","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 Waterloo","funders":"Microsoft Research","keywords":"Computer science; Information retrieval; Web search query; Web query classification; Query expansion; Search engine; Context (archaeology); Rank (graph theory); Ranking (information retrieval); Query optimization; Sargable; Query language; World Wide Web; Mathematics","score_opus":0.04596005764990982,"score_gpt":0.3114538059106155,"score_spread":0.2654937482607057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2019942346","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":[0.9988961,0.00012912936,0.00008855688,0.000053667558,0.0000043003765,0.0000047174804,0.00008458275,0.000011121274,0.000727848],"genre_scores_gemma":[0.9993686,0.000037557355,0.00008553057,0.000024841776,0.000010733284,0.0000029654334,0.0001363464,0.000009471539,0.00032399799],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989133,0.00050701085,0.00010102678,0.00011592403,0.00021348202,0.00014932429],"domain_scores_gemma":[0.93015885,0.060584202,0.004073802,0.0010072291,0.0015330906,0.0026428944],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024264825,0.00022737005,0.00039434564,0.000793554,0.00033043174,0.0017328952,0.0002838831,0.00082382874,0.0030520628],"category_scores_gemma":[0.03803864,0.00024990996,0.0003579608,0.0006727166,0.00028083817,0.0011671804,0.0004464988,0.0008452478,0.00039490583],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008840705,0.002282625,0.94616616,0.00014663982,0.00031039154,0.00031878226,0.00066684134,0.003950827,0.014123481,0.0005626364,0.00084726093,0.02178359],"study_design_scores_gemma":[0.000099653356,0.0008399393,0.97608364,0.000023171347,0.00030651694,0.000298861,0.0003857964,0.019415358,0.0016094638,0.0005931138,0.00030886734,0.00003563104],"about_ca_topic_score_codex":0.006836972,"about_ca_topic_score_gemma":0.009680076,"teacher_disagreement_score":0.006836972,"about_ca_system_score_codex":0.0003582315,"about_ca_system_score_gemma":0.00056814036,"threshold_uncertainty_score":0.013594389},"labels":[],"label_agreement":null},{"id":"W2025568499","doi":"10.1007/s10115-012-0538-1","title":"Efficient greedy feature selection for unsupervised learning","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Face and Expression Recognition","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 Waterloo","funders":"","keywords":"Feature selection; Dimensionality reduction; Computer science; Artificial intelligence; Greedy algorithm; Machine learning; Pattern recognition (psychology); Curse of dimensionality; Feature (linguistics); Feature learning; Unsupervised learning; Selection (genetic algorithm); Dimension (graph theory); Data mining; Algorithm; Mathematics","score_opus":0.013086839058054958,"score_gpt":0.237677893728842,"score_spread":0.22459105467078702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2025568499","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":[0.008496701,0.00020465472,0.9889966,0.00012168762,0.000030693158,0.000059262165,0.00014756649,0.0015231131,0.00041965878],"genre_scores_gemma":[0.30557588,0.00027470718,0.68501997,0.0003004976,0.00013604044,0.0006072868,0.0023228233,0.00049567095,0.005267175],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985158,0.0004636955,0.00010968347,0.00032503688,0.00036920863,0.00021664763],"domain_scores_gemma":[0.9977502,0.0013555934,0.00011056121,0.00033380033,0.0003651651,0.00008458199],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015042552,0.0013780108,0.0025662028,0.0012827188,0.00096095103,0.0011310531,0.0028299855,0.0015080508,0.0037415845],"category_scores_gemma":[0.0049296725,0.000917985,0.0014215107,0.0018801402,0.001006509,0.0016120118,0.0018854521,0.0016341887,0.0018443915],"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":[0.00059214764,0.00030477735,0.0012952622,0.00016427822,0.00017772368,0.00016003654,0.00008150204,0.17676528,0.013483602,0.009667578,0.014639473,0.7826682],"study_design_scores_gemma":[0.000053623477,0.000055407407,0.0004023163,0.0000061302603,0.000025125702,0.000059661328,0.000018742681,0.98372525,0.002868221,0.011907968,0.0008634177,0.000014208739],"about_ca_topic_score_codex":0.007707199,"about_ca_topic_score_gemma":0.011249158,"teacher_disagreement_score":0.007707199,"about_ca_system_score_codex":0.0009909511,"about_ca_system_score_gemma":0.0024960414,"threshold_uncertainty_score":0.015324712},"labels":[],"label_agreement":null},{"id":"W2026962484","doi":"10.1007/s10115-006-0002-1","title":"A collaborative filtering framework based on fuzzy association rules and multiple-level similarity","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":114,"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":"National Research Council Canada; Hong Kong Polytechnic University; University of Rochester","keywords":"Collaborative filtering; Recommender system; Computer science; Association rule learning; Similarity (geometry); Popularity; Fuzzy logic; Data mining; Product (mathematics); Information retrieval; Quality (philosophy); The Internet; Machine learning; Artificial intelligence; World Wide Web; Mathematics","score_opus":0.014067022931369966,"score_gpt":0.24111992388868606,"score_spread":0.2270529009573161,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2026962484","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":[0.0021502236,0.00030929188,0.9964676,0.00010564072,0.000039417195,0.000035157012,0.00003302104,0.0001287119,0.0007309041],"genre_scores_gemma":[0.12251322,0.00069234846,0.87236357,0.00014450705,0.00021319598,0.000142239,0.00022629907,0.000046330293,0.0036583948],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9943494,0.0016319305,0.0004199911,0.0011580279,0.0022133547,0.0002273281],"domain_scores_gemma":[0.9936243,0.003177907,0.0003299385,0.00088147784,0.0017635653,0.00022282384],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057649156,0.0008993491,0.003073223,0.004016074,0.0019230848,0.0037117784,0.004292514,0.0031969931,0.0027525614],"category_scores_gemma":[0.011066842,0.0008810489,0.0024806703,0.0046489453,0.001254803,0.0059657497,0.0020964136,0.0021061,0.0012809181],"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":[0.00030130075,0.0005448804,0.0028006558,0.0005228191,0.0010220645,0.0004599663,0.000981222,0.22609922,0.0065538073,0.27603891,0.006257436,0.47841778],"study_design_scores_gemma":[0.000041965017,0.00011596431,0.00052344985,0.000044959077,0.00020509824,0.0003105053,0.00005683815,0.9217408,0.0015625464,0.06991953,0.0053945775,0.00008384184],"about_ca_topic_score_codex":0.011057874,"about_ca_topic_score_gemma":0.01103358,"teacher_disagreement_score":0.011057874,"about_ca_system_score_codex":0.0010959159,"about_ca_system_score_gemma":0.0017774277,"threshold_uncertainty_score":0.030488133},"labels":[],"label_agreement":null},{"id":"W2027417703","doi":"10.1007/s10115-010-0287-y","title":"An information gain-based approach for recommending useful product reviews","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":64,"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":"Helpfulness; Computer science; Product (mathematics); Recommender system; Ranking (information retrieval); Quality (philosophy); Order (exchange); Task (project management); Information retrieval; Data science; Business; Engineering","score_opus":0.03250743570292643,"score_gpt":0.2886427697291577,"score_spread":0.2561353340262313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2027417703","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":[0.14743298,0.006187667,0.8279379,0.002145537,0.0005627064,0.0010306265,0.0015171065,0.0025501705,0.010635312],"genre_scores_gemma":[0.7109576,0.0013721777,0.27876347,0.00040488646,0.0007892257,0.00039273754,0.001148431,0.0000721033,0.006099379],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974769,0.0004163119,0.00020977242,0.00033546373,0.0014462707,0.00011533023],"domain_scores_gemma":[0.99561876,0.0020369168,0.00024926657,0.00016206909,0.001839573,0.000093397255],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023840931,0.0011173775,0.0018986452,0.0070535787,0.0008148514,0.0017484548,0.0013703495,0.001715071,0.0019738325],"category_scores_gemma":[0.008329907,0.00043065706,0.0012555544,0.0037789212,0.00047093528,0.0025647946,0.00071798015,0.0010161225,0.0008293616],"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":[0.0011869675,0.0013395507,0.011914807,0.00068543013,0.00090004405,0.000337252,0.0002578143,0.027598675,0.018792374,0.005585945,0.015513563,0.9158876],"study_design_scores_gemma":[0.0001908849,0.0007892297,0.009662986,0.00007044531,0.00063302147,0.0005826137,0.0001069824,0.964583,0.009652946,0.008892618,0.0047327867,0.00010245032],"about_ca_topic_score_codex":0.003749022,"about_ca_topic_score_gemma":0.0074480623,"teacher_disagreement_score":0.0070535787,"about_ca_system_score_codex":0.0011012071,"about_ca_system_score_gemma":0.0010612807,"threshold_uncertainty_score":0.0126084685},"labels":[],"label_agreement":null},{"id":"W2028187934","doi":"10.1007/s10115-015-0835-6","title":"Efficient discovery of contrast subspaces for object explanation and characterization","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Image Retrieval and Classification Techniques","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":"Simon Fraser University","funders":"National Natural Science Foundation of China; China Postdoctoral Science Foundation; China Computer Federation","keywords":"Linear subspace; Contrast (vision); Pruning; Computer science; Object (grammar); Kernel (algebra); Range (aeronautics); Stability (learning theory); Set (abstract data type); Mathematics; Artificial intelligence; Characterization (materials science); Data mining; Pattern recognition (psychology); Algorithm; Machine learning; Combinatorics","score_opus":0.01964182712551702,"score_gpt":0.24757300972635435,"score_spread":0.22793118260083733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2028187934","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":[0.050902378,0.0002658675,0.94649124,0.00018320912,0.000012414834,0.0000540037,0.00019034646,0.0007641275,0.0011363802],"genre_scores_gemma":[0.4515618,0.00026356473,0.5455882,0.00009067671,0.000043796525,0.000073097,0.00089961535,0.00016313714,0.0013160979],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99956614,0.00009571621,0.0000243567,0.00010876376,0.00013451785,0.00007052584],"domain_scores_gemma":[0.99880385,0.0005791715,0.000098840035,0.00025222028,0.00020386168,0.000062026236],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006391823,0.0008089988,0.0011173004,0.0020433422,0.0004683372,0.0017062894,0.0012335217,0.001029363,0.0022919364],"category_scores_gemma":[0.0031794054,0.00042033455,0.0010222279,0.0012765008,0.0006539704,0.0022794984,0.0019369555,0.001308986,0.00072387315],"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":[0.0006024546,0.00028140118,0.005065963,0.00023437756,0.0001860349,0.00024065273,0.00027721553,0.041840974,0.09921486,0.034666106,0.0041260584,0.81326383],"study_design_scores_gemma":[0.000034033797,0.000093409646,0.001502617,0.000013434279,0.000053302403,0.00017262372,0.0001198313,0.9379897,0.023540739,0.034599926,0.0018616748,0.000018746698],"about_ca_topic_score_codex":0.0018975829,"about_ca_topic_score_gemma":0.0026728117,"teacher_disagreement_score":0.0022919364,"about_ca_system_score_codex":0.00047578718,"about_ca_system_score_gemma":0.00091424375,"threshold_uncertainty_score":0.007667303},"labels":[],"label_agreement":null},{"id":"W2028936054","doi":"10.1007/s101150050009","title":"An Index Structure for Data Mining and Clustering","year":2000,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Fudan University; National Science Foundation","keywords":"Cluster analysis; Euclidean distance; Data mining; Computer science; Metric (unit); Set (abstract data type); Index (typography); Visualization; Mathematics; Data set; Computation; Approximation error; Algorithm; Pattern recognition (psychology); Artificial intelligence","score_opus":0.02296360961186923,"score_gpt":0.2681822105133437,"score_spread":0.24521860090147446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2028936054","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":[0.005291731,0.0011830876,0.98423946,0.00067642325,0.00029762156,0.00020462487,0.0025564576,0.0026424008,0.0029081213],"genre_scores_gemma":[0.039620485,0.001027263,0.9479585,0.00028422847,0.00030749955,0.0003944876,0.0058113337,0.00044544888,0.0041507618],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99736553,0.00043529237,0.00044292634,0.00034273585,0.0012885011,0.00012506709],"domain_scores_gemma":[0.9931471,0.0016555558,0.00058902113,0.0023578648,0.0018611625,0.0003893021],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032931108,0.00071367493,0.0018991486,0.0051311282,0.0023170258,0.004012326,0.0021302,0.0012421245,0.004042132],"category_scores_gemma":[0.015369188,0.00067817234,0.0010799067,0.009693438,0.0010301134,0.007321326,0.0030643255,0.0020420484,0.0035299524],"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":[0.00038531568,0.00027610688,0.0028201106,0.00059304683,0.00011986406,0.00017529227,0.0005992815,0.014592313,0.007878007,0.19348873,0.05655429,0.72251767],"study_design_scores_gemma":[0.00012517223,0.00039381505,0.0020397166,0.00030502825,0.00019180533,0.00088165846,0.0002458033,0.2646913,0.010862662,0.5632091,0.15692483,0.00012900113],"about_ca_topic_score_codex":0.0021788576,"about_ca_topic_score_gemma":0.003615917,"teacher_disagreement_score":0.0051311282,"about_ca_system_score_codex":0.0012498916,"about_ca_system_score_gemma":0.0029214018,"threshold_uncertainty_score":0.017415822},"labels":[],"label_agreement":null},{"id":"W2029902517","doi":"10.1007/s10115-012-0511-z","title":"How you move reveals who you are: understanding human behavior by analyzing trajectory data","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":80,"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 New Brunswick","funders":"","keywords":"Computer science; Process (computing); Inference; Field (mathematics); Human behavior; Data science; Trajectory; Artificial intelligence; Knowledge extraction; Business process discovery; Human–computer interaction; Work in process; Business process; Engineering","score_opus":0.07373967419833824,"score_gpt":0.28297431501211756,"score_spread":0.20923464081377932,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2029902517","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":[0.94681036,0.00047268826,0.04586838,0.00088550855,0.000022156912,0.00006524283,0.003136613,0.00032401463,0.0024151192],"genre_scores_gemma":[0.9856839,0.00026797035,0.01249106,0.00003146369,0.000008575304,0.000023303712,0.0010178358,0.000027591046,0.00044822373],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999747,0.00010622821,0.00001665212,0.000062171544,0.000040008734,0.000027912596],"domain_scores_gemma":[0.9976271,0.0013067155,0.00043888175,0.00023524647,0.0002755587,0.00011662861],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004530261,0.00033594243,0.00027540172,0.0015895133,0.0004420284,0.0010800579,0.0003502307,0.0006866451,0.0010166672],"category_scores_gemma":[0.0058079944,0.00016068028,0.00019611775,0.0021658498,0.0004174779,0.001553057,0.00046165776,0.000502735,0.000554823],"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":[0.000555159,0.00026172635,0.74750894,0.00026146753,0.00022301769,0.00049785443,0.0116380025,0.020906523,0.008871657,0.0038475704,0.006051277,0.1993768],"study_design_scores_gemma":[0.000026504695,0.0003781372,0.5899519,0.0002358457,0.00021477029,0.000865067,0.034136042,0.32379425,0.0066437465,0.027251331,0.016366262,0.00013621301],"about_ca_topic_score_codex":0.024865828,"about_ca_topic_score_gemma":0.031651583,"teacher_disagreement_score":0.024865828,"about_ca_system_score_codex":0.0003945993,"about_ca_system_score_gemma":0.00059963303,"threshold_uncertainty_score":0.049442172},"labels":[],"label_agreement":null},{"id":"W2031366103","doi":"10.1007/s10115-002-8192-7","title":"Knowledge Discovery Through Self-Organizing Maps: Data Visualization and Query Processing","year":2002,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Visualization and Analytics","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 Toronto","funders":"","keywords":"Computer science; Knowledge extraction; Visualization; Data mining; Set (abstract data type); Information retrieval; Heuristic; Premise; Data visualization; Information visualization; Data science; Artificial intelligence","score_opus":0.047145981158898155,"score_gpt":0.3028134019006688,"score_spread":0.2556674207417706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2031366103","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":[0.038057067,0.0022844542,0.9440732,0.0017555653,0.00013291274,0.00017558725,0.0012659475,0.0068405094,0.0054145996],"genre_scores_gemma":[0.36565202,0.0023799306,0.627491,0.00015201312,0.00011618107,0.00024776292,0.0013246719,0.00055376533,0.002082688],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979317,0.0007632367,0.00012450363,0.00025328493,0.0008172108,0.000110082256],"domain_scores_gemma":[0.99514943,0.002845012,0.00027879298,0.0007516456,0.00077192165,0.00020321652],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032345979,0.00082841655,0.0009413622,0.00608657,0.0010462462,0.0065375087,0.0012654292,0.001090137,0.003126432],"category_scores_gemma":[0.011115626,0.00058675953,0.0008446951,0.007684327,0.0011124368,0.0057184114,0.0024045317,0.0010131617,0.00087731023],"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":[0.00057528884,0.00038448133,0.0078835655,0.0011331469,0.00032941677,0.0003587412,0.003068978,0.046867374,0.01023592,0.10343149,0.02637996,0.7993516],"study_design_scores_gemma":[0.00008294633,0.0000982123,0.003520109,0.00013569088,0.00019033068,0.00041542007,0.00184674,0.6693265,0.030770168,0.25319216,0.040296987,0.00012478596],"about_ca_topic_score_codex":0.0048545096,"about_ca_topic_score_gemma":0.004842831,"teacher_disagreement_score":0.0065375087,"about_ca_system_score_codex":0.0009627545,"about_ca_system_score_gemma":0.0018887635,"threshold_uncertainty_score":0.017106354},"labels":[],"label_agreement":null},{"id":"W2031582216","doi":"10.1007/s10115-006-0062-2","title":"Toward understanding the importance of gesture in distributed scientific collaboration","year":2007,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Speech and dialogue systems","field":"Computer Science","cited_by":8,"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 Victoria","funders":"University of Victoria","keywords":"Artifact (error); Computer science; Gesture; Coding (social sciences); Set (abstract data type); Data science; Human–computer interaction; Software; Artificial intelligence; Sociology","score_opus":0.03044051358999905,"score_gpt":0.25883442101330445,"score_spread":0.2283939074233054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2031582216","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":[0.23255555,0.003019229,0.685172,0.0066277604,0.00026272013,0.00016355314,0.00020507429,0.00048467694,0.07150945],"genre_scores_gemma":[0.92287296,0.0011453098,0.07325347,0.00022495691,0.00008817753,0.00009402881,0.000078422905,0.00008304465,0.002159575],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99874604,0.0006293973,0.000065685,0.0002396927,0.00017578085,0.00014336065],"domain_scores_gemma":[0.99266934,0.00529576,0.00056051905,0.00071864773,0.0005261295,0.00022957972],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002272963,0.0005658026,0.000493765,0.0014693093,0.0013361782,0.0072178645,0.0011566872,0.002828991,0.0044315387],"category_scores_gemma":[0.016766487,0.00088404136,0.0006011801,0.00096273574,0.0059775272,0.015038581,0.0035128535,0.0027153767,0.0006243352],"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":[0.0006055194,0.00015894017,0.015782453,0.0008797938,0.00008923442,0.00069087686,0.03536339,0.027427962,0.039318155,0.6991892,0.0020193008,0.17847523],"study_design_scores_gemma":[0.000073853145,0.00010192347,0.009462976,0.0002322192,0.00008842133,0.0003746116,0.013192292,0.1487848,0.008286304,0.80614734,0.013151901,0.00010324948],"about_ca_topic_score_codex":0.004519685,"about_ca_topic_score_gemma":0.0029493526,"teacher_disagreement_score":0.99866384,"about_ca_system_score_codex":0.0010913828,"about_ca_system_score_gemma":0.0014228745,"threshold_uncertainty_score":0.014824986},"labels":[],"label_agreement":null},{"id":"W2032469362","doi":"10.1007/s10115-006-0032-8","title":"CanTree: a canonical-order tree for incremental frequent-pattern mining","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":140,"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":"Computer science; Data mining; Tree (set theory); GSP Algorithm; Database transaction; Tree structure; A priori and a posteriori; Association rule learning; Apriori algorithm; Database; Mathematics; Algorithm; Binary tree","score_opus":0.01329467542214002,"score_gpt":0.24506175368626992,"score_spread":0.2317670782641299,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2032469362","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":[0.011099182,0.0011142149,0.9371065,0.00036089084,0.00024840143,0.00059865456,0.011422102,0.035313714,0.0027363952],"genre_scores_gemma":[0.049336765,0.00066888914,0.9275955,0.00015667427,0.00007818708,0.00037731088,0.017496742,0.0016080376,0.00268185],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9986927,0.00025012152,0.00016913483,0.00019097835,0.00059378066,0.00010325359],"domain_scores_gemma":[0.9958216,0.0017424977,0.0002051335,0.0009794168,0.0010076294,0.00024368105],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015159169,0.0010606641,0.0015075727,0.004260251,0.0010099462,0.002600781,0.002450834,0.0010646568,0.0086219935],"category_scores_gemma":[0.0110673215,0.00089736417,0.0014359833,0.0054732054,0.00052272185,0.003007544,0.0015351088,0.0016473422,0.0037425216],"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":[0.0011303263,0.00036172007,0.0038665838,0.0013799357,0.00027318104,0.0006747421,0.0004561777,0.0217341,0.008582065,0.028379953,0.101332664,0.8318286],"study_design_scores_gemma":[0.0006083248,0.00047464864,0.0024623964,0.00035971726,0.00038773808,0.0015824201,0.00033671464,0.7047969,0.015611367,0.12110339,0.15204732,0.00022902263],"about_ca_topic_score_codex":0.006871248,"about_ca_topic_score_gemma":0.016837912,"teacher_disagreement_score":0.0086219935,"about_ca_system_score_codex":0.0006007674,"about_ca_system_score_gemma":0.0028870625,"threshold_uncertainty_score":0.028843403},"labels":[],"label_agreement":null},{"id":"W2034485749","doi":"10.1007/s10115-008-0190-y","title":"Effectiveness of NAQ-tree as index structure for similarity search in high-dimensional metric space","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Image and Video Retrieval Techniques","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 Calgary","funders":"","keywords":"Nearest neighbor search; Metric space; Tree (set theory); Tree traversal; Mathematics; Computer science; Cover tree; Data mining; Metric (unit); Disjoint sets; Algorithm; Pattern recognition (psychology); Artificial intelligence; Combinatorics; Cluster analysis; Discrete mathematics","score_opus":0.009903275750799953,"score_gpt":0.28998444828426806,"score_spread":0.2800811725334681,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2034485749","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":[0.4494004,0.01656079,0.5051788,0.0019556147,0.00059779343,0.0004153872,0.002581719,0.0038239337,0.019485498],"genre_scores_gemma":[0.73936707,0.0023104292,0.25246146,0.00020792376,0.00019106595,0.00014235424,0.0022755465,0.00016955036,0.0028745984],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9955676,0.0015504105,0.00032982486,0.0003295595,0.0020094174,0.00021320677],"domain_scores_gemma":[0.9883754,0.006844673,0.00049199065,0.0015389458,0.002364268,0.0003846915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030522353,0.00040667737,0.00201349,0.0034154058,0.0010957374,0.00200328,0.001547606,0.0016207518,0.002852324],"category_scores_gemma":[0.020699851,0.00023790982,0.000461464,0.0051802336,0.00064959854,0.0059336084,0.0013739648,0.0005739086,0.0010568117],"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":[0.0028984004,0.0010011528,0.010167956,0.0011127396,0.00024380947,0.00024130497,0.0005847433,0.09116068,0.020419417,0.05022824,0.024920689,0.7970208],"study_design_scores_gemma":[0.0002299957,0.0007612457,0.002822898,0.000063429914,0.00012029435,0.0006471159,0.00029872768,0.9476881,0.007318309,0.033312827,0.006682797,0.000054334847],"about_ca_topic_score_codex":0.0074714487,"about_ca_topic_score_gemma":0.005127252,"teacher_disagreement_score":0.0074714487,"about_ca_system_score_codex":0.0010534985,"about_ca_system_score_gemma":0.0021628498,"threshold_uncertainty_score":0.016141951},"labels":[],"label_agreement":null},{"id":"W2038021410","doi":"10.1007/s10115-010-0290-3","title":"Study of select items in different data sources by grouping","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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":"Measure (data warehouse); Computer science; Set (abstract data type); Data mining; Database; Association rule learning; Information retrieval","score_opus":0.022332729292336054,"score_gpt":0.2703498052012024,"score_spread":0.24801707590886635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2038021410","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":[0.84137416,0.0017052009,0.14843868,0.0005946215,0.00011522637,0.00064393994,0.0027364485,0.00052870996,0.0038630483],"genre_scores_gemma":[0.8315214,0.000734617,0.16029349,0.00008731737,0.00011065163,0.00023629487,0.0046861856,0.00013968,0.0021904418],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958727,0.0010701382,0.00051373465,0.0008317692,0.0014405831,0.0002711444],"domain_scores_gemma":[0.9785499,0.013466986,0.0016823998,0.0022305965,0.0033948468,0.000675387],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037671386,0.00077274657,0.0012676222,0.009019513,0.0018565889,0.003934144,0.0015886705,0.0009962161,0.0028909636],"category_scores_gemma":[0.017470416,0.0003978365,0.0018204423,0.009930285,0.00077583635,0.0049205865,0.0012378431,0.0008558542,0.0010442153],"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":[0.0036901643,0.001196653,0.46611944,0.002171104,0.0017317445,0.0013789387,0.00539901,0.00937406,0.029235035,0.01199526,0.0054426165,0.46226606],"study_design_scores_gemma":[0.00063045777,0.003815452,0.44116873,0.0009087085,0.006551427,0.005321964,0.039979443,0.31372762,0.08118379,0.06303122,0.04326843,0.0004126662],"about_ca_topic_score_codex":0.0018961962,"about_ca_topic_score_gemma":0.002310748,"teacher_disagreement_score":0.009019513,"about_ca_system_score_codex":0.00062466227,"about_ca_system_score_gemma":0.0010848014,"threshold_uncertainty_score":0.019922733},"labels":[],"label_agreement":null},{"id":"W2043301550","doi":"10.1007/s10115-003-0101-1","title":"Data Mining: How Research Meets Practical Development?","year":2003,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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 Waterloo","funders":"","keywords":"Data science; Computer science; Data mining; Panel discussion; Development (topology); Business","score_opus":0.19245540659425892,"score_gpt":0.3901175826814103,"score_spread":0.19766217608715136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2043301550","genre_codex":"commentary","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":[0.0019608561,0.15114613,0.089480124,0.73471314,0.0057799355,0.00020557428,0.00030338272,0.00050999434,0.01590085],"genre_scores_gemma":[0.16706468,0.3342386,0.36971208,0.08712785,0.02937904,0.0015021887,0.00087290193,0.0007580621,0.009344655],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9474573,0.03334611,0.004461061,0.002660342,0.01101732,0.0010579112],"domain_scores_gemma":[0.55890167,0.33576322,0.010106628,0.029420555,0.054944377,0.010863642],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.13472348,0.003167832,0.0053270995,0.008756171,0.0037803324,0.02202194,0.007917255,0.015326348,0.010668081],"category_scores_gemma":[0.24473633,0.0023570633,0.0014619903,0.00889241,0.03702589,0.07013687,0.010910106,0.015485746,0.005318611],"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":[0.00019275506,0.00039409485,0.0048522945,0.004807663,0.0002354541,0.00022772372,0.0011621522,0.0015050026,0.00041009195,0.57316285,0.13382497,0.27922493],"study_design_scores_gemma":[0.00014041431,0.00012546808,0.00091501576,0.0040776827,0.000112712834,0.00034999702,0.0031747485,0.004298895,0.00045954718,0.8291125,0.1571601,0.00007293135],"about_ca_topic_score_codex":0.003501137,"about_ca_topic_score_gemma":0.0026669162,"teacher_disagreement_score":0.13472348,"about_ca_system_score_codex":0.0059383325,"about_ca_system_score_gemma":0.025028488,"threshold_uncertainty_score":0.7124944},"labels":[],"label_agreement":null},{"id":"W2044979776","doi":"10.1007/s10115-012-0515-8","title":"Anonymizing bag-valued sparse data by semantic similarity-based clustering","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":12,"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; Semantic similarity; Cluster analysis; Information retrieval; Vocabulary; Data publishing; WordNet; Similarity (geometry); Differential privacy; Data mining; Artificial intelligence","score_opus":0.05574438451177503,"score_gpt":0.28777105456625324,"score_spread":0.2320266700544782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2044979776","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":[0.018996915,0.00015086382,0.97905236,0.00035738188,0.0000616936,0.00004013472,0.0003006481,0.00031499716,0.00072502665],"genre_scores_gemma":[0.57664996,0.0006402652,0.41588235,0.00034966724,0.00032342388,0.00015808533,0.0032191072,0.00016860239,0.002608586],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99328,0.0025427428,0.0004704649,0.0010562297,0.00224929,0.0004013192],"domain_scores_gemma":[0.98709285,0.0044557485,0.00097486307,0.0060672946,0.0012168335,0.00019244546],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004118244,0.0006759642,0.002095867,0.0027618979,0.001356128,0.002887871,0.002258467,0.0015631536,0.0011342582],"category_scores_gemma":[0.020238038,0.0007095561,0.0013503279,0.004980617,0.002018085,0.0059613907,0.004650351,0.0021455388,0.00070203195],"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":[0.0011286306,0.00037154273,0.0039004334,0.00038843104,0.00041880313,0.00037350843,0.0011107846,0.39690575,0.014556402,0.21069705,0.009605803,0.3605428],"study_design_scores_gemma":[0.000032744192,0.00007598528,0.0005816198,0.000028486493,0.00005649558,0.00029273808,0.00030283988,0.7985351,0.0069391746,0.18982857,0.0032835002,0.000042681557],"about_ca_topic_score_codex":0.0014801438,"about_ca_topic_score_gemma":0.0016316102,"teacher_disagreement_score":0.004118244,"about_ca_system_score_codex":0.001056242,"about_ca_system_score_gemma":0.0017697462,"threshold_uncertainty_score":0.021779656},"labels":[],"label_agreement":null},{"id":"W2046943275","doi":"10.1007/s10115-012-0524-7","title":"Common understanding in a multi-agent system using ontology-guided learning","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Semantic Web and Ontologies","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; Western University","funders":"","keywords":"Computer science; Ontology; Flexibility (engineering); Set (abstract data type); Process (computing); Multi-agent system; Intelligent agent; Common knowledge (logic); Autonomous agent; Artificial intelligence; Human–computer interaction; Description logic; Epistemic modal logic","score_opus":0.13382357076986137,"score_gpt":0.31820717832592593,"score_spread":0.18438360755606456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2046943275","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":[0.069348805,0.00007775679,0.92222244,0.0008765716,0.000022739907,0.00017179322,0.00005630501,0.0006910654,0.0065326416],"genre_scores_gemma":[0.63246363,0.00008744317,0.36379972,0.00014324284,0.000015640066,0.00018508542,0.00019901547,0.00009318741,0.0030130087],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99797016,0.000742085,0.0002162585,0.00046462036,0.0004344403,0.0001723111],"domain_scores_gemma":[0.9964631,0.0019290365,0.00026948,0.0005921562,0.00050221646,0.00024399809],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034602997,0.00054487144,0.0010353413,0.0013350791,0.0026584957,0.0040749554,0.0025025345,0.0022272577,0.0029199854],"category_scores_gemma":[0.009144153,0.0005997502,0.0014415297,0.00091654423,0.002492952,0.009201942,0.0055638,0.0022174455,0.00041462985],"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":[0.00072711817,0.0012015565,0.00881041,0.0003370112,0.000411869,0.001279129,0.011010722,0.353501,0.010764312,0.35755903,0.0032648724,0.251133],"study_design_scores_gemma":[0.000054277905,0.000077271834,0.0005091573,0.00003964355,0.000101102625,0.000103472375,0.0008294621,0.8216052,0.0038112162,0.16892043,0.003903798,0.000044999208],"about_ca_topic_score_codex":0.012789729,"about_ca_topic_score_gemma":0.01481305,"teacher_disagreement_score":0.012789729,"about_ca_system_score_codex":0.0016936281,"about_ca_system_score_gemma":0.0032345455,"threshold_uncertainty_score":0.02543056},"labels":[],"label_agreement":null},{"id":"W2049619678","doi":"10.1007/s10115-012-0486-9","title":"Skyline distance: a measure of multidimensional competence","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","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":"Simon Fraser University","funders":"","keywords":"Skyline; Computer science; Scalability; Measure (data warehouse); Data mining; sort; Point (geometry); Theoretical computer science; Information retrieval; Database; Mathematics","score_opus":0.01418241925939989,"score_gpt":0.23315883989896102,"score_spread":0.21897642063956113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2049619678","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":[0.23811528,0.001372171,0.73383844,0.0010351879,0.00011741888,0.0001653607,0.0027953738,0.0005358832,0.022024866],"genre_scores_gemma":[0.78090084,0.00067591085,0.2120197,0.00013623708,0.000121369776,0.00024041942,0.0027505392,0.00010289111,0.0030520698],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959984,0.0012208062,0.0004407036,0.0006641467,0.0014519225,0.00022396885],"domain_scores_gemma":[0.9871395,0.0056096343,0.002072598,0.0014240608,0.002633033,0.0011211337],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025956044,0.0005334052,0.0006600671,0.006403836,0.0009108819,0.0029777396,0.00092489395,0.0009894044,0.004071974],"category_scores_gemma":[0.023727791,0.00021102285,0.00078756985,0.005286068,0.0015551207,0.006757742,0.0028622241,0.001041017,0.00087935687],"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":[0.00084296527,0.00031131605,0.12952425,0.0010448877,0.0006524926,0.0003969467,0.0059788874,0.035674486,0.014756851,0.4361924,0.013689452,0.36093503],"study_design_scores_gemma":[0.00011455277,0.0008391589,0.1518329,0.0003528309,0.0001933929,0.0016026156,0.00529543,0.19074716,0.010449226,0.58685046,0.05140413,0.00031824858],"about_ca_topic_score_codex":0.001711581,"about_ca_topic_score_gemma":0.0014341426,"teacher_disagreement_score":0.006403836,"about_ca_system_score_codex":0.00094522047,"about_ca_system_score_gemma":0.0009461846,"threshold_uncertainty_score":0.013727009},"labels":[],"label_agreement":null},{"id":"W2050175694","doi":"10.1007/s10115-012-0527-4","title":"A new approach for maximizing bichromatic reverse nearest neighbor search","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","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":"Simon Fraser University","funders":"National Natural Science Foundation of China; National Science Foundation","keywords":"k-nearest neighbors algorithm; Best bin first; Nearest neighbor search; Large margin nearest neighbor; Nearest-neighbor chain algorithm; Computer science; Nearest neighbor graph; Metric space; Metric (unit); Fixed-radius near neighbors; Algorithm; Space (punctuation); Mathematics; Data mining; Artificial intelligence; Discrete mathematics; Cluster analysis","score_opus":0.036402218891202674,"score_gpt":0.26517865462145235,"score_spread":0.22877643573024967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2050175694","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":[0.002720377,0.00021653231,0.9951918,0.00008013338,0.00004214847,0.000037217964,0.00003434721,0.00019669217,0.0014807645],"genre_scores_gemma":[0.07147072,0.00029010812,0.9240741,0.00016719259,0.00009664541,0.00014150774,0.00018216851,0.00020523385,0.0033722285],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969015,0.0010045654,0.00017591832,0.00047865388,0.0012548921,0.00018449342],"domain_scores_gemma":[0.99732155,0.0008063291,0.00017549533,0.0006057321,0.0009966777,0.00009427691],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029786457,0.0011767829,0.0018417457,0.0020901016,0.00097108324,0.001483731,0.0037042196,0.0017188698,0.0035898944],"category_scores_gemma":[0.009467943,0.0007315173,0.0008486923,0.0026133286,0.000979381,0.0030607705,0.0031545102,0.0016334779,0.0013955488],"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":[0.00040424487,0.00027632262,0.0015816799,0.00041394337,0.00014999221,0.00011609406,0.00023405996,0.2356356,0.01911618,0.13192321,0.011843917,0.5983048],"study_design_scores_gemma":[0.000025938487,0.00006748514,0.00027673866,0.000023451084,0.000035418427,0.00016924457,0.000033998607,0.9572448,0.004143518,0.032692596,0.005260457,0.000026461983],"about_ca_topic_score_codex":0.0028101932,"about_ca_topic_score_gemma":0.0055102487,"teacher_disagreement_score":0.0037042196,"about_ca_system_score_codex":0.0009600619,"about_ca_system_score_gemma":0.0015879438,"threshold_uncertainty_score":0.015752792},"labels":[],"label_agreement":null},{"id":"W2054521721","doi":"10.1007/s10115-009-0245-8","title":"Mining incomplete survey data through classification","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":49,"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","keywords":"Missing data; Data mining; Computer science; Classifier (UML); Complete information; Data set; Artificial intelligence; Machine learning; Mathematics","score_opus":0.11564483165384673,"score_gpt":0.3234492842129648,"score_spread":0.20780445255911809,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2054521721","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":[0.09135821,0.0017616424,0.8978154,0.0010837483,0.00014716004,0.00035763625,0.0025089746,0.0017212934,0.0032458869],"genre_scores_gemma":[0.6101188,0.0013985249,0.3744561,0.00033937258,0.0003208993,0.00051203865,0.008116314,0.000113889335,0.0046240995],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99648297,0.001098482,0.00031770763,0.00076330046,0.0009849096,0.00035260207],"domain_scores_gemma":[0.9875729,0.006105102,0.0012923805,0.0028637545,0.0019149428,0.00025085662],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034401584,0.00078998646,0.0023563101,0.0070870807,0.0010835428,0.0033421586,0.0022766364,0.0014000216,0.002261367],"category_scores_gemma":[0.014866972,0.00058085687,0.0017556555,0.0089424625,0.00090636325,0.0038804745,0.0019622853,0.0014188726,0.0018280691],"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":[0.00031276676,0.00042907422,0.08690983,0.00046543285,0.00032281585,0.00023602581,0.0005261197,0.025537256,0.002790309,0.014353046,0.012782644,0.85533464],"study_design_scores_gemma":[0.000051053426,0.00015547275,0.019215753,0.00019628236,0.00034646803,0.0005779616,0.0009903309,0.871932,0.0068319025,0.08171186,0.017929606,0.00006118715],"about_ca_topic_score_codex":0.005739502,"about_ca_topic_score_gemma":0.007566735,"teacher_disagreement_score":0.0070870807,"about_ca_system_score_codex":0.001003916,"about_ca_system_score_gemma":0.0018381062,"threshold_uncertainty_score":0.018193483},"labels":[],"label_agreement":null},{"id":"W2054765427","doi":"10.1007/s10115-011-0467-4","title":"A countably infinite mixture model for clustering and feature selection","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":52,"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; Concordia University","funders":"","keywords":"Mixture model; Cluster analysis; Model selection; Computer science; Artificial intelligence; Feature selection; Dirichlet process; Dirichlet distribution; Machine learning; Bayesian inference; Inference; Pattern recognition (psychology); Data mining; Mathematics; Bayesian probability","score_opus":0.028644784435366628,"score_gpt":0.25466713485173914,"score_spread":0.2260223504163725,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2054765427","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":[0.00087694783,0.00031461546,0.9981573,0.00016858727,0.000016813858,0.000014894277,0.000059991606,0.000068135174,0.00032270845],"genre_scores_gemma":[0.12083402,0.001724557,0.86607176,0.000483067,0.00036788057,0.0007573895,0.0012233729,0.0002855387,0.008252461],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9933289,0.0035470892,0.0003676968,0.0011716016,0.0012797841,0.0003049647],"domain_scores_gemma":[0.9853056,0.011117771,0.00062260684,0.0013423066,0.0012959626,0.0003157322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008247054,0.0015241983,0.0038564277,0.0037413675,0.0020021915,0.004424692,0.008513694,0.004212843,0.005240204],"category_scores_gemma":[0.029752756,0.002046646,0.0034545395,0.006218964,0.003559905,0.007841675,0.003938571,0.0056276633,0.0022220693],"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":[0.00015101403,0.00009760749,0.0006613346,0.00027148923,0.00024306495,0.00011793511,0.00029212362,0.274262,0.0008596445,0.6372823,0.004785927,0.0809755],"study_design_scores_gemma":[0.000014831973,0.000011973739,0.00012036118,0.000026491534,0.000031482014,0.000045375273,0.000013755323,0.71681625,0.00016309673,0.28111094,0.0016122634,0.000033139913],"about_ca_topic_score_codex":0.0095969755,"about_ca_topic_score_gemma":0.009844458,"teacher_disagreement_score":0.0095969755,"about_ca_system_score_codex":0.0030330701,"about_ca_system_score_gemma":0.0020810054,"threshold_uncertainty_score":0.043615103},"labels":[],"label_agreement":null},{"id":"W2057822545","doi":"10.1007/s10115-014-0736-0","title":"Finding top- $$k\\, r$$ k r -cliques for keyword search from graphs in polynomial delay","year":2014,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Algorithms and Data Compression","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":"York University","funders":"","keywords":"Computer science; Clique; Substructure; Keyword search; Graph; Set (abstract data type); Theoretical computer science; Context (archaeology); Combinatorics; Mathematics; Information retrieval","score_opus":0.017632627432734232,"score_gpt":0.2689965599923172,"score_spread":0.25136393255958295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2057822545","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":[0.3030703,0.0042010914,0.6256744,0.012013,0.00047562012,0.0025454666,0.01606083,0.014499541,0.021459708],"genre_scores_gemma":[0.46521515,0.00090595306,0.5063624,0.0011398124,0.0002396568,0.00062338676,0.011656096,0.0017585572,0.012099048],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99739265,0.0005571557,0.00020288327,0.0009305328,0.00045523662,0.0004615442],"domain_scores_gemma":[0.9879848,0.008643326,0.0006473311,0.0014206932,0.00068536116,0.0006185755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015200974,0.0018416169,0.003239273,0.0030536463,0.0022958925,0.004244952,0.0047641336,0.0032835214,0.014030264],"category_scores_gemma":[0.015874432,0.0015919821,0.0027503292,0.0047224932,0.0017158565,0.008203542,0.0040204558,0.0022334335,0.00451434],"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":[0.00497056,0.0020197614,0.007929267,0.0038383019,0.0007823089,0.000834398,0.0018264509,0.24057421,0.053806897,0.06317663,0.11686241,0.50337875],"study_design_scores_gemma":[0.0005421519,0.0002175903,0.0013537864,0.0001012722,0.00021854379,0.00047745873,0.00073710654,0.84930515,0.011246721,0.12980735,0.0059118806,0.000080921076],"about_ca_topic_score_codex":0.013818193,"about_ca_topic_score_gemma":0.035113636,"teacher_disagreement_score":0.014030264,"about_ca_system_score_codex":0.0034646716,"about_ca_system_score_gemma":0.006392735,"threshold_uncertainty_score":0.046935976},"labels":[],"label_agreement":null},{"id":"W2064848265","doi":"10.1007/s10115-011-0405-5","title":"Effectiveness of NAQ-tree in handling reverse nearest-neighbor queries in high-dimensional metric space","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":4,"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; Pruning; Metric (unit); k-nearest neighbors algorithm; Nearest neighbor search; Euclidean distance; Tree (set theory); Data mining; Metric space; Artificial intelligence; Mathematics","score_opus":0.014143569775840109,"score_gpt":0.2202502816969014,"score_spread":0.20610671192106128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064848265","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":[0.52379537,0.009431058,0.43800312,0.002806716,0.0006923896,0.00038491687,0.0018878042,0.0073771654,0.015621519],"genre_scores_gemma":[0.7299175,0.0013074761,0.2626583,0.0003818622,0.00014036919,0.000092864495,0.0026127098,0.00026144233,0.0026274845],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99293035,0.0021225165,0.00073616067,0.00065057183,0.0030873984,0.00047313637],"domain_scores_gemma":[0.9753798,0.014516188,0.0008368762,0.004366026,0.0043755793,0.0005255104],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0055373223,0.0006349573,0.0022266654,0.001962444,0.0016602128,0.00257624,0.0025960577,0.0021616379,0.002400784],"category_scores_gemma":[0.031020323,0.00036298696,0.00066382607,0.0035391876,0.0008575823,0.007635797,0.0018928626,0.00084278936,0.0012945425],"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":[0.0046579763,0.0014436172,0.020762302,0.0010833781,0.0003972519,0.0004623795,0.00095734646,0.2078406,0.02084329,0.025486974,0.038282868,0.67778206],"study_design_scores_gemma":[0.00013487617,0.0002423219,0.0013729981,0.00002725851,0.000069594294,0.0004979912,0.00040331788,0.97652435,0.0054776454,0.011631745,0.0035856103,0.00003230385],"about_ca_topic_score_codex":0.012987172,"about_ca_topic_score_gemma":0.009305545,"teacher_disagreement_score":0.012987172,"about_ca_system_score_codex":0.001244695,"about_ca_system_score_gemma":0.002816612,"threshold_uncertainty_score":0.029284477},"labels":[],"label_agreement":null},{"id":"W2073825221","doi":"10.1007/s10115-007-0090-6","title":"Robust projected clustering","year":2007,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Statistical Methods and Models","field":"Mathematics","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":"Simon Fraser University; University of Alberta","funders":"Yale University","keywords":"Cluster analysis; Categorical variable; Disjoint sets; Linear subspace; Outlier; Data mining; CURE data clustering algorithm; Cluster (spacecraft); Subspace topology; Set (abstract data type); Computer science; Data set; Correlation clustering; Clustering high-dimensional data; Single-linkage clustering; Algorithm; Pattern recognition (psychology); Mathematics; Artificial intelligence; Machine learning; Combinatorics","score_opus":0.1299341584742713,"score_gpt":0.38931893886732927,"score_spread":0.25938478039305796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2073825221","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":[0.003052884,0.00036234193,0.9921829,0.00019975776,0.000099300625,0.00007035005,0.00038873038,0.0010685184,0.00257517],"genre_scores_gemma":[0.16516872,0.00076036784,0.8101547,0.00039574708,0.0003207678,0.00041890342,0.0055272905,0.0012083241,0.016045222],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9941738,0.0017918171,0.00027556162,0.0019467176,0.0015202488,0.00029183627],"domain_scores_gemma":[0.9950842,0.0010822621,0.00028152333,0.0018735202,0.0014966093,0.00018198106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030048057,0.0021832567,0.003138893,0.0031605107,0.0016673204,0.0035241086,0.0035245656,0.0031038253,0.009476632],"category_scores_gemma":[0.012162055,0.0013013256,0.0034270706,0.002928799,0.0017529648,0.0030905383,0.004389039,0.0027128276,0.0076417583],"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":[0.0005996759,0.00015752632,0.00088168593,0.000494365,0.00057615736,0.0001310083,0.00018217172,0.30061513,0.010288724,0.1031293,0.029212719,0.55373156],"study_design_scores_gemma":[0.000025108928,0.000049197428,0.0004786802,0.00003260446,0.00005059741,0.000108506414,0.000045135806,0.9114035,0.00388268,0.07660711,0.007279193,0.00003754938],"about_ca_topic_score_codex":0.004153752,"about_ca_topic_score_gemma":0.0043353103,"teacher_disagreement_score":0.009476632,"about_ca_system_score_codex":0.0013073313,"about_ca_system_score_gemma":0.002715041,"threshold_uncertainty_score":0.03170252},"labels":[],"label_agreement":null},{"id":"W2077179394","doi":"10.1007/s10115-006-0035-5","title":"Handicapping attacker's confidence: an alternative to k-anonymization","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","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":"Simon Fraser University","funders":"","keywords":"Property (philosophy); Computer science; Data mining; Limit (mathematics); Domain (mathematical analysis); Set (abstract data type); Monotonic function; Limiting; Dual (grammatical number); Data set; Information sensitivity; Transformation (genetics); Information retrieval; Artificial intelligence; Computer security; Mathematics","score_opus":0.021496679816238064,"score_gpt":0.27713976483394576,"score_spread":0.2556430850177077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077179394","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":[0.041915014,0.0003326784,0.93600136,0.0027635312,0.00019323351,0.00014402316,0.0001366234,0.00083074433,0.017682793],"genre_scores_gemma":[0.9169798,0.00016561133,0.07635539,0.00046409565,0.00015135808,0.000064090265,0.000053039235,0.000085610256,0.0056809266],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.988303,0.005256792,0.0005436804,0.0015979327,0.0031099247,0.0011886498],"domain_scores_gemma":[0.9559052,0.015663989,0.002451063,0.023205927,0.0021571512,0.00061672856],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0053049913,0.00081388175,0.0020871905,0.0013350263,0.0027489564,0.003694285,0.0028306933,0.0044995565,0.004015748],"category_scores_gemma":[0.029955966,0.00063709344,0.001327636,0.0019751347,0.0054033524,0.011069999,0.009334718,0.0039766207,0.0011821475],"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":[0.0010904084,0.00015624466,0.0015651437,0.00018531222,0.00020305591,0.0008306211,0.0018957476,0.041511558,0.00950708,0.7664293,0.0060791187,0.17054641],"study_design_scores_gemma":[0.000109487926,0.0001694623,0.0004194932,0.00007708288,0.00014492459,0.0015542795,0.0005524227,0.26311564,0.017220257,0.7039043,0.01260009,0.00013251501],"about_ca_topic_score_codex":0.000624957,"about_ca_topic_score_gemma":0.0005172882,"teacher_disagreement_score":0.0053049913,"about_ca_system_score_codex":0.0008499459,"about_ca_system_score_gemma":0.0014520441,"threshold_uncertainty_score":0.028055787},"labels":[],"label_agreement":null},{"id":"W2077378918","doi":"10.1007/s10115-009-0252-9","title":"A binary decision diagram based approach for mining frequent subsequences","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":48,"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; Bottleneck; Data mining; Prefix; Sequence (biology); Binary decision diagram; Representation (politics); Theoretical computer science","score_opus":0.022212111231806726,"score_gpt":0.2688086613249732,"score_spread":0.2465965500931665,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077378918","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":[0.008779426,0.0002639309,0.9865739,0.00019838371,0.0000783022,0.00040622393,0.0011407597,0.0012905154,0.0012685151],"genre_scores_gemma":[0.087221764,0.00028271854,0.9078753,0.00010793413,0.000041068448,0.0005440522,0.002225805,0.00006854393,0.00163277],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99668014,0.0008990314,0.0005351149,0.00063592027,0.0011263686,0.00012346607],"domain_scores_gemma":[0.9915455,0.005728547,0.00046825517,0.0003976678,0.0016492457,0.00021071202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028497824,0.0007054725,0.0011481505,0.004950769,0.0009336949,0.002582649,0.0013437119,0.0012767226,0.005750876],"category_scores_gemma":[0.01289001,0.0005311655,0.0011711105,0.0038055985,0.0006068251,0.0027355107,0.0009251632,0.0010808078,0.0015777234],"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":[0.0016688128,0.00079754245,0.016397556,0.0023484125,0.00040133003,0.0010882638,0.00070283335,0.057766594,0.02440337,0.08915106,0.011152012,0.7941223],"study_design_scores_gemma":[0.00032757668,0.000701722,0.003866563,0.00033945646,0.0004982288,0.0015005086,0.00029996096,0.84523046,0.021229744,0.089700565,0.03615619,0.00014906634],"about_ca_topic_score_codex":0.0037951896,"about_ca_topic_score_gemma":0.0041584945,"teacher_disagreement_score":0.005750876,"about_ca_system_score_codex":0.00079399126,"about_ca_system_score_gemma":0.0019808684,"threshold_uncertainty_score":0.019238532},"labels":[],"label_agreement":null},{"id":"W2078765398","doi":"10.1007/s10115-011-0400-x","title":"Early classification on time series","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":123,"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; Beijing Institute of Technology; University of Windsor; Simon Fraser University; National Science Foundation","keywords":"Series (stratigraphy); Classifier (UML); Time series; Computer science; k-nearest neighbors algorithm; Benchmark (surveying); Data mining; Artificial intelligence; Pattern recognition (psychology); Machine learning; Geography","score_opus":0.02641227350953722,"score_gpt":0.20786184290149232,"score_spread":0.18144956939195508,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2078765398","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":[0.148766,0.011669989,0.81064945,0.0040441537,0.0024292741,0.00011327434,0.00069308595,0.0010104437,0.020624338],"genre_scores_gemma":[0.8127771,0.00822032,0.12590587,0.00060205813,0.0042984523,0.00012722539,0.0021471868,0.00025430293,0.045667518],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9985178,0.00038075104,0.00011919754,0.00028651874,0.00054247136,0.00015333362],"domain_scores_gemma":[0.9886628,0.005780545,0.00069206726,0.0016579822,0.0028515195,0.00035502575],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002923914,0.00078602455,0.0009662483,0.004905713,0.00083443365,0.0030554074,0.0009911422,0.0012035329,0.004669551],"category_scores_gemma":[0.015084368,0.00037367485,0.00079321954,0.003334575,0.0010481911,0.004689715,0.0011799252,0.0024974856,0.0023160768],"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":[0.00054375525,0.00018839134,0.013474877,0.00047870525,0.000117680866,0.00026213177,0.00037356984,0.027529607,0.009754282,0.1895827,0.01782446,0.73986983],"study_design_scores_gemma":[0.000035085668,0.00024124759,0.012671009,0.00020893315,0.00012452183,0.00035985725,0.00022921251,0.5777108,0.013330394,0.35992473,0.03509586,0.000068380206],"about_ca_topic_score_codex":0.0017558711,"about_ca_topic_score_gemma":0.0012794434,"teacher_disagreement_score":0.004905713,"about_ca_system_score_codex":0.0013754914,"about_ca_system_score_gemma":0.0006484394,"threshold_uncertainty_score":0.015621185},"labels":[],"label_agreement":null},{"id":"W2081471536","doi":"10.1007/s10115-007-0097-z","title":"A compact multi-resolution index for variable length queries in time series databases","year":2007,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":25,"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":"Search engine indexing; Scalability; Computer science; Ranging; Series (stratigraphy); Piecewise; Data mining; Algorithm; Univariate; Variable (mathematics); Database; Mathematics; Artificial intelligence; Multivariate statistics; Machine learning","score_opus":0.022915543555870287,"score_gpt":0.25871462556820385,"score_spread":0.23579908201233357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2081471536","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":[0.028322268,0.0036055762,0.95632625,0.00064491894,0.00041514888,0.00023183905,0.0021712733,0.0060496423,0.002233131],"genre_scores_gemma":[0.26273477,0.0017429002,0.72341186,0.00044434273,0.0006852656,0.00039271035,0.0059913713,0.000675197,0.0039216415],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9963257,0.00061202934,0.0007404373,0.00041688385,0.001662826,0.00024215432],"domain_scores_gemma":[0.9892135,0.003981391,0.000716152,0.0038711359,0.0017327633,0.00048502831],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029254884,0.0007970466,0.0031631256,0.0040407847,0.0011260106,0.0046794456,0.0026930508,0.0013163083,0.0050540604],"category_scores_gemma":[0.018827368,0.00071565073,0.0006328709,0.006486262,0.0007737616,0.0081626,0.0037870565,0.0017656032,0.0031670206],"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":[0.002457277,0.00053866283,0.0026399402,0.00069874176,0.00017206308,0.000557343,0.0005687116,0.038727675,0.03613842,0.05531668,0.04282402,0.81936044],"study_design_scores_gemma":[0.00038629273,0.00058720965,0.0013981573,0.00013899352,0.00015048272,0.0013875394,0.00028273265,0.85269374,0.022409057,0.080330685,0.040067326,0.00016785129],"about_ca_topic_score_codex":0.0015769387,"about_ca_topic_score_gemma":0.0014957893,"teacher_disagreement_score":0.0050540604,"about_ca_system_score_codex":0.0008617378,"about_ca_system_score_gemma":0.0015853068,"threshold_uncertainty_score":0.016907513},"labels":[],"label_agreement":null},{"id":"W2081778498","doi":"10.1007/s10115-008-0188-5","title":"Protecting buying agents in e-marketplaces by direct experience trust modelling","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Blockchain Technology Applications and Security","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":"University of Ottawa","funders":"","keywords":"Quality (philosophy); Context (archaeology); Computer science; Trustworthiness; Value (mathematics); Business; Risk analysis (engineering); Marketing; Internet privacy; Computer security; Microeconomics; Economics","score_opus":0.017957197094414755,"score_gpt":0.25437709880640913,"score_spread":0.23641990171199437,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2081778498","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":[0.17816134,0.00017292937,0.8079571,0.0009348687,0.00004382195,0.00012870449,0.00008159372,0.0002819743,0.012237687],"genre_scores_gemma":[0.9824211,0.000060568615,0.014670319,0.000030871175,0.000009473993,0.000036646794,0.00003410759,0.000017317308,0.0027194954],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99621993,0.0018777094,0.00024558906,0.0003798998,0.00079517235,0.00048166962],"domain_scores_gemma":[0.98052406,0.0115057025,0.0017767164,0.003801469,0.0018227204,0.00056930556],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046417997,0.0005139509,0.00086425856,0.0005799021,0.00079374877,0.0031396332,0.0016681674,0.0021773258,0.0045296936],"category_scores_gemma":[0.023234189,0.0006221807,0.0007772941,0.00062433013,0.0024922797,0.009737631,0.0029251794,0.0017703782,0.0006050025],"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":[0.00075166015,0.00024661364,0.004504094,0.00018667684,0.00011034046,0.00048037522,0.0014032244,0.5627456,0.0035780019,0.37709826,0.0012415795,0.04765359],"study_design_scores_gemma":[0.000034396897,0.00007746613,0.0002942012,0.000016867563,0.000021803573,0.00005428723,0.00015415857,0.866014,0.0011048375,0.13137682,0.000830422,0.000020864125],"about_ca_topic_score_codex":0.003757725,"about_ca_topic_score_gemma":0.0028033108,"teacher_disagreement_score":0.0046417997,"about_ca_system_score_codex":0.0012519838,"about_ca_system_score_gemma":0.0014058909,"threshold_uncertainty_score":0.024548471},"labels":[],"label_agreement":null},{"id":"W2081877265","doi":"10.1007/s10115-012-0504-y","title":"An efficient concept-based retrieval model for enhancing text retrieval quality","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","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 Waterloo; D2L (Canada)","funders":"","keywords":"Computer science; Sentence; Information retrieval; Natural language processing; Weighting; Term (time); Semantics (computer science); Latent semantic analysis; Term Discrimination; Representation (politics); Artificial intelligence; Graph; Document retrieval; Relevance (law); Concept search; Theoretical computer science; Search engine","score_opus":0.03897948914390884,"score_gpt":0.30871514418177404,"score_spread":0.2697356550378652,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2081877265","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":[0.019612933,0.00050500006,0.977948,0.00024109616,0.000063803025,0.00009121676,0.00017481153,0.0007645087,0.0005987367],"genre_scores_gemma":[0.47244087,0.00083957583,0.52089435,0.00025927986,0.00025051567,0.000491984,0.0008192457,0.00020006057,0.0038041463],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99901664,0.0002965868,0.00008754757,0.0002118111,0.00031230357,0.00007519556],"domain_scores_gemma":[0.9981437,0.00084968505,0.00012005099,0.00017684461,0.00065659825,0.000053110074],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020468875,0.00066446216,0.0013112675,0.0015422367,0.00048855704,0.0014082346,0.0015493894,0.0012832701,0.0018776093],"category_scores_gemma":[0.0050486275,0.00036478005,0.0010325984,0.0015833902,0.0004336629,0.0035721618,0.00073849614,0.0011017387,0.0010499181],"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":[0.0010851385,0.0006749158,0.0020586823,0.0005953547,0.00028738604,0.0001487729,0.00032101315,0.3070586,0.05074766,0.0392626,0.0116788335,0.5860811],"study_design_scores_gemma":[0.000031218948,0.00006080605,0.00025680193,0.0000069099597,0.000054412292,0.0000458482,0.000011072073,0.99156123,0.0030079228,0.004221866,0.0007254585,0.000016501228],"about_ca_topic_score_codex":0.0042777737,"about_ca_topic_score_gemma":0.004198247,"teacher_disagreement_score":0.0042777737,"about_ca_system_score_codex":0.0010475179,"about_ca_system_score_gemma":0.001399417,"threshold_uncertainty_score":0.010825098},"labels":[],"label_agreement":null},{"id":"W2084144913","doi":"10.1007/s10115-011-0399-z","title":"Best papers from the Fifth International Conference on Advanced Data Mining and Applications (ADMA 2009)","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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":"Simon Fraser University","funders":"","keywords":"Computer science; Data science; Data mining","score_opus":0.06692820956384166,"score_gpt":0.2819794512247109,"score_spread":0.21505124166086925,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2084144913","genre_codex":"editorial","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":[0.006247621,0.18005398,0.060323354,0.1023885,0.51295567,0.00071832875,0.013536625,0.0026188865,0.12115704],"genre_scores_gemma":[0.016912114,0.1131637,0.035564832,0.010437516,0.08561319,0.0002891276,0.021765323,0.0020308301,0.7142233],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969393,0.0003206395,0.00028296176,0.0004302604,0.0017963225,0.00023053099],"domain_scores_gemma":[0.98790866,0.0015744581,0.00042874867,0.0008911379,0.0065895338,0.0026073386],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005463561,0.0015226883,0.0028307026,0.00575458,0.0014046918,0.0075998395,0.0024317168,0.0019334494,0.14412726],"category_scores_gemma":[0.0099070305,0.0005193435,0.0016865693,0.0068457033,0.0005958132,0.0044774255,0.0026782446,0.002411497,0.07318584],"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":[0.00015429998,0.0000762238,0.0003740527,0.0005493392,0.00004620972,0.00008916727,0.000033882217,0.0003540347,0.00084486464,0.0012247798,0.7882355,0.20801756],"study_design_scores_gemma":[0.000036986516,0.000088307504,0.0012775618,0.00034031927,0.000077613724,0.00022864674,0.00007168417,0.0009828996,0.0014201752,0.003493043,0.9919492,0.00003357095],"about_ca_topic_score_codex":0.0023865143,"about_ca_topic_score_gemma":0.005526237,"teacher_disagreement_score":0.14412726,"about_ca_system_score_codex":0.0017976544,"about_ca_system_score_gemma":0.0038143871,"threshold_uncertainty_score":0.4821537},"labels":[],"label_agreement":null},{"id":"W2087414729","doi":"10.1007/s10115-006-0020-z","title":"Detecting outlying subspaces for high-dimensional data: the new task, algorithms, and performance","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":143,"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; Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Linear subspace; Outlier; Subspace topology; Pruning; Computer science; Dimension (graph theory); Heuristic; Anomaly detection; Task (project management); Algorithm; Pattern recognition (psychology); Clustering high-dimensional data; Measure (data warehouse); Point (geometry); Process (computing); Artificial intelligence; Data point; Mathematics; Data mining; Cluster analysis; Combinatorics","score_opus":0.0196734787435495,"score_gpt":0.2493453736811592,"score_spread":0.2296718949376097,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2087414729","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":[0.027255697,0.004847516,0.96340686,0.0027483208,0.00013216306,0.00004250123,0.00014922103,0.00094637106,0.00047143697],"genre_scores_gemma":[0.27874908,0.0045144293,0.71262807,0.00043749233,0.0010925664,0.000120353994,0.0007359012,0.00033164732,0.0013904286],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99143046,0.0033334089,0.000658816,0.0013764706,0.002931431,0.00026952283],"domain_scores_gemma":[0.93289816,0.048427284,0.0030990862,0.009165387,0.0053110206,0.0010989881],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011843909,0.001541212,0.004350826,0.0031748281,0.0014625981,0.006851447,0.003087921,0.004166258,0.0016816588],"category_scores_gemma":[0.05218237,0.0010304541,0.0011734734,0.0049439026,0.003372387,0.011052349,0.0041548824,0.005085617,0.0011687067],"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":[0.0006335271,0.00031069678,0.016106145,0.0005342786,0.0002904817,0.00016405598,0.0005113992,0.05490437,0.008648601,0.024703702,0.010403298,0.8827895],"study_design_scores_gemma":[0.0000536512,0.00017657186,0.0041660257,0.0000535115,0.00007739431,0.00066019583,0.00036195226,0.87242955,0.0047988994,0.1129695,0.0041551143,0.000097754426],"about_ca_topic_score_codex":0.0029224409,"about_ca_topic_score_gemma":0.002151246,"teacher_disagreement_score":0.011843909,"about_ca_system_score_codex":0.00083235506,"about_ca_system_score_gemma":0.0013705884,"threshold_uncertainty_score":0.06263733},"labels":[],"label_agreement":null},{"id":"W2094160481","doi":"10.1007/s10115-003-0105-x","title":"A Linguistic Approach to Extracting Acronym Expansions from Text","year":2004,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Lexicography and Language Studies","field":"Arts and Humanities","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":"Simon Fraser University","funders":"","keywords":"Acronym; Natural language processing; Lexicon; Computer science; Artificial intelligence; Linguistics","score_opus":0.025754580464048802,"score_gpt":0.24229241205647079,"score_spread":0.216537831592422,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2094160481","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":[0.04634039,0.0030441864,0.9150664,0.0015874732,0.0005684851,0.0009162962,0.007893429,0.01362648,0.010956819],"genre_scores_gemma":[0.066058725,0.00093430665,0.9180133,0.00021981346,0.00019794355,0.00029049182,0.008724123,0.0004904273,0.0050708544],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99795073,0.000470596,0.00032910408,0.000579277,0.0005630154,0.00010729752],"domain_scores_gemma":[0.99540895,0.002106582,0.0003866541,0.00047653567,0.0014664172,0.00015485298],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012380035,0.0012523648,0.0011422129,0.011377778,0.0023203879,0.0038586664,0.0015231741,0.0013326203,0.0066969926],"category_scores_gemma":[0.0049551255,0.00074677693,0.0014874253,0.00986371,0.0012497302,0.0043321294,0.002059023,0.0019415028,0.0047468417],"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":[0.00028657998,0.00035179665,0.0031180894,0.0014694326,0.00017439587,0.0011125247,0.0021559913,0.0025458643,0.062351637,0.016978417,0.025727252,0.883728],"study_design_scores_gemma":[0.0003615612,0.00080795126,0.024047127,0.0009912773,0.0014982008,0.00860681,0.010573564,0.30056342,0.11233412,0.11378937,0.42575094,0.0006756943],"about_ca_topic_score_codex":0.004689381,"about_ca_topic_score_gemma":0.010255782,"teacher_disagreement_score":0.011377778,"about_ca_system_score_codex":0.00078101957,"about_ca_system_score_gemma":0.0025491915,"threshold_uncertainty_score":0.022403717},"labels":[],"label_agreement":null},{"id":"W2102549231","doi":"10.1007/s10115-006-0023-9","title":"Node similarity in the citation graph","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":69,"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 Ottawa; Dalhousie University","funders":"Southeast University; Dalhousie University","keywords":"Computer science; Graph; Complementarity (molecular biology); Theoretical computer science; Similarity (geometry); Citation; Random geometric graph; Similitude; Information retrieval; Data mining; Artificial intelligence; Line graph; Voltage graph; World Wide Web","score_opus":0.01017648760587226,"score_gpt":0.24321623998598907,"score_spread":0.23303975238011682,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102549231","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":[0.559788,0.00637731,0.40334043,0.0032494117,0.00025621918,0.00014904902,0.003078832,0.0006850169,0.023075787],"genre_scores_gemma":[0.95194405,0.0021456664,0.039922066,0.0001259539,0.0002910499,0.000103984385,0.0011762233,0.000074638876,0.004216377],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980393,0.00076261756,0.00013144937,0.00039189772,0.0005924597,0.00008229219],"domain_scores_gemma":[0.9749126,0.019177185,0.0019652916,0.0009620233,0.002402098,0.00058083],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0018134734,0.00033418968,0.0011316577,0.012091901,0.0012088375,0.002728873,0.0011300435,0.0020098237,0.0035894897],"category_scores_gemma":[0.031974554,0.00039381764,0.00046607215,0.012723211,0.0012209854,0.0061744316,0.0011898063,0.0007189678,0.000551267],"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":[0.00044817958,0.00023435421,0.044117335,0.0013250785,0.000524707,0.00057785853,0.0013590722,0.100221954,0.0063162087,0.61744565,0.013509125,0.21392047],"study_design_scores_gemma":[0.000043960637,0.00007224855,0.015727408,0.00013943375,0.00030690889,0.00059657654,0.00028075263,0.30066824,0.0018190411,0.67009735,0.010187512,0.000060590795],"about_ca_topic_score_codex":0.0026335157,"about_ca_topic_score_gemma":0.0029730236,"teacher_disagreement_score":0.9879081,"about_ca_system_score_codex":0.0010972791,"about_ca_system_score_gemma":0.00074185047,"threshold_uncertainty_score":0.012008011},"labels":[],"label_agreement":null},{"id":"W2110958602","doi":"10.1007/s10115-011-0415-3","title":"Knowledge-empowered agent information system for privacy payoff in eCommerce","year":2011,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy, Security, and Data Protection","field":"Social Sciences","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 Ottawa","funders":"","keywords":"Reputation; Personally identifiable information; Computer science; Asset (computer security); Stochastic game; Information system; Valuation (finance); Knowledge management; Business; Internet privacy; Computer security; Economics","score_opus":0.050463113740772295,"score_gpt":0.2996881609855993,"score_spread":0.24922504724482702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110958602","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":[0.24619888,0.0007397865,0.6853235,0.004774096,0.00020890629,0.00025418427,0.00022631648,0.001284416,0.06098999],"genre_scores_gemma":[0.96387625,0.00007966996,0.031054895,0.00013220562,0.000034599907,0.00004561577,0.000042643605,0.000018675364,0.004715524],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99792635,0.00097413175,0.00013399396,0.00028034064,0.0004600628,0.00022520463],"domain_scores_gemma":[0.99478585,0.0027227171,0.00031030492,0.0011702524,0.0006877991,0.00032311227],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003183243,0.0002944365,0.0005633912,0.00074704067,0.0016143287,0.0036670081,0.0010831155,0.0019071447,0.005736417],"category_scores_gemma":[0.009032538,0.00019885447,0.00049841567,0.0005837398,0.0010479984,0.006020159,0.0029389246,0.0014831757,0.0007072314],"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":[0.0012546858,0.0006740978,0.0035214468,0.00018870746,0.00011461437,0.0006317645,0.0010894934,0.04194597,0.010966445,0.76237595,0.0064924266,0.17074448],"study_design_scores_gemma":[0.00014305777,0.0002558602,0.0012735409,0.00004521631,0.00013837536,0.00045179867,0.0002924894,0.50723785,0.012943358,0.46427456,0.012885616,0.000058251342],"about_ca_topic_score_codex":0.0006787391,"about_ca_topic_score_gemma":0.0007270822,"teacher_disagreement_score":0.005736417,"about_ca_system_score_codex":0.00080147,"about_ca_system_score_gemma":0.0013069572,"threshold_uncertainty_score":0.019190192},"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":"W2121566986","doi":"10.1007/s10115-008-0179-6","title":"PADS: a simple yet effective pattern-aware dynamic search method for fast maximal frequent pattern mining","year":2008,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":21,"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; Executable; Data mining; Benchmark (surveying); Pattern search; Simple (philosophy); Linear subspace; Tree (set theory); Algorithm; Mathematics; Programming language","score_opus":0.019725056746500533,"score_gpt":0.2976702125706421,"score_spread":0.27794515582414153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121566986","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":[0.01438856,0.00088106876,0.9580609,0.00025127575,0.0001759569,0.0002345753,0.0020679582,0.022282215,0.0016574081],"genre_scores_gemma":[0.08911966,0.00045252586,0.9009691,0.00023536972,0.00012629229,0.00044586588,0.004239587,0.0007699828,0.0036416184],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987141,0.0002449017,0.00015833213,0.00025620675,0.00055089116,0.0000755943],"domain_scores_gemma":[0.9978886,0.0011380079,0.00013920755,0.00045389243,0.000289117,0.00009127304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010314083,0.0011404803,0.0016696063,0.004487724,0.0007601981,0.0017723166,0.0023994548,0.0009681152,0.00789005],"category_scores_gemma":[0.0057793283,0.00085657986,0.0010423523,0.004902567,0.00045834915,0.0030681125,0.002471954,0.0011623717,0.0037253331],"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":[0.0010417092,0.00023318763,0.0026804805,0.0006447144,0.0002740172,0.00030765217,0.00013067911,0.008301948,0.015638484,0.0053025954,0.028524535,0.93692],"study_design_scores_gemma":[0.00089753215,0.0007380916,0.0027541765,0.00012330033,0.0005138077,0.002203855,0.0003142006,0.8152127,0.04202738,0.06030845,0.07475117,0.00015531333],"about_ca_topic_score_codex":0.0010151762,"about_ca_topic_score_gemma":0.0022993411,"teacher_disagreement_score":0.00789005,"about_ca_system_score_codex":0.0003044397,"about_ca_system_score_gemma":0.0015245068,"threshold_uncertainty_score":0.026394844},"labels":[],"label_agreement":null},{"id":"W2126734246","doi":"10.1007/s10115-009-0198-y","title":"Boosting support vector machines for imbalanced data sets","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":279,"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":"National Institutes of Health","keywords":"Support vector machine; Boosting (machine learning); Classifier (UML); Machine learning; Artificial intelligence; Computer science; Margin classifier; Data mining; Structured support vector machine; Relevance vector machine; Pattern recognition (psychology)","score_opus":0.03635550629993588,"score_gpt":0.3092264469905259,"score_spread":0.27287094069059004,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126734246","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":[0.039036393,0.0020982546,0.9563919,0.000492308,0.00023681579,0.0000759656,0.00015002824,0.00075033074,0.00076790684],"genre_scores_gemma":[0.6567967,0.0015605524,0.33659083,0.00021573454,0.0007992375,0.00021673697,0.0011875647,0.00014681095,0.0024858653],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996557,0.0012109112,0.00026476086,0.00048614107,0.0012590085,0.00022211634],"domain_scores_gemma":[0.98898995,0.006535327,0.00067491434,0.0014454712,0.0020895654,0.00026469032],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008011001,0.0009689268,0.0026823375,0.0024037687,0.0007545329,0.0014901381,0.0017015678,0.0012072163,0.00100659],"category_scores_gemma":[0.024216317,0.0006982004,0.00089063693,0.0022780534,0.0007072627,0.0025356635,0.001753574,0.0022736422,0.00067971135],"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":[0.0005129984,0.0003408183,0.004731841,0.00034844302,0.00022557542,0.00013319768,0.00016608465,0.29803935,0.0044112843,0.0138503015,0.009289028,0.66795117],"study_design_scores_gemma":[0.000013630624,0.00003460746,0.00043718133,0.000011615996,0.00002235099,0.000027019823,0.000013850334,0.98641866,0.0009436852,0.01130033,0.0007718347,0.0000052578957],"about_ca_topic_score_codex":0.0015280596,"about_ca_topic_score_gemma":0.0012885063,"teacher_disagreement_score":0.008011001,"about_ca_system_score_codex":0.000646519,"about_ca_system_score_gemma":0.0008980531,"threshold_uncertainty_score":0.042366743},"labels":[],"label_agreement":null},{"id":"W2126966660","doi":"10.1007/s10115-005-0233-6","title":"Capabilities of outlier detection schemes in large datasets, framework and methodologies","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":86,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"","keywords":"Outlier; Computer science; Anomaly detection; Data mining; Scheme (mathematics); Credit card fraud; Matching (statistics); Artificial intelligence; Machine learning; Pattern recognition (psychology); Credit card; Mathematics; Statistics","score_opus":0.016619104645449252,"score_gpt":0.28303760100419423,"score_spread":0.266418496358745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126966660","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":[0.04848084,0.0009675084,0.94541585,0.0014186734,0.00006323633,0.00009411418,0.00039625418,0.0017264077,0.0014370427],"genre_scores_gemma":[0.5292632,0.000826155,0.46743217,0.0002241384,0.00016838966,0.00014850842,0.0008248018,0.00019680212,0.0009157824],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98985684,0.004410428,0.00086280203,0.0015146713,0.0030011993,0.0003540853],"domain_scores_gemma":[0.92983675,0.04826687,0.0037456267,0.011562516,0.0056166584,0.000971498],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.023465926,0.0010107524,0.001525185,0.0037581942,0.0013990232,0.005631189,0.0025258728,0.0023554796,0.0016682873],"category_scores_gemma":[0.08315993,0.0006009713,0.0010292762,0.0035891167,0.0018242681,0.009106221,0.004438931,0.0024824159,0.000672808],"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":[0.0010752862,0.00047718722,0.025114663,0.0007794263,0.0005804109,0.00031881302,0.0008515539,0.27094954,0.012862398,0.10749846,0.0071244384,0.57236785],"study_design_scores_gemma":[0.000038146125,0.00016123576,0.0020167404,0.00006107907,0.00007445203,0.0002885987,0.00020128387,0.87546504,0.0069127884,0.11177493,0.0029606896,0.000045090004],"about_ca_topic_score_codex":0.0021097239,"about_ca_topic_score_gemma":0.0016389779,"teacher_disagreement_score":0.023465926,"about_ca_system_score_codex":0.0013439264,"about_ca_system_score_gemma":0.0018182894,"threshold_uncertainty_score":0.12410116},"labels":[],"label_agreement":null},{"id":"W2127094475","doi":"10.1007/s10115-004-0189-y","title":"Optimizing complex queries based on similarities of subqueries","year":2005,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","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":"IBM (Canada)","funders":"International Business Machines Corporation","keywords":"Computer science; Query optimization; Joins; Key (lock); Similarity (geometry); Information retrieval; Graph; Database; Query expansion; Query language; Data mining; Theoretical computer science; Artificial intelligence","score_opus":0.021749295487071173,"score_gpt":0.24014789906072057,"score_spread":0.2183986035736494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127094475","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":[0.5585521,0.0018244023,0.42886448,0.00093654956,0.00014781795,0.00047862812,0.0006744443,0.004070262,0.0044513494],"genre_scores_gemma":[0.78566873,0.00040602687,0.20927785,0.00020345776,0.0002521379,0.00015891099,0.001480659,0.00072801125,0.0018242385],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9885958,0.0024250757,0.0011142934,0.0019537706,0.005129637,0.000781434],"domain_scores_gemma":[0.9681253,0.022058742,0.0021108652,0.0037264512,0.003234958,0.00074372994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042685303,0.001362877,0.0038502524,0.0040926323,0.0013013397,0.0039841505,0.0030557748,0.0019846805,0.0047929217],"category_scores_gemma":[0.035014752,0.0009880578,0.0014182127,0.00553178,0.0013184368,0.009306245,0.0033037832,0.0024838403,0.00081256364],"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":[0.005373107,0.001522611,0.03359206,0.0007263157,0.0008318726,0.000503324,0.001158137,0.25069138,0.041364092,0.0281405,0.009393995,0.62670255],"study_design_scores_gemma":[0.00015222126,0.0005630767,0.004200055,0.000021019769,0.00020457235,0.00024236686,0.00033518235,0.9504631,0.009574764,0.03251589,0.0016882614,0.00003946913],"about_ca_topic_score_codex":0.0037800507,"about_ca_topic_score_gemma":0.0059034103,"teacher_disagreement_score":0.0047929217,"about_ca_system_score_codex":0.0019910745,"about_ca_system_score_gemma":0.001935298,"threshold_uncertainty_score":0.022574425},"labels":[],"label_agreement":null},{"id":"W2127392028","doi":"10.1007/s10115-010-0298-8","title":"User-centric query refinement and processing using granularity-based strategies","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Rough Sets and Fuzzy Logic","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 Regina","funders":"","keywords":"Computer science; Granularity; Unification; Ontology; Context (archaeology); Information retrieval; Data mining; Data science","score_opus":0.015292020251002863,"score_gpt":0.25043735842326176,"score_spread":0.23514533817225888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127392028","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":[0.025272183,0.00039864233,0.9685348,0.0003617057,0.000034782257,0.000253701,0.00010783692,0.002785395,0.002250985],"genre_scores_gemma":[0.43177965,0.00026707072,0.5648986,0.00024524034,0.00006408945,0.00020283372,0.00038602352,0.0003752569,0.0017813027],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98768616,0.0046133273,0.0011015508,0.0013247969,0.0044571813,0.00081697456],"domain_scores_gemma":[0.980877,0.009472259,0.0008688848,0.005647324,0.0026432036,0.0004912899],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009965888,0.0010920423,0.0022793987,0.003309837,0.0014476686,0.00475855,0.0027958783,0.0014066155,0.0039015473],"category_scores_gemma":[0.024479685,0.00095498824,0.0014313849,0.0032267836,0.0013682545,0.0054826946,0.0042614145,0.0019389542,0.000996827],"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":[0.0035880327,0.0006866741,0.008438812,0.00093869073,0.0007454907,0.00075338245,0.0061924234,0.043798134,0.13992392,0.104911916,0.012726584,0.677296],"study_design_scores_gemma":[0.00031092623,0.00046652084,0.0033886798,0.0001453276,0.0006251562,0.00095629337,0.0022441542,0.76987815,0.11489598,0.08495924,0.02189633,0.00023319336],"about_ca_topic_score_codex":0.005038487,"about_ca_topic_score_gemma":0.0054765707,"teacher_disagreement_score":0.009965888,"about_ca_system_score_codex":0.0012042291,"about_ca_system_score_gemma":0.0022054971,"threshold_uncertainty_score":0.05270523},"labels":[],"label_agreement":null},{"id":"W2127408419","doi":"10.1007/s10115-009-0202-6","title":"Integrating multiple document features in language models for expert finding","year":2009,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Expert finding and Q&A systems","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":"York University","funders":"Engineering and Physical Sciences Research Council; Commonwealth Scientific and Industrial Research Organisation","keywords":"Computer science; Information retrieval; Language model; Paragraph; Document retrieval; Document management system; Question answering; PageRank; Intranet; Key (lock); Subject-matter expert; Expert system; Artificial intelligence; Natural language processing; World Wide Web; The Internet","score_opus":0.016461128839897702,"score_gpt":0.2821247681681098,"score_spread":0.2656636393282121,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127408419","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":[0.048213847,0.0007381566,0.9417662,0.0009040574,0.00016840525,0.00016456252,0.0016873667,0.0042816103,0.0020758337],"genre_scores_gemma":[0.6081293,0.0005258964,0.38345695,0.0002703084,0.00020896042,0.0003669549,0.0026245015,0.0007805098,0.003636634],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99785054,0.0008852026,0.0002520218,0.00040150408,0.00047854902,0.00013214402],"domain_scores_gemma":[0.98369217,0.012643154,0.000625264,0.0011048198,0.0016751055,0.0002594715],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004282814,0.0009947096,0.0012755436,0.0027065356,0.0009497906,0.0031260059,0.0022165321,0.0020117117,0.0039658137],"category_scores_gemma":[0.019875465,0.00084479613,0.0022301958,0.002360844,0.0007333153,0.006504217,0.0012738439,0.002765212,0.0022153154],"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":[0.001527261,0.0008296387,0.010523809,0.000787719,0.0006222136,0.00050940365,0.00075633056,0.41545305,0.011200022,0.029308291,0.012449038,0.5160333],"study_design_scores_gemma":[0.0000331689,0.00006438082,0.00048261546,0.000025470164,0.000115352035,0.00008082993,0.000046541143,0.9729768,0.0019474543,0.02276626,0.0014275305,0.00003371894],"about_ca_topic_score_codex":0.008427418,"about_ca_topic_score_gemma":0.012404598,"teacher_disagreement_score":0.008427418,"about_ca_system_score_codex":0.0015814247,"about_ca_system_score_gemma":0.0016387664,"threshold_uncertainty_score":0.022649944},"labels":[],"label_agreement":null},{"id":"W2134522724","doi":"10.1007/s10115-006-0015-9","title":"Applying language modeling to session identification from database trace logs","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Service-Oriented Architecture and Web Services","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":false,"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":"Computer science; Session (web analytics); TRACE (psycholinguistics); Database; Language model; Identification (biology); Data modeling; Artificial intelligence; Data mining; Natural language processing; Machine learning; Information retrieval","score_opus":0.009350929948738405,"score_gpt":0.2408473831906952,"score_spread":0.2314964532419568,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134522724","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":[0.026431896,0.000102900034,0.96679395,0.00022687967,0.00003371153,0.000112773494,0.00035975562,0.0051874393,0.0007505982],"genre_scores_gemma":[0.52249306,0.00029900315,0.47102973,0.00015356365,0.000062306404,0.00030222232,0.0019456757,0.00094736397,0.0027670723],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99664503,0.0014083463,0.0002889399,0.0006003531,0.000771346,0.00028595954],"domain_scores_gemma":[0.98309314,0.01176067,0.0008947903,0.002075425,0.0019123552,0.0002636517],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043169768,0.00094481034,0.0010948615,0.0028756713,0.0011075647,0.0039530015,0.0021834632,0.0016131403,0.001537169],"category_scores_gemma":[0.019751757,0.0007957405,0.0019021871,0.002197366,0.0010033953,0.0042719753,0.0013961763,0.0018080608,0.0010595693],"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":[0.00094252545,0.0009745886,0.025045546,0.0007640666,0.00045383172,0.0013791946,0.003301383,0.4822869,0.03371984,0.07936772,0.006765586,0.36499888],"study_design_scores_gemma":[0.000010921112,0.000028084392,0.00041481014,0.000019862182,0.000050117196,0.00008968358,0.000103656756,0.9721897,0.0050652577,0.02073175,0.001267762,0.000028305822],"about_ca_topic_score_codex":0.017716346,"about_ca_topic_score_gemma":0.016290942,"teacher_disagreement_score":0.017716346,"about_ca_system_score_codex":0.0015710434,"about_ca_system_score_gemma":0.0035767318,"threshold_uncertainty_score":0.035226464},"labels":[],"label_agreement":null},{"id":"W2135868586","doi":"10.1007/s10115-010-0341-9","title":"Classifier-based acronym extraction for business documents","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","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":"École de Technologie Supérieure","funders":"","keywords":"Acronym; Computer science; Classifier (UML); Information extraction; Precision and recall; Natural language processing; Artificial intelligence; Information retrieval; Data mining; Linguistics","score_opus":0.013582429754063777,"score_gpt":0.29530810560313925,"score_spread":0.2817256758490755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135868586","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":[0.11138631,0.0048693847,0.75383025,0.002072001,0.0019033055,0.0017674795,0.05588575,0.0478805,0.020404976],"genre_scores_gemma":[0.14836518,0.0013536694,0.7916337,0.00026048018,0.00024883205,0.00043646112,0.050631378,0.0010667229,0.006003579],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99801123,0.00028389477,0.00037335278,0.0005406141,0.0006170304,0.0001738467],"domain_scores_gemma":[0.9952172,0.0015916397,0.0005214002,0.0005270334,0.0019381257,0.00020461307],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011063088,0.0011919582,0.0013460111,0.011873848,0.0017884474,0.002769464,0.0010996065,0.0012981837,0.006391279],"category_scores_gemma":[0.007566184,0.00041986757,0.0014548531,0.008095327,0.00035924432,0.0032276472,0.0012275116,0.0013462142,0.0072675343],"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":[0.0005388529,0.00033585436,0.007283065,0.0015236697,0.00020277115,0.0011515926,0.0006089104,0.0027733892,0.055720445,0.0092993975,0.08014411,0.8404179],"study_design_scores_gemma":[0.00033187077,0.00044870627,0.01951351,0.0010312939,0.0010136769,0.0048873974,0.002360484,0.31936187,0.17369176,0.039212905,0.43781784,0.0003286669],"about_ca_topic_score_codex":0.005980211,"about_ca_topic_score_gemma":0.008971906,"teacher_disagreement_score":0.011873848,"about_ca_system_score_codex":0.0011529444,"about_ca_system_score_gemma":0.0034596585,"threshold_uncertainty_score":0.021380961},"labels":[],"label_agreement":null},{"id":"W2139151761","doi":"10.1007/s10115-006-0031-9","title":"Mining gene–sample–time microarray data: a coherent gene cluster discovery approach","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"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":"Division of Biological Infrastructure; National Institutes of Health","keywords":"Microarray analysis techniques; Data mining; Computational biology; Gene; Sample (material); Microarray databases; Microarray; Cluster analysis; Set (abstract data type); Computer science; Data set; Gene chip analysis; DNA microarray; Biology; Bioinformatics; Gene expression; Genetics; Artificial intelligence; Physics","score_opus":0.016347351486732498,"score_gpt":0.24262301528054223,"score_spread":0.22627566379380973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139151761","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":[0.019585066,0.00040689178,0.9779478,0.00029742336,0.000026239897,0.0001543171,0.00056763785,0.0007932398,0.00022146663],"genre_scores_gemma":[0.10772234,0.00023993441,0.88865376,0.00015711845,0.00007174046,0.0003207173,0.0023814884,0.00012339486,0.0003296345],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99460584,0.0013572341,0.00053454767,0.0014270138,0.0018261848,0.00024924314],"domain_scores_gemma":[0.9933949,0.0037948024,0.00051486166,0.0011238729,0.00096798287,0.00020357376],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005850302,0.0009944135,0.0026068983,0.0055561867,0.0015142139,0.002829279,0.0027505714,0.0015544092,0.0010985417],"category_scores_gemma":[0.012308806,0.0008377763,0.0028707378,0.0062267487,0.0012506208,0.0021179477,0.002697977,0.0015616548,0.0006365405],"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":[0.0013265132,0.0013623731,0.03177888,0.0014746186,0.0025811985,0.0007639532,0.001061578,0.04161033,0.15009624,0.027140643,0.00894816,0.73185563],"study_design_scores_gemma":[0.00034016476,0.00093971734,0.028802138,0.0001221373,0.0019350523,0.0011516947,0.00062791526,0.8126445,0.051339705,0.09019411,0.01165525,0.0002475967],"about_ca_topic_score_codex":0.00183932,"about_ca_topic_score_gemma":0.0036021755,"teacher_disagreement_score":0.005850302,"about_ca_system_score_codex":0.00081928546,"about_ca_system_score_gemma":0.0029790013,"threshold_uncertainty_score":0.030939698},"labels":[],"label_agreement":null},{"id":"W2145373745","doi":"10.1007/s10115-003-0135-4","title":"Ontologies for Knowledge Management: An Information Systems Perspective","year":2004,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":201,"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 Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Ontology; Computer science; Knowledge management; Perspective (graphical); Data science; Interdependence; Representation (politics); Knowledge representation and reasoning; Management science; Epistemology; Artificial intelligence; Sociology; Engineering","score_opus":0.021858794082854682,"score_gpt":0.27250576757426637,"score_spread":0.2506469734914117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145373745","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":[0.0063414797,0.092590585,0.6831421,0.07201799,0.002382085,0.00014645532,0.0005913618,0.00054819835,0.14223976],"genre_scores_gemma":[0.36814365,0.11515844,0.4707858,0.011581068,0.0084744785,0.000705742,0.0015160324,0.00037008486,0.023264736],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99480313,0.0023083442,0.0005815489,0.0005644877,0.0013673063,0.0003751387],"domain_scores_gemma":[0.99283314,0.0048163026,0.00047890205,0.0007974415,0.0006370759,0.00043711238],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0076881205,0.0014496654,0.0025806613,0.010551116,0.0033094266,0.019991737,0.004548378,0.007125156,0.005406705],"category_scores_gemma":[0.008595577,0.001541071,0.0020504303,0.012623373,0.020650014,0.04507128,0.005714162,0.006999721,0.001475792],"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":[0.0000041103276,0.000012386832,0.000040204322,0.00008915177,0.000016725917,0.000040077764,0.0002343348,0.00058325316,0.00004968256,0.9927511,0.0015586219,0.00462024],"study_design_scores_gemma":[0.0000051513666,0.000004092967,0.00004495496,0.00012719024,0.000015876349,0.000051842962,0.00024294452,0.001621469,0.000101303696,0.9708356,0.026940104,0.000009526494],"about_ca_topic_score_codex":0.0076807896,"about_ca_topic_score_gemma":0.0051226914,"teacher_disagreement_score":0.019991737,"about_ca_system_score_codex":0.006116792,"about_ca_system_score_gemma":0.0039391792,"threshold_uncertainty_score":0.044380665},"labels":[],"label_agreement":null},{"id":"W2146624380","doi":"10.1007/s10115-006-0024-8","title":"Answering ad hoc aggregate queries from data streams using prefix aggregate trees","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Database Systems and Queries","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":"Simon Fraser University","funders":"National University of Singapore","keywords":"Computer science; Aggregate (composite); Scalability; Trie; Prefix; Data stream mining; Data stream; Data warehouse; Data mining; Database; Data structure; Information retrieval","score_opus":0.02539816418324298,"score_gpt":0.257263077859474,"score_spread":0.23186491367623102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146624380","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":[0.1131701,0.0012277295,0.8752964,0.00091005076,0.0001874523,0.00030576315,0.0020676611,0.005271998,0.0015627947],"genre_scores_gemma":[0.5105448,0.0011902299,0.47687584,0.0003918043,0.00043619648,0.00029760913,0.0078266235,0.00041217514,0.002024767],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99531674,0.0006835784,0.0007602776,0.0007508041,0.0021715292,0.0003169777],"domain_scores_gemma":[0.9871508,0.008255642,0.0008368188,0.0019050223,0.0014803977,0.00037130387],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041783457,0.0011667412,0.0028649333,0.0026944242,0.0014230304,0.00483431,0.0019344856,0.0022088469,0.0017865705],"category_scores_gemma":[0.014575249,0.0010208664,0.0013653113,0.0053178696,0.00095009926,0.010493503,0.003330774,0.002215965,0.00076951215],"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":[0.0036348489,0.0011094976,0.020980138,0.0014989657,0.0009109913,0.0016016675,0.0023826957,0.1981337,0.06432297,0.04017056,0.026028441,0.63922554],"study_design_scores_gemma":[0.00009372772,0.00016553041,0.0010137148,0.00003498153,0.00015587856,0.00031633832,0.0006740466,0.92352015,0.015106265,0.05393293,0.0049498104,0.00003652437],"about_ca_topic_score_codex":0.0025180145,"about_ca_topic_score_gemma":0.0038546934,"teacher_disagreement_score":0.00483431,"about_ca_system_score_codex":0.0008822839,"about_ca_system_score_gemma":0.0015764859,"threshold_uncertainty_score":0.022097468},"labels":[],"label_agreement":null},{"id":"W2151207331","doi":"10.1007/s10115-008-0127-5","title":"Multirelational classification: a multiple view approach","year":2008,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":40,"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":"Shanghai Jiao Tong University","keywords":"Computer science; Relational database; Construct (python library); Table (database); Set (abstract data type); Feature (linguistics); Data mining; Relevance (law); Artificial intelligence; Information retrieval","score_opus":0.04522474571286368,"score_gpt":0.24907032545918464,"score_spread":0.20384557974632095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151207331","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":[0.007803008,0.0015858274,0.9851151,0.001056134,0.00020160111,0.00009105446,0.00031163814,0.00048249596,0.0033531508],"genre_scores_gemma":[0.31451404,0.001873411,0.6765507,0.00062722,0.00063514384,0.00025066308,0.0015774913,0.00030085188,0.0036705115],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99187297,0.0032289317,0.000484297,0.0014076887,0.002538481,0.00046757807],"domain_scores_gemma":[0.9912418,0.004168655,0.0006740282,0.0017087483,0.0018398099,0.0003669517],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006538544,0.0011364558,0.0021955653,0.005617841,0.0015398778,0.0081944065,0.0035864557,0.0022258137,0.0051287822],"category_scores_gemma":[0.014912399,0.00083687744,0.0036286318,0.0051946067,0.0012463809,0.008015023,0.003057084,0.002801332,0.0017424187],"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":[0.00056620105,0.000517417,0.010096086,0.0007786528,0.0013179367,0.0005707039,0.001053694,0.02499255,0.010515593,0.16699712,0.021252496,0.7613415],"study_design_scores_gemma":[0.000058812777,0.00018277572,0.0037646207,0.00025423383,0.0007292874,0.0007526894,0.0008893063,0.6755484,0.005668349,0.2910421,0.020980924,0.00012843784],"about_ca_topic_score_codex":0.0026960047,"about_ca_topic_score_gemma":0.0030484363,"teacher_disagreement_score":0.0081944065,"about_ca_system_score_codex":0.0011783683,"about_ca_system_score_gemma":0.0015567398,"threshold_uncertainty_score":0.034579515},"labels":[],"label_agreement":null},{"id":"W2153689444","doi":"10.1007/s10115-010-0311-2","title":"The k-anonymity and l-diversity approaches for privacy preservation in social networks against neighborhood attacks","year":2010,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":229,"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":"Data publishing; Social network (sociolinguistics); Computer science; Anonymity; k-anonymity; Adversary; Internet privacy; Computer security; Information privacy; Identity (music); Diversity (politics); Data science; Publishing; Social media; World Wide Web; Sociology","score_opus":0.04483096747457849,"score_gpt":0.25951115515531614,"score_spread":0.21468018768073766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153689444","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":[0.0484931,0.0015242551,0.9311755,0.0044773226,0.00014912101,0.00012452494,0.00024994373,0.00016748771,0.013638711],"genre_scores_gemma":[0.9125812,0.0008794064,0.08064794,0.0005910388,0.00042696437,0.00021464081,0.00011026888,0.00004837058,0.004500086],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9854506,0.0071720057,0.00062185136,0.001881471,0.003593889,0.0012802521],"domain_scores_gemma":[0.96670514,0.017925298,0.0027639647,0.008933272,0.0026799971,0.0009923275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0071834153,0.0006698808,0.0020918278,0.0024258,0.0053294622,0.0059006074,0.003309782,0.0038464135,0.0020550985],"category_scores_gemma":[0.02520445,0.00066825637,0.0019850167,0.0026981158,0.0069668097,0.016143167,0.008561075,0.005128773,0.00061234157],"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":[0.0005584576,0.00012382523,0.0018636642,0.00033267555,0.00022486631,0.00028424195,0.0019277863,0.06564105,0.0038313682,0.82499176,0.0041049835,0.09611527],"study_design_scores_gemma":[0.000062933854,0.0001588116,0.0005718829,0.00008273919,0.00010436099,0.0005973472,0.0007766874,0.21867123,0.0058396533,0.76485217,0.008163278,0.00011894609],"about_ca_topic_score_codex":0.00092022994,"about_ca_topic_score_gemma":0.0007327708,"teacher_disagreement_score":0.0071834153,"about_ca_system_score_codex":0.0024667226,"about_ca_system_score_gemma":0.0022166555,"threshold_uncertainty_score":0.037989974},"labels":[],"label_agreement":null},{"id":"W2155224116","doi":"10.1007/s10115-012-0495-8","title":"Estimating feature ratings through an effective review selection approach","year":2013,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":13,"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; Feature (linguistics); Purchasing; Feature selection; Product (mathematics); Recommender system; Machine learning; Selection (genetic algorithm); Sentiment analysis; Artificial intelligence; Data mining; Information retrieval; Marketing","score_opus":0.01256800742941423,"score_gpt":0.2659206055018179,"score_spread":0.2533525980724037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2155224116","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":[0.06664125,0.0030471303,0.92366225,0.00058792584,0.00029574693,0.00046100022,0.0007528163,0.001518447,0.0030335449],"genre_scores_gemma":[0.6093815,0.0012076913,0.38000894,0.00022173973,0.000949306,0.00042217842,0.0019303565,0.00012750611,0.0057508466],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.995937,0.0013471182,0.00034797995,0.0006312073,0.0015454947,0.000191267],"domain_scores_gemma":[0.9882881,0.005195532,0.0008750958,0.0006042421,0.0048491647,0.0001878998],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038322138,0.0011500322,0.0017227246,0.0046589402,0.0005883146,0.0019519727,0.0012173448,0.001188,0.0016771196],"category_scores_gemma":[0.015526925,0.00038767103,0.0010967342,0.0029566218,0.0002501567,0.0018111608,0.00061446754,0.00074768375,0.0016395648],"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":[0.00064555724,0.00044794698,0.015071817,0.00057583867,0.00079214375,0.00033634194,0.00015374548,0.016472483,0.031988125,0.002715764,0.021295322,0.90950495],"study_design_scores_gemma":[0.00012350245,0.0005592709,0.021664556,0.000056035493,0.0008291542,0.000587978,0.00013602097,0.94541395,0.016865246,0.0052207573,0.008449137,0.000094470626],"about_ca_topic_score_codex":0.002551422,"about_ca_topic_score_gemma":0.005132187,"teacher_disagreement_score":0.0046589402,"about_ca_system_score_codex":0.00047520836,"about_ca_system_score_gemma":0.001009763,"threshold_uncertainty_score":0.02026695},"labels":[],"label_agreement":null},{"id":"W2184821933","doi":"10.1007/s10115-015-0888-6","title":"Mining contentious documents","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","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 Alberta","funders":"","keywords":"Computer science; Information retrieval; Natural language processing; Data science","score_opus":0.037327002593624904,"score_gpt":0.26374361425195963,"score_spread":0.22641661165833474,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2184821933","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":[0.450472,0.017184662,0.49250874,0.0031568927,0.00064164796,0.0008890263,0.012794656,0.004954062,0.017398316],"genre_scores_gemma":[0.7850077,0.0051479866,0.17455351,0.00034561587,0.0012871067,0.00037442974,0.018356003,0.0005329968,0.0143946195],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99666303,0.00067761954,0.00034346114,0.00083934347,0.0011851962,0.00029134162],"domain_scores_gemma":[0.98930115,0.0057641477,0.000969129,0.0015288823,0.002031063,0.0004055273],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001910277,0.0011397614,0.001358361,0.017416166,0.002125046,0.004419041,0.0017564856,0.0016922856,0.0030629917],"category_scores_gemma":[0.015475426,0.00086028816,0.0016420175,0.010902338,0.0009528828,0.0064950846,0.001591307,0.0015510404,0.0024519921],"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":[0.0014691912,0.001016339,0.043413162,0.001971798,0.0010682911,0.0020778982,0.0024652013,0.013914859,0.04632259,0.028107166,0.034559038,0.8236145],"study_design_scores_gemma":[0.00029816298,0.0010269382,0.061127782,0.00081247994,0.0025476175,0.0063098865,0.004351198,0.48786873,0.11366956,0.17431714,0.14740196,0.0002685072],"about_ca_topic_score_codex":0.0022415943,"about_ca_topic_score_gemma":0.0038996928,"teacher_disagreement_score":0.017416166,"about_ca_system_score_codex":0.0010456947,"about_ca_system_score_gemma":0.0018892887,"threshold_uncertainty_score":0.010246754},"labels":[],"label_agreement":null},{"id":"W2237502633","doi":"10.1007/s10115-015-0909-5","title":"A feature location approach supported by time-aware weighting of terms associated with developer expertise profiles","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software Engineering 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 Lethbridge","funders":"","keywords":"Computer science; Feature (linguistics); Weighting; Context (archaeology); Data mining; Metadata; Feature model; Software; Source code; Information retrieval; Term (time); Set (abstract data type); World Wide Web","score_opus":0.008221255373777811,"score_gpt":0.2118801598688599,"score_spread":0.20365890449508212,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2237502633","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":[0.10201124,0.001728563,0.8844143,0.0004632368,0.00026856,0.00026783333,0.002422932,0.0049737957,0.0034494854],"genre_scores_gemma":[0.40606517,0.0005524205,0.58184904,0.00013839663,0.00018415693,0.00020209774,0.0034279525,0.0006625889,0.0069182087],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99737656,0.00043381067,0.0002838864,0.00068482285,0.0010195913,0.00020137837],"domain_scores_gemma":[0.992191,0.002064954,0.00096126046,0.0013366733,0.0030360387,0.00040997274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002081121,0.0009599379,0.0010231584,0.008062457,0.00078562804,0.0020720558,0.0015308106,0.0012623619,0.0024482638],"category_scores_gemma":[0.012007576,0.00041422923,0.0009557488,0.0066773137,0.00030411998,0.0033235902,0.0020467688,0.0010789422,0.0017740628],"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":[0.0008364763,0.0004632462,0.015146765,0.00052692444,0.00025812341,0.0002793504,0.0008961269,0.005905837,0.09198401,0.005713934,0.012733757,0.86525536],"study_design_scores_gemma":[0.00023161391,0.0009333374,0.061230958,0.000358679,0.0012155115,0.002458565,0.0016805,0.7100278,0.10815516,0.043126628,0.07013902,0.00044224973],"about_ca_topic_score_codex":0.007100055,"about_ca_topic_score_gemma":0.015903711,"teacher_disagreement_score":0.008062457,"about_ca_system_score_codex":0.0007499826,"about_ca_system_score_gemma":0.0017374958,"threshold_uncertainty_score":0.01411742},"labels":[],"label_agreement":null},{"id":"W2289757941","doi":"10.1007/s10115-016-0929-9","title":"Detecting emerging and evolving novelties with locally adaptive density ratio estimation","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Anomaly Detection Techniques and Applications","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 Waterloo","funders":"","keywords":"Computer science; Novelty; Robustness (evolution); Kernel density estimation; Benchmark (surveying); Kernel (algebra); Artificial intelligence; Set (abstract data type); Density estimation; Focus (optics); Machine learning; Data mining; Pattern recognition (psychology); Mathematics; Statistics","score_opus":0.009207441277482483,"score_gpt":0.21909077333278848,"score_spread":0.209883332055306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2289757941","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":[0.10718953,0.00043524016,0.8904106,0.00020979425,0.000036407124,0.000047970592,0.000081491846,0.000834634,0.00075428793],"genre_scores_gemma":[0.78769016,0.00026029258,0.21037735,0.00008072436,0.00009178429,0.00005573969,0.0002082793,0.00008097626,0.0011547004],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991291,0.00019875547,0.00004441263,0.0002501108,0.00028349564,0.0000941802],"domain_scores_gemma":[0.99554425,0.0024758363,0.0007324217,0.0005258052,0.0005671419,0.00015461726],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016947694,0.0005631834,0.0011302808,0.0027345677,0.00034085443,0.0012738998,0.0016825172,0.0011326812,0.00075644656],"category_scores_gemma":[0.009243264,0.0003815119,0.0007433737,0.0016014422,0.0009692407,0.002415435,0.001998712,0.0013507516,0.00040421746],"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":[0.0006857896,0.00034906424,0.037295394,0.0002761061,0.00038663027,0.00068838673,0.00046631045,0.1299897,0.058006085,0.014416826,0.0028433404,0.7545965],"study_design_scores_gemma":[0.000014019948,0.00007405788,0.004673796,0.0000073973565,0.00003962778,0.00041165345,0.000073509036,0.97607565,0.007232444,0.010907066,0.0004665161,0.000024303092],"about_ca_topic_score_codex":0.0010787772,"about_ca_topic_score_gemma":0.0010873299,"teacher_disagreement_score":0.0027345677,"about_ca_system_score_codex":0.0004296686,"about_ca_system_score_gemma":0.00033126917,"threshold_uncertainty_score":0.008962929},"labels":[],"label_agreement":null},{"id":"W2292231671","doi":"10.1007/s10115-015-0906-8","title":"Managing dimensionality in data privacy anonymization","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","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 Calgary","funders":"","keywords":"Curse of dimensionality; Computer science; Data mining; Closeness; Process (computing); Cluster analysis; Variety (cybernetics); Machine learning; Mathematics; Artificial intelligence","score_opus":0.07068525535148379,"score_gpt":0.30138513635749714,"score_spread":0.23069988100601335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2292231671","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":[0.027923638,0.0015890286,0.956005,0.0076364023,0.00022320499,0.00013747944,0.0003700867,0.00032570085,0.0057895686],"genre_scores_gemma":[0.74645406,0.002123264,0.24583343,0.0009601437,0.00049760746,0.00022216648,0.00046103421,0.00011234534,0.0033360112],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9721124,0.015056526,0.0018475046,0.0027803038,0.006862339,0.0013408236],"domain_scores_gemma":[0.93236655,0.0266008,0.0038805113,0.032895286,0.003467994,0.00078892417],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015641432,0.000640392,0.0022174446,0.0021868038,0.0040856013,0.009264796,0.0022987123,0.0026097042,0.0016306977],"category_scores_gemma":[0.060578812,0.0009735006,0.001506044,0.004840464,0.0076389266,0.023642613,0.011315108,0.005955392,0.00047835245],"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":[0.00029020777,0.00014819718,0.0028217693,0.00026621134,0.00014101628,0.0002136536,0.0014872849,0.058904406,0.0019438971,0.8255437,0.0056082457,0.10263144],"study_design_scores_gemma":[0.000022896209,0.000037710033,0.00040018297,0.00007066378,0.00006223562,0.00027375817,0.00060791615,0.14038363,0.00473957,0.84283704,0.01052421,0.000040207062],"about_ca_topic_score_codex":0.0014354311,"about_ca_topic_score_gemma":0.0011857174,"teacher_disagreement_score":0.015641432,"about_ca_system_score_codex":0.002830898,"about_ca_system_score_gemma":0.0041924343,"threshold_uncertainty_score":0.082720816},"labels":[],"label_agreement":null},{"id":"W2293288274","doi":"10.1007/s10115-015-0868-x","title":"Models of time series with time granulation","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Time Series Analysis and Forecasting","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":"University of Alberta","funders":"King Abdulaziz University","keywords":"Piecewise; Series (stratigraphy); Representation (politics); Algorithm; Mathematics; Classification of discontinuities; Particle swarm optimization; Segmentation; Time series; Approximation error; Mathematical optimization; Computer science; Artificial intelligence; Statistics; Mathematical analysis","score_opus":0.007936365335809168,"score_gpt":0.1822148914027256,"score_spread":0.17427852606691643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2293288274","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":[0.04958126,0.0010450342,0.9441139,0.00086206273,0.00020881696,0.000051603765,0.00031528404,0.00046166818,0.0033604281],"genre_scores_gemma":[0.912993,0.0015034558,0.07387873,0.0001885069,0.00030723953,0.00021864475,0.0004646841,0.0001829471,0.010262647],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99934775,0.00020588422,0.00006529706,0.00013468882,0.00015626515,0.00009012094],"domain_scores_gemma":[0.99455404,0.0038558592,0.00069893803,0.0004290458,0.00033034425,0.00013180057],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020092193,0.0007691159,0.001552221,0.0011474524,0.00048244555,0.002229833,0.0019583232,0.0020821164,0.002461529],"category_scores_gemma":[0.00966398,0.0007804456,0.001614414,0.0016184296,0.0017473074,0.0032464634,0.0014792898,0.0017256308,0.00040032476],"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":[0.000047118127,0.000019587995,0.0006036413,0.000056685956,0.00004938973,0.00008633402,0.0000805385,0.8893215,0.00054713496,0.10194798,0.00052565924,0.0067144],"study_design_scores_gemma":[0.000004206668,0.0000044829567,0.00007019644,0.0000028987956,0.0000072572357,0.0000100443585,0.0000022479637,0.98161066,0.00007479266,0.018005544,0.00020356652,0.000004157215],"about_ca_topic_score_codex":0.006945663,"about_ca_topic_score_gemma":0.002748729,"teacher_disagreement_score":0.006945663,"about_ca_system_score_codex":0.0012377424,"about_ca_system_score_gemma":0.0007914742,"threshold_uncertainty_score":0.013810456},"labels":[],"label_agreement":null},{"id":"W2337195508","doi":"10.1007/s10115-016-0939-7","title":"Correlation analysis techniques for uncertain time series","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Time Series Analysis and Forecasting","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":"Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada; Concordia University","keywords":"Multiset; Computer science; Timestamp; Autocorrelation; Time series; Probability density function; Probabilistic logic; Normalization (sociology); Data mining; Pruning; Correlation; Random variable; Computation; Algorithm; Artificial intelligence; Machine learning; Mathematics; Statistics","score_opus":0.009764429531435057,"score_gpt":0.2261206984437926,"score_spread":0.21635626891235754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2337195508","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":[0.0039289226,0.0008056012,0.9943936,0.0000722891,0.00003519691,0.000013439823,0.000043723467,0.00014617723,0.0005610578],"genre_scores_gemma":[0.40410396,0.0047921925,0.584713,0.00015119306,0.0006383776,0.00025631365,0.00076219277,0.00032604783,0.0042566564],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99770314,0.00078292267,0.00019661052,0.0003115612,0.0008652233,0.00014059644],"domain_scores_gemma":[0.98926824,0.007771944,0.0007713231,0.0009448835,0.0011341716,0.00010945626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003340247,0.0008486996,0.0012535499,0.003268367,0.00065866887,0.0014236078,0.001164515,0.00077984785,0.002251076],"category_scores_gemma":[0.016826794,0.0005466408,0.0015095745,0.0044282223,0.00069721934,0.0021085998,0.0011466892,0.0020563449,0.0006963056],"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":[0.0002735666,0.00014169404,0.003793316,0.00045347234,0.00058068184,0.0004488176,0.000270875,0.2978498,0.007991578,0.17380859,0.0049512,0.5094365],"study_design_scores_gemma":[0.000007760677,0.000035970257,0.000853868,0.000033343083,0.000074662,0.00010336504,0.000026153175,0.9499107,0.0020952371,0.04430988,0.002527858,0.000021106393],"about_ca_topic_score_codex":0.0027294816,"about_ca_topic_score_gemma":0.0020820384,"teacher_disagreement_score":0.003340247,"about_ca_system_score_codex":0.0006259928,"about_ca_system_score_gemma":0.0013025606,"threshold_uncertainty_score":0.017665088},"labels":[],"label_agreement":null},{"id":"W2513061279","doi":"10.1007/s10115-016-0986-0","title":"EFIM: a fast and memory efficient algorithm for high-utility itemset mining","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":239,"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 Moncton","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Computer science; Key (lock); Data mining; Projection (relational algebra); Database transaction; Task (project management); Tree (set theory); Algorithm; High memory; Database; Mathematics; Parallel computing","score_opus":0.012865336366108314,"score_gpt":0.2381222319819108,"score_spread":0.22525689561580248,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2513061279","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":[0.018990753,0.0018868919,0.95252657,0.00041821328,0.00024271524,0.0004443645,0.003413398,0.019902842,0.0021742387],"genre_scores_gemma":[0.04341999,0.0003351131,0.94803405,0.00018689065,0.00007432727,0.00033708205,0.004750748,0.00042981916,0.0024318912],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99858737,0.00026443895,0.00022292086,0.00028160575,0.00047898185,0.00016464829],"domain_scores_gemma":[0.99723285,0.0013730223,0.00015451336,0.0005448204,0.00058149354,0.00011332957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002128504,0.0018484956,0.0020001745,0.0054847547,0.0013023472,0.0019326555,0.0044230707,0.0019505444,0.008165998],"category_scores_gemma":[0.010128777,0.00094134064,0.0014245124,0.0072214874,0.00048040575,0.0039431923,0.002838069,0.0016915842,0.0045075133],"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":[0.0008522374,0.00031287275,0.0029290074,0.00046859443,0.00018869963,0.00023511435,0.00011924483,0.016619638,0.0039351233,0.0045986027,0.030191928,0.93954897],"study_design_scores_gemma":[0.0004705612,0.00041781634,0.0023246922,0.00020411193,0.00019815628,0.0013262179,0.00020574892,0.893647,0.019048937,0.04372665,0.03832556,0.000104590144],"about_ca_topic_score_codex":0.0032487232,"about_ca_topic_score_gemma":0.005698227,"teacher_disagreement_score":0.008165998,"about_ca_system_score_codex":0.0007999422,"about_ca_system_score_gemma":0.0022958629,"threshold_uncertainty_score":0.027317941},"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":"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":"W2784269545","doi":"10.1007/s10115-017-1147-9","title":"Localized user-driven topic discovery via boosted ensemble of nonnegative matrix factorization","year":2018,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Text Analysis Techniques","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 Waterloo","funders":"National Research Foundation of Korea","keywords":"Matrix decomposition; Computer science; Factorization; Matrix (chemical analysis); Non-negative matrix factorization; Mathematics; Artificial intelligence; Algorithm; Physics","score_opus":0.009477174681492609,"score_gpt":0.27222000813632263,"score_spread":0.26274283345483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2784269545","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":[0.026083862,0.0013050697,0.9687497,0.0004799293,0.0003150174,0.00009856149,0.00047086593,0.0014632125,0.0010338527],"genre_scores_gemma":[0.5172196,0.0013075038,0.46705967,0.000573029,0.0014297295,0.00039342412,0.004744101,0.00041598317,0.0068571116],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972031,0.001004944,0.00014272452,0.0007511267,0.00058541447,0.00031266618],"domain_scores_gemma":[0.99375665,0.0035376637,0.00033690146,0.0006434082,0.0013773915,0.00034794872],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032289773,0.00179771,0.0025947369,0.0025829573,0.0013075193,0.0018941684,0.0026271609,0.0020872282,0.002144259],"category_scores_gemma":[0.009759893,0.00075553556,0.0020301696,0.0026097416,0.0007990813,0.0037012007,0.002373175,0.002362843,0.002109441],"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":[0.002054002,0.0012876759,0.007850456,0.00070858997,0.0007308743,0.0004143031,0.0009896205,0.12702367,0.04567176,0.017101744,0.03632962,0.7598377],"study_design_scores_gemma":[0.000025516436,0.000060221606,0.00046105238,0.000013762197,0.000064433356,0.000073199124,0.000045821787,0.98703045,0.002513256,0.008176457,0.0015110135,0.000024798217],"about_ca_topic_score_codex":0.005197224,"about_ca_topic_score_gemma":0.010697315,"teacher_disagreement_score":0.005197224,"about_ca_system_score_codex":0.00058626255,"about_ca_system_score_gemma":0.0018017602,"threshold_uncertainty_score":0.017076671},"labels":[],"label_agreement":null},{"id":"W2883420210","doi":"10.1007/s10115-018-1235-5","title":"Overview of the crowdsourcing process","year":2018,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Mobile Crowdsensing and Crowdsourcing","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":"University of Waterloo","funders":"","keywords":"Crowdsourcing; Computer science; Process (computing); Task (project management); Quality (philosophy); Data science; Human intelligence; Incentive; Artificial intelligence; Machine learning; World Wide Web; Engineering","score_opus":0.020536533582253488,"score_gpt":0.26745217543955035,"score_spread":0.24691564185729686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2883420210","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":[0.003125776,0.0192296,0.89590496,0.007920758,0.0010409635,0.0012064497,0.0022575762,0.0015676089,0.06774631],"genre_scores_gemma":[0.147669,0.052699648,0.70699054,0.0026314831,0.0048007974,0.0026366767,0.0052341474,0.0010902571,0.07624749],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99329424,0.0018734122,0.00051099045,0.0014440047,0.0024333652,0.0004441176],"domain_scores_gemma":[0.99348277,0.0030275837,0.00031993457,0.0012128573,0.0016183811,0.0003385098],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058136946,0.0017076074,0.0018642757,0.0058768326,0.003424972,0.009762814,0.003526879,0.0052323597,0.016320534],"category_scores_gemma":[0.011557774,0.0019197463,0.002048744,0.0072994465,0.003190617,0.006882859,0.0072239223,0.003967068,0.010931197],"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":[0.0001213275,0.00015255266,0.0011848971,0.0021954924,0.000111099434,0.00029469078,0.0011502849,0.023420185,0.003242262,0.707544,0.030646566,0.22993651],"study_design_scores_gemma":[0.000034448374,0.000055483055,0.0010331551,0.0006560921,0.00006169404,0.0003673113,0.00030862924,0.060583256,0.0025958833,0.43740022,0.49678388,0.000119973774],"about_ca_topic_score_codex":0.01603959,"about_ca_topic_score_gemma":0.009058144,"teacher_disagreement_score":0.016320534,"about_ca_system_score_codex":0.004133658,"about_ca_system_score_gemma":0.0068963687,"threshold_uncertainty_score":0.054597616},"labels":[],"label_agreement":null},{"id":"W2889435016","doi":"10.1007/s10115-018-1250-6","title":"Categorizing relational facts from the web with fuzzy rough sets","year":2018,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Rough Sets and Fuzzy Logic","field":"Computer Science","cited_by":7,"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 Winnipeg","funders":"Natural Sciences and Engineering Research Council of Canada; Türkiye Bilimsel ve Teknolojik Araştırma Kurumu; University of Winnipeg","keywords":"Rough set; Artificial intelligence; Computer science; Set (abstract data type); Noun; Fuzzy logic; Statistical relational learning; Fuzzy set; Machine learning; Natural language processing; Mathematics; Relational database; Data mining","score_opus":0.02216141262980433,"score_gpt":0.2288226861964595,"score_spread":0.20666127356665517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889435016","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":[0.30914968,0.003195754,0.6640818,0.0013493001,0.0002790026,0.0006453369,0.004855751,0.0013293887,0.01511399],"genre_scores_gemma":[0.7654785,0.001639136,0.22595698,0.00013333141,0.00013988905,0.00022254448,0.0052079586,0.000053604632,0.0011680253],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99648654,0.0008049485,0.0005849988,0.00046984418,0.0014431347,0.00021041285],"domain_scores_gemma":[0.9943891,0.0031967768,0.00056718173,0.00083291647,0.0008619539,0.00015211459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021920614,0.00062595966,0.0009049109,0.017318474,0.0010060223,0.005772765,0.0008366478,0.0010175277,0.0024002278],"category_scores_gemma":[0.01532085,0.0003856827,0.0016016321,0.006945213,0.0013229443,0.008929916,0.0020681452,0.0014424697,0.00072005397],"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":[0.00094986736,0.00071207614,0.048383202,0.0017233493,0.0010842348,0.003938478,0.0069197835,0.06447446,0.011170914,0.20289484,0.008760061,0.6489888],"study_design_scores_gemma":[0.00007061094,0.00021580918,0.030351745,0.00089597097,0.0008663906,0.0009770594,0.006350554,0.33937997,0.0093688,0.5862659,0.025041526,0.00021566497],"about_ca_topic_score_codex":0.004274162,"about_ca_topic_score_gemma":0.004000715,"teacher_disagreement_score":0.017318474,"about_ca_system_score_codex":0.0011073548,"about_ca_system_score_gemma":0.0008315428,"threshold_uncertainty_score":0.011592865},"labels":[],"label_agreement":null},{"id":"W2906685695","doi":"10.1007/s10115-018-1278-7","title":"Combining semantic and term frequency similarities for text clustering","year":2019,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Text and Document Classification Technologies","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":"Dalhousie University","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":"Cluster analysis; Semantic similarity; Computer science; Document clustering; Similarity (geometry); Measure (data warehouse); Information retrieval; Similarity measure; Natural language processing; Term (time); Artificial intelligence; Word (group theory); Data mining; Mathematics","score_opus":0.01570786528100131,"score_gpt":0.24286185679499328,"score_spread":0.22715399151399196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2906685695","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":[0.123882644,0.0036071506,0.86066514,0.0003190131,0.00040587352,0.00032091964,0.0021774264,0.0033271585,0.0052945637],"genre_scores_gemma":[0.56906706,0.0013717745,0.41879496,0.000112664195,0.0006654824,0.00032991607,0.005411552,0.00042164724,0.0038249388],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970804,0.0005349146,0.00035498446,0.00052632333,0.0013171491,0.00018626063],"domain_scores_gemma":[0.9957747,0.0017542881,0.00034760684,0.00048206936,0.0015053789,0.00013603213],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00211578,0.00063412817,0.0012625848,0.014377883,0.0011301856,0.0021374961,0.0010457116,0.0012252686,0.0026668238],"category_scores_gemma":[0.009617088,0.00028983626,0.001536634,0.0117167495,0.0005454635,0.003095946,0.0011834403,0.00066196517,0.00218431],"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":[0.000933483,0.00042562053,0.010976605,0.000657147,0.0006392734,0.0001525617,0.0003585493,0.015423425,0.03027804,0.0074733747,0.0064370353,0.92624485],"study_design_scores_gemma":[0.00021910647,0.00067473133,0.03179697,0.00021953083,0.0012676757,0.0013484521,0.0008799106,0.84382343,0.03290542,0.06828828,0.01831172,0.00026477472],"about_ca_topic_score_codex":0.0032486923,"about_ca_topic_score_gemma":0.0054411558,"teacher_disagreement_score":0.014377883,"about_ca_system_score_codex":0.00065730495,"about_ca_system_score_gemma":0.0013432731,"threshold_uncertainty_score":0.011189461},"labels":[],"label_agreement":null},{"id":"W2979871440","doi":"10.1007/s10115-019-01405-7","title":"$${\\textsf {SecDM}}$$: privacy-preserving data outsourcing framework with differential privacy","year":2019,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":13,"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; University of Manitoba; McGill University","funders":"","keywords":"Computer science; Cloud computing; Homomorphic encryption; Encryption; Differential privacy; Information privacy; Outsourcing; Scalability; Computer security; Service provider; Cryptosystem; Database; Data mining; Service (business); Business","score_opus":0.01773065424694377,"score_gpt":0.24963610835367206,"score_spread":0.2319054541067283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979871440","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":[0.0061459383,0.00039035047,0.9715802,0.0025355928,0.00016562128,0.00012877311,0.0004104868,0.001076323,0.01756672],"genre_scores_gemma":[0.5754619,0.001451655,0.3825494,0.0012626549,0.00028457283,0.0003643726,0.00083232054,0.00049090665,0.037302226],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.997267,0.0006421304,0.00010048451,0.0003246205,0.0011259313,0.00053984585],"domain_scores_gemma":[0.99891615,0.00018604312,0.00007539415,0.0004941,0.00021072563,0.000117593125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023889078,0.00064664474,0.00088333286,0.0010251836,0.0016955232,0.0039007156,0.0030718397,0.0013570731,0.009276581],"category_scores_gemma":[0.002183534,0.00051780697,0.0009339511,0.002472636,0.0025691285,0.0065380693,0.00725035,0.0037972939,0.0032489996],"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":[0.0000746729,0.000058607635,0.00007672725,0.0000624556,0.000010967233,0.00010148224,0.000097960605,0.0054799127,0.00151983,0.95403063,0.009958215,0.028528713],"study_design_scores_gemma":[0.000047417954,0.000054152282,0.00013526836,0.00004031562,0.000026842039,0.00048049277,0.00011371343,0.12327073,0.006675471,0.81652975,0.052570757,0.000055185326],"about_ca_topic_score_codex":0.002280348,"about_ca_topic_score_gemma":0.0022870807,"teacher_disagreement_score":0.009276581,"about_ca_system_score_codex":0.0022938247,"about_ca_system_score_gemma":0.0031417815,"threshold_uncertainty_score":0.031033218},"labels":[],"label_agreement":null},{"id":"W3027993156","doi":"10.1007/s10115-020-01472-1","title":"Survival neural networks for time-to-event prediction in longitudinal study","year":2020,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":14,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier Universitaire de Sherbrooke; Université de Sherbrooke","funders":"National Institute of Justice; Fonds de Recherche du Québec - Santé; Canadian Institutes of Health Research; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; National Science Foundation","keywords":"Censoring (clinical trials); Covariate; Survival analysis; Artificial neural network; Computer science; Event (particle physics); Accelerated failure time model; Prognostics; Brier score; Artificial intelligence; Machine learning; Statistics; Data mining; Mathematics","score_opus":0.09601427600990584,"score_gpt":0.35848118935826145,"score_spread":0.2624669133483556,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3027993156","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":[0.053106226,0.0046797185,0.9374929,0.002105991,0.00021079752,0.000088322035,0.0009075765,0.00048573088,0.0009226718],"genre_scores_gemma":[0.863917,0.00391881,0.12216696,0.0003886597,0.0005145321,0.00052139227,0.0018922715,0.00007992577,0.00660053],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981621,0.0010738254,0.00014155035,0.00035748858,0.00015398166,0.00011107184],"domain_scores_gemma":[0.9587115,0.03794772,0.0010400368,0.0010615853,0.00092226197,0.00031681388],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011397324,0.0008347084,0.0018412075,0.0021158806,0.00069328526,0.0012757148,0.0022485692,0.0020778687,0.0032127025],"category_scores_gemma":[0.040954646,0.00066566205,0.0009780038,0.0020087527,0.0011041973,0.0024142023,0.0013540135,0.0029143307,0.00058605167],"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":[0.0004607086,0.00024489645,0.01570112,0.0002994692,0.0005206149,0.00019110861,0.00021334941,0.739401,0.00033090607,0.05796012,0.0041105063,0.18056619],"study_design_scores_gemma":[0.000010435599,0.000015869076,0.0007448884,0.000028317434,0.000036514313,0.000013794946,0.000013279838,0.96443284,0.00007291057,0.034383833,0.00023867343,0.000008632525],"about_ca_topic_score_codex":0.014208613,"about_ca_topic_score_gemma":0.011783677,"teacher_disagreement_score":0.014208613,"about_ca_system_score_codex":0.0013464707,"about_ca_system_score_gemma":0.0013882476,"threshold_uncertainty_score":0.060275495},"labels":[],"label_agreement":null},{"id":"W3096565974","doi":"10.1007/s10115-020-01521-9","title":"CANE: community-aware network embedding via adversarial training","year":2020,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Graph Neural Networks","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":"Novelis (Canada)","funders":"","keywords":"Computer science; Node (physics); Discriminative model; Pairwise comparison; Embedding; Adversarial system; Machine learning; Representation (politics); Feature learning; Data mining; Artificial intelligence; Community structure; Theoretical computer science; Mathematics","score_opus":0.0314213780843983,"score_gpt":0.2559193188299602,"score_spread":0.22449794074556192,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3096565974","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":[0.008892289,0.0004938703,0.97842914,0.00067057484,0.00029660302,0.00017727418,0.0008907954,0.0064992458,0.003650283],"genre_scores_gemma":[0.38711107,0.0006046887,0.57653797,0.0013001015,0.0003367198,0.0007096621,0.0061487607,0.0020132228,0.025237834],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992988,0.0002476155,0.00001767821,0.00018444128,0.00016951888,0.00008192426],"domain_scores_gemma":[0.9978848,0.0010170626,0.000102661164,0.00057194196,0.00028247182,0.00014094615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013440894,0.0018754194,0.0013879661,0.0013334157,0.00080165983,0.0009579289,0.0035552615,0.0033437854,0.009324199],"category_scores_gemma":[0.006385839,0.00075459405,0.001041911,0.001303993,0.0010701184,0.0028082298,0.0034760316,0.004111093,0.0038389587],"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":[0.00037595333,0.00036878744,0.0008961227,0.0002804911,0.00022920546,0.0002497818,0.00011786868,0.6152396,0.004554525,0.03283675,0.05879717,0.2860537],"study_design_scores_gemma":[0.000015594402,0.00001869482,0.00004987378,0.000010306061,0.000007528077,0.000028175433,0.0000080882555,0.9857538,0.0006550082,0.011856069,0.0015905873,0.0000062986715],"about_ca_topic_score_codex":0.0068390705,"about_ca_topic_score_gemma":0.0135573475,"teacher_disagreement_score":0.009324199,"about_ca_system_score_codex":0.0008890697,"about_ca_system_score_gemma":0.0010942181,"threshold_uncertainty_score":0.031192541},"labels":[],"label_agreement":null},{"id":"W3100780777","doi":"10.1007/s10115-017-1033-5","title":"Measuring in-network node similarity based on neighborhoods: a unified parametric approach","year":2017,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","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":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Similarity (geometry); Node (physics); Parametric statistics; Data mining; Computer science; Mathematics; Artificial intelligence; Theoretical computer science; Statistics; Engineering","score_opus":0.035922509109027265,"score_gpt":0.26357355829460927,"score_spread":0.227651049185582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3100780777","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":[0.076360546,0.00023341835,0.9211592,0.000093128525,0.000011670517,0.00004997253,0.00012317163,0.00012761369,0.0018412581],"genre_scores_gemma":[0.8821823,0.00029946273,0.11632556,0.000021914375,0.00004712187,0.00012759882,0.00014751962,0.000034687055,0.0008138448],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976427,0.0009113431,0.0001245274,0.0005464425,0.0006574815,0.00011755595],"domain_scores_gemma":[0.99193233,0.004667685,0.0011959384,0.0011930865,0.00079170737,0.00021930838],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025442785,0.0005264485,0.0012246054,0.0043475307,0.00067375856,0.0024313084,0.0019041472,0.0011282332,0.0011977758],"category_scores_gemma":[0.017275874,0.0004043645,0.00070595636,0.004184264,0.0015806601,0.0066595087,0.002438153,0.0007436908,0.00021982339],"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":[0.00023208864,0.00034238742,0.021575635,0.00031559748,0.00035489196,0.00025680388,0.00090764184,0.500663,0.01110138,0.2360223,0.0011746536,0.22705366],"study_design_scores_gemma":[0.000004660313,0.00007278647,0.0042271926,0.00001747335,0.000056221055,0.00013765538,0.0001874911,0.92623496,0.0011649912,0.06727456,0.0005983727,0.000023546621],"about_ca_topic_score_codex":0.001721305,"about_ca_topic_score_gemma":0.001441065,"teacher_disagreement_score":0.0043475307,"about_ca_system_score_codex":0.0007934871,"about_ca_system_score_gemma":0.0006539311,"threshold_uncertainty_score":0.013455629},"labels":[],"label_agreement":null},{"id":"W3195438473","doi":"10.1007/s10115-021-01605-0","title":"Model complexity of deep learning: a survey","year":2021,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":388,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Deep learning; Artificial intelligence; Generalization; Machine learning; Model selection; Game complexity; Process (computing); Computational complexity theory; Worst-case complexity; Algorithm; Mathematics","score_opus":0.041583257627835754,"score_gpt":0.2735118115492377,"score_spread":0.23192855392140194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3195438473","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":[0.01982539,0.13453028,0.8135599,0.015967656,0.0006585559,0.000075305,0.0013265049,0.0005445912,0.013511794],"genre_scores_gemma":[0.50748163,0.23614195,0.23323867,0.0036862122,0.005305241,0.0005302477,0.0035370141,0.00068433414,0.009394791],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976375,0.0007653854,0.00019879668,0.00043657742,0.0008244429,0.00013728467],"domain_scores_gemma":[0.97972816,0.016834885,0.000619177,0.0012761129,0.0012781748,0.000263439],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043166056,0.0016242153,0.0028340763,0.0018991911,0.0006552453,0.004785905,0.0033019579,0.0023068497,0.004022646],"category_scores_gemma":[0.023436924,0.0012096686,0.0014000755,0.0028214925,0.0019071539,0.010755478,0.0031462377,0.004528206,0.0008751925],"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":[0.00018701277,0.00021146123,0.003924327,0.002181076,0.00026915665,0.000110381436,0.00012613856,0.23288883,0.0005905586,0.4925283,0.01610288,0.2508798],"study_design_scores_gemma":[0.000013899932,0.000045292574,0.00043919106,0.00020218486,0.000040012892,0.00008750469,0.000039617687,0.38046268,0.00044726304,0.6098202,0.008378144,0.000024026298],"about_ca_topic_score_codex":0.0039192326,"about_ca_topic_score_gemma":0.0031941875,"teacher_disagreement_score":0.004785905,"about_ca_system_score_codex":0.0036357706,"about_ca_system_score_gemma":0.0025868472,"threshold_uncertainty_score":0.026379466},"labels":[],"label_agreement":null},{"id":"W3215783632","doi":"10.1007/s10115-021-01625-w","title":"An integrated framework for diagnosing process faults with incomplete features","year":2021,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Fault Detection and Control Systems","field":"Engineering","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":"Missing data; Imputation (statistics); Pooling; Dimensionality reduction; Data mining; Computer science; Curse of dimensionality; Principal component analysis; Artificial intelligence; Pattern recognition (psychology); Machine learning","score_opus":0.007338811282059134,"score_gpt":0.2464068347811218,"score_spread":0.23906802349906267,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3215783632","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":[0.002249019,0.000107649415,0.996077,0.000057036024,0.000011812308,0.000040631752,0.00007310173,0.0008953853,0.0004883186],"genre_scores_gemma":[0.18067579,0.00023927975,0.8168383,0.00008716534,0.00004767429,0.00014685906,0.0004441646,0.00012241698,0.0013983046],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99810624,0.00024357974,0.00014373576,0.00046148678,0.00082916726,0.00021573754],"domain_scores_gemma":[0.9981421,0.0006232063,0.00020960066,0.00035740645,0.0005616394,0.0001059952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018325644,0.0014394411,0.0018095796,0.002412881,0.00076691963,0.003424033,0.0031939372,0.0018152058,0.0021478042],"category_scores_gemma":[0.0039133215,0.00066664605,0.0019524002,0.0014930555,0.0011781966,0.002879331,0.002486823,0.0014980856,0.0006327327],"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":[0.00037909343,0.00034994705,0.0030407608,0.00047868845,0.0004768836,0.0013353648,0.00046080217,0.56658,0.018476667,0.11625087,0.004107426,0.2880634],"study_design_scores_gemma":[0.00001980114,0.00005530245,0.0002704685,0.00003526701,0.00010769673,0.00013116137,0.000039948285,0.9616623,0.003574617,0.031476047,0.0026045963,0.000022842762],"about_ca_topic_score_codex":0.012562535,"about_ca_topic_score_gemma":0.016251087,"teacher_disagreement_score":0.012562535,"about_ca_system_score_codex":0.0014446485,"about_ca_system_score_gemma":0.0033087928,"threshold_uncertainty_score":0.024978817},"labels":[],"label_agreement":null},{"id":"W4283720480","doi":"10.1007/s10115-022-01703-7","title":"BertHANK: hierarchical attention networks with enhanced knowledge and pre-trained model for answer selection","year":2022,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","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":"York University","funders":"","keywords":"Selection (genetic algorithm); Computer science; Artificial intelligence; Machine learning; Model selection","score_opus":0.011614359370299665,"score_gpt":0.2366141620126121,"score_spread":0.22499980264231242,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283720480","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":[0.021390704,0.0017069129,0.92851824,0.0010276764,0.00045552463,0.00026761153,0.0023749839,0.039043337,0.0052149906],"genre_scores_gemma":[0.35670865,0.0007468638,0.6031391,0.0009444948,0.00032142006,0.0006452149,0.0072141383,0.0024442824,0.027835814],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99921083,0.00021709803,0.0000310509,0.00026633116,0.00015580465,0.000118830874],"domain_scores_gemma":[0.9985726,0.00084196276,0.000047173355,0.0002039564,0.00024931776,0.00008511741],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016127425,0.0015780132,0.0011772801,0.0013431228,0.0008725575,0.0013050479,0.0031351934,0.0030984713,0.013434308],"category_scores_gemma":[0.0061694155,0.0010293905,0.0010421928,0.0013672658,0.00055154366,0.004161507,0.0029644084,0.003082603,0.0052142455],"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":[0.0010751999,0.0003655997,0.00093218003,0.00035065116,0.0002677906,0.0002190921,0.00025868468,0.09616026,0.0131795015,0.013227494,0.07606995,0.7978936],"study_design_scores_gemma":[0.00006224343,0.00003503421,0.00022232335,0.000013477849,0.000039028135,0.000024828409,0.000018972856,0.97988635,0.0040190043,0.012904226,0.0027554925,0.000019029449],"about_ca_topic_score_codex":0.025516426,"about_ca_topic_score_gemma":0.03904148,"teacher_disagreement_score":0.025516426,"about_ca_system_score_codex":0.0014902877,"about_ca_system_score_gemma":0.0015698806,"threshold_uncertainty_score":0.05073577},"labels":[],"label_agreement":null},{"id":"W4310276756","doi":"10.1007/s10115-022-01792-4","title":"DNETC: dynamic network embedding preserving both triadic closure evolution and community structures","year":2022,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","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":"Université de Montréal","funders":"Sichuan Province Science and Technology Support Program; National Natural Science Foundation of China","keywords":"Computer science; Embedding; Closure (psychology); Dynamic network analysis; Focus (optics); Theoretical computer science; Social network (sociolinguistics); Artificial intelligence; Algorithm; Topology (electrical circuits); Mathematics","score_opus":0.0101272855086007,"score_gpt":0.2636848183653573,"score_spread":0.2535575328567566,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310276756","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":[0.014409776,0.00027798754,0.9739057,0.00041783997,0.00026092454,0.000116178126,0.0016718765,0.0029767947,0.0059629823],"genre_scores_gemma":[0.3263229,0.0007131382,0.63709,0.0004173148,0.00022093,0.00042857346,0.008535997,0.0019764688,0.024294721],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99919635,0.00015758829,0.00003363649,0.0002556808,0.0002874803,0.00006942681],"domain_scores_gemma":[0.99742633,0.00068440294,0.00018383023,0.0009709143,0.00050119666,0.00023337081],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010017233,0.00096587965,0.00096029387,0.0017518688,0.0008446062,0.0014832343,0.001933511,0.001048325,0.005375798],"category_scores_gemma":[0.0059713796,0.00041204618,0.00069571525,0.0019321828,0.0010152952,0.0038535695,0.0029496483,0.0023106807,0.0015106386],"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":[0.000490424,0.00035259817,0.0010796158,0.00050477864,0.0001302034,0.00028531704,0.00029361024,0.19966595,0.014139804,0.2776145,0.056874935,0.44856828],"study_design_scores_gemma":[0.000035231777,0.000043764278,0.00024321824,0.00003300569,0.000022031762,0.00012799197,0.000055079436,0.843977,0.005207943,0.13455202,0.015681304,0.000021408176],"about_ca_topic_score_codex":0.0068651503,"about_ca_topic_score_gemma":0.011006635,"teacher_disagreement_score":0.0068651503,"about_ca_system_score_codex":0.001237869,"about_ca_system_score_gemma":0.0015522181,"threshold_uncertainty_score":0.017983854},"labels":[],"label_agreement":null},{"id":"W4327590965","doi":"10.1007/s10115-023-01850-5","title":"Towards intelligent database systems using clusters of SQL transactions","year":2023,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Database Systems and Queries","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":"Huawei Technologies (Canada)","funders":"","keywords":"Computer science; Online transaction processing; Troubleshooting; Database transaction; Database; DBSCAN; Credit card; Cluster analysis; Transaction processing; Data mining; SQL; Overhead (engineering); Artificial intelligence; Operating system; World Wide Web","score_opus":0.036835463230279156,"score_gpt":0.2875389323398121,"score_spread":0.25070346910953295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4327590965","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":[0.0143580325,0.00038332015,0.9773766,0.00035689538,0.00008555419,0.00016297259,0.00013036527,0.005610904,0.001535406],"genre_scores_gemma":[0.11731863,0.00051619275,0.8766449,0.0002113275,0.00009397063,0.00016490283,0.0006434356,0.00041583952,0.0039908127],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973104,0.0007212698,0.00030689125,0.0005052381,0.0009931062,0.00016321536],"domain_scores_gemma":[0.9937495,0.0017155465,0.00032385392,0.002376423,0.0015403334,0.00029434246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004038421,0.00089885015,0.0016897154,0.0016994026,0.00095838535,0.004647743,0.0028695317,0.001205182,0.0037172963],"category_scores_gemma":[0.009229136,0.0014357659,0.0011941686,0.0030426423,0.0012705203,0.0077427193,0.0032768317,0.002396765,0.0026078508],"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":[0.0013575379,0.00063052156,0.00683278,0.00069106615,0.00052242324,0.0003164846,0.0012028501,0.12833638,0.036666695,0.15666659,0.026399875,0.6403768],"study_design_scores_gemma":[0.00015599422,0.00027415407,0.0010182476,0.00008156579,0.00021059328,0.00020402586,0.00039226943,0.75755274,0.024290511,0.18526591,0.030485315,0.000068747475],"about_ca_topic_score_codex":0.0028180466,"about_ca_topic_score_gemma":0.004285206,"teacher_disagreement_score":0.004647743,"about_ca_system_score_codex":0.0008357335,"about_ca_system_score_gemma":0.001607596,"threshold_uncertainty_score":0.021357477},"labels":[],"label_agreement":null},{"id":"W4362589576","doi":"10.1007/s10115-023-01852-3","title":"Mining frequent generators and closures in data streams with FGC-Stream","year":2023,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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 à Montréal","funders":"","keywords":"Computer science; Sliding window protocol; Data stream mining; Data mining; Exploit; Association rule learning; Factoring; Data stream; STREAMS; Window (computing); Computer security","score_opus":0.029363750963375318,"score_gpt":0.26678572819741725,"score_spread":0.23742197723404193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362589576","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":[0.2347716,0.0016256642,0.74157935,0.0009055607,0.00030543082,0.00071904395,0.012217961,0.0062748347,0.0016005246],"genre_scores_gemma":[0.5205088,0.00073787203,0.45108843,0.00017541356,0.00033527098,0.0004902874,0.02468143,0.0002568169,0.0017257128],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975858,0.0003067527,0.00029702074,0.0005850206,0.0009914628,0.00023399405],"domain_scores_gemma":[0.9872694,0.0072215796,0.0013049309,0.001361632,0.0022510067,0.0005913974],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032122112,0.0011085037,0.0014044873,0.007083253,0.0011945434,0.0027670944,0.0017513352,0.0012146657,0.0016137034],"category_scores_gemma":[0.01731845,0.0005690866,0.0015266554,0.004317969,0.00083029026,0.0029812611,0.0014186287,0.0014971053,0.00064680865],"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":[0.002539031,0.0012642308,0.11654679,0.0014571601,0.0006523001,0.00570815,0.0013071338,0.105635114,0.016584247,0.036382854,0.028044723,0.68387824],"study_design_scores_gemma":[0.00011488122,0.00020211704,0.005832409,0.0001020176,0.000115485935,0.0013380266,0.00036946073,0.94646114,0.0048271185,0.034041304,0.006561076,0.00003494276],"about_ca_topic_score_codex":0.0043692617,"about_ca_topic_score_gemma":0.0050055915,"teacher_disagreement_score":0.007083253,"about_ca_system_score_codex":0.00065968285,"about_ca_system_score_gemma":0.0024181183,"threshold_uncertainty_score":0.016988039},"labels":[],"label_agreement":null},{"id":"W4393495083","doi":"10.1007/s10115-023-02059-2","title":"Entity linking for English and other languages: a survey","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Natural Language Processing Techniques","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":"Ontario Tech University","funders":"Aston University; UK Research and Innovation","keywords":"Linguistics; Computer science; Philosophy","score_opus":0.01679273697061635,"score_gpt":0.2900945133550044,"score_spread":0.2733017763843881,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393495083","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":[0.030093648,0.7776113,0.08401592,0.0033932591,0.00078881084,0.00020756082,0.0043458804,0.003480224,0.09606334],"genre_scores_gemma":[0.085247256,0.81759363,0.069088675,0.0024648944,0.00086616835,0.00019022294,0.009555182,0.0011315802,0.013862388],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99844956,0.0003279442,0.00025559962,0.00038330382,0.00046154446,0.00012207967],"domain_scores_gemma":[0.9922983,0.0057428577,0.00045529762,0.00044019968,0.0009460681,0.00011729945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031218615,0.0006968872,0.0010591939,0.010379954,0.0006299846,0.0030510416,0.0011793167,0.00090629794,0.008285613],"category_scores_gemma":[0.008048123,0.0005689901,0.00075628294,0.016086292,0.00065167865,0.010054532,0.0017133566,0.0009230351,0.0050911666],"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":[0.000039219718,0.000079195044,0.0033462024,0.004933392,0.000051450013,0.00017784201,0.0006207338,0.00055832835,0.0017398552,0.008344359,0.020435285,0.9596742],"study_design_scores_gemma":[0.0000073612746,0.000058713522,0.0067338343,0.0031647454,0.00010468414,0.0015289149,0.0011273861,0.0021018211,0.0049227104,0.006710496,0.9734865,0.000052847536],"about_ca_topic_score_codex":0.0012589481,"about_ca_topic_score_gemma":0.0011632185,"teacher_disagreement_score":0.010379954,"about_ca_system_score_codex":0.0004450954,"about_ca_system_score_gemma":0.00097063335,"threshold_uncertainty_score":0.027718127},"labels":[],"label_agreement":null},{"id":"W4401314125","doi":"10.1007/s10115-024-02185-5","title":"An efficient approach for incremental erasable utility pattern mining from non-binary data","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","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":"National Research Foundation of Korea","keywords":"Computer science; Data mining; Binary number; Task (project management); Database; Mathematics","score_opus":0.04000251640079476,"score_gpt":0.2932303513575283,"score_spread":0.2532278349567335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401314125","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":[0.018884774,0.00046340656,0.9748174,0.00021228321,0.00006124932,0.0001912493,0.0005337298,0.0038577442,0.0009781953],"genre_scores_gemma":[0.15064165,0.00026301248,0.84436506,0.00011858634,0.00006392386,0.00019286755,0.0012725133,0.00016590235,0.00291656],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985738,0.00019915639,0.00014805862,0.00023485947,0.00071441306,0.00012971014],"domain_scores_gemma":[0.9971834,0.0013367911,0.00015931402,0.00062719325,0.00060416525,0.000089240246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010775052,0.00064324436,0.0010946759,0.0027223572,0.00058916333,0.001618055,0.0023859015,0.0006824392,0.002380143],"category_scores_gemma":[0.0065494273,0.0004804291,0.0009732841,0.0030212426,0.00049608655,0.0022856873,0.0020098477,0.0012516559,0.0012084696],"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":[0.0003474447,0.00022287242,0.0029917364,0.00023261867,0.000089614936,0.00041055444,0.00018825497,0.021596223,0.012542823,0.012645695,0.0064409296,0.9422913],"study_design_scores_gemma":[0.000064218235,0.00012950363,0.0012369042,0.000032525724,0.00006688264,0.00068183674,0.000114333314,0.94638443,0.011723514,0.032176327,0.0073551866,0.000034208133],"about_ca_topic_score_codex":0.0028157134,"about_ca_topic_score_gemma":0.006929664,"teacher_disagreement_score":0.0028157134,"about_ca_system_score_codex":0.00042903423,"about_ca_system_score_gemma":0.0016692748,"threshold_uncertainty_score":0.007962346},"labels":[],"label_agreement":null},{"id":"W4401504220","doi":"10.1007/s10115-024-02196-2","title":"MoRGH: movie recommender system using GNNs on heterogeneous graphs","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":12,"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":"Recommender system; Computer science; Artificial intelligence; Information retrieval","score_opus":0.028685631643001173,"score_gpt":0.2700012999917217,"score_spread":0.2413156683487205,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401504220","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":[0.10374404,0.0020974434,0.7975407,0.0012277386,0.0006567385,0.00083651434,0.0108913975,0.071267515,0.011737922],"genre_scores_gemma":[0.39904952,0.00060646643,0.5711455,0.0004682867,0.00016044555,0.0002458765,0.013448357,0.0008714306,0.014004054],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996748,0.000093061884,0.000020076643,0.000096085336,0.00008332219,0.000032784395],"domain_scores_gemma":[0.9990426,0.00033279473,0.000031454343,0.0003235842,0.00020756702,0.000061987594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007773534,0.00071241375,0.000871462,0.0016976625,0.0008208127,0.0009265914,0.0011187893,0.0010618845,0.0058805444],"category_scores_gemma":[0.0035437224,0.00040203024,0.00068152463,0.0014255769,0.00018128166,0.0018140941,0.0010708291,0.00089053944,0.0020404651],"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":[0.0017001741,0.0006961023,0.008776933,0.0006466809,0.0008978506,0.0004968887,0.00022523591,0.17161155,0.01049434,0.009266205,0.12632434,0.6688637],"study_design_scores_gemma":[0.00008004865,0.00006955424,0.00079287746,0.000013075001,0.000060492104,0.000072872404,0.000031907755,0.983706,0.0021677262,0.0050415015,0.007938308,0.000025603224],"about_ca_topic_score_codex":0.027412921,"about_ca_topic_score_gemma":0.055533983,"teacher_disagreement_score":0.027412921,"about_ca_system_score_codex":0.00071951794,"about_ca_system_score_gemma":0.0006751038,"threshold_uncertainty_score":0.05450672},"labels":[],"label_agreement":null},{"id":"W4402322874","doi":"10.1007/s10115-024-02204-5","title":"CAERS-CF: enhancing convolutional autoencoder recommendations through collaborative filtering","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Recommender Systems and Techniques","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 Ottawa","funders":"","keywords":"Autoencoder; Collaborative filtering; Computer science; Artificial intelligence; Convolutional neural network; Pattern recognition (psychology); Information retrieval; Recommender system; Deep learning","score_opus":0.017814532704902904,"score_gpt":0.28284898012906495,"score_spread":0.26503444742416205,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402322874","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":[0.03956294,0.0013228385,0.94965214,0.00039975322,0.0003318318,0.00009122607,0.00044231073,0.0042502563,0.003946647],"genre_scores_gemma":[0.53012305,0.0007893948,0.44863534,0.0005861844,0.0002518958,0.00011200421,0.0011092679,0.0003014749,0.0180913],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991862,0.00017465527,0.000035829675,0.00020305571,0.00031350495,0.00008672497],"domain_scores_gemma":[0.9975153,0.0012570265,0.00009841727,0.0004726868,0.00057326275,0.00008321589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013986742,0.0010387315,0.0010822265,0.0011866513,0.0005110328,0.00097545027,0.0015819955,0.0020781562,0.0040889475],"category_scores_gemma":[0.005814378,0.00051876984,0.00065261277,0.0012464178,0.00043040424,0.0016136704,0.0009834545,0.0017314637,0.0014148416],"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":[0.000556996,0.0005607279,0.003325877,0.00018456751,0.00036548765,0.00019004603,0.00014946035,0.22050184,0.01777168,0.0071579856,0.018557515,0.7306778],"study_design_scores_gemma":[0.00001957542,0.000052761377,0.00042789,0.000010836637,0.000031596486,0.00004100002,0.000008588623,0.9928203,0.0034069077,0.0016358085,0.0015323169,0.0000123040545],"about_ca_topic_score_codex":0.034230646,"about_ca_topic_score_gemma":0.06519826,"teacher_disagreement_score":0.034230646,"about_ca_system_score_codex":0.0006997942,"about_ca_system_score_gemma":0.0011621641,"threshold_uncertainty_score":0.06806278},"labels":[],"label_agreement":null},{"id":"W4403155837","doi":"10.1007/s10115-024-02248-7","title":"Deep-transfer learning inspired natural language processing system for software requirements classification","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software Engineering 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":"Computer science; Artificial intelligence; Transfer of learning; Natural language processing; Natural language; Deep learning; Software; Machine learning; Programming language","score_opus":0.0214657515274903,"score_gpt":0.28490399424704965,"score_spread":0.26343824271955935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403155837","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":[0.09917299,0.0003920372,0.8707159,0.00057623425,0.00012665916,0.00024126169,0.001673524,0.020002393,0.007098961],"genre_scores_gemma":[0.61114985,0.000267145,0.37096277,0.00055475073,0.00005719994,0.00028859475,0.00441062,0.0002669018,0.012042177],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997296,0.000045208155,0.000020439678,0.00008203543,0.00007750637,0.000045255216],"domain_scores_gemma":[0.9995628,0.00015526358,0.000038942966,0.00006622779,0.00014457948,0.000032242944],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004312079,0.0004911506,0.00041633451,0.00080115284,0.00033616603,0.0005081446,0.0009636979,0.0006290975,0.004044031],"category_scores_gemma":[0.0010545794,0.00019131179,0.00060348934,0.0006454354,0.00019345172,0.001064277,0.0007184073,0.0012914612,0.001171231],"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":[0.000227827,0.00072550215,0.002797851,0.00020487804,0.00008437915,0.00025882543,0.00015972329,0.06259241,0.027754977,0.006419834,0.021601856,0.877172],"study_design_scores_gemma":[0.000013400876,0.000065156346,0.00055428257,0.0000092338405,0.000015541302,0.000047077476,0.00002765542,0.98500234,0.0074311104,0.004600511,0.0022251043,0.0000086719],"about_ca_topic_score_codex":0.0084959855,"about_ca_topic_score_gemma":0.013670697,"teacher_disagreement_score":0.0084959855,"about_ca_system_score_codex":0.00089629943,"about_ca_system_score_gemma":0.0015135717,"threshold_uncertainty_score":0.016893089},"labels":[],"label_agreement":null},{"id":"W4404366193","doi":"10.1007/s10115-024-02269-2","title":"An evidence-based approach for open-domain question answering","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"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":"Computer science; Question answering; Information retrieval; Relevance (law); Open domain; Benchmark (surveying); Context (archaeology); Graph; Construct (python library); Domain (mathematical analysis); Rank (graph theory); Knowledge graph; Artificial intelligence; Natural language processing; Theoretical computer science","score_opus":0.05172166446128868,"score_gpt":0.3110074057323632,"score_spread":0.2592857412710745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404366193","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":[0.003533508,0.00069479825,0.99125683,0.0012074136,0.00006746026,0.00020945593,0.00050776254,0.00070343097,0.0018193129],"genre_scores_gemma":[0.14353503,0.0006039762,0.8507194,0.00038307987,0.0002351769,0.0004095359,0.0018820334,0.00012907085,0.0021026877],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99002516,0.0042963447,0.0011576712,0.0015503146,0.0026567655,0.0003137636],"domain_scores_gemma":[0.9604143,0.03180639,0.001133372,0.0023874668,0.0035804948,0.0006781162],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0088766515,0.0011675428,0.0021584837,0.008564285,0.0017152188,0.0054906723,0.0048756907,0.0044106846,0.0077177626],"category_scores_gemma":[0.046794035,0.0011998102,0.0030599614,0.0059729707,0.0018357543,0.008595634,0.0057880343,0.0043659937,0.0021195726],"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":[0.0007208426,0.0009017002,0.0047578067,0.0014387921,0.00071806245,0.00078599993,0.0018627165,0.04780253,0.007245023,0.19426104,0.013748316,0.7257572],"study_design_scores_gemma":[0.00013399987,0.00012212877,0.0012800212,0.00032268482,0.00042201197,0.00044572944,0.00049679185,0.6428141,0.0044425814,0.33252805,0.016899427,0.00009245564],"about_ca_topic_score_codex":0.0051901164,"about_ca_topic_score_gemma":0.008744657,"teacher_disagreement_score":0.0088766515,"about_ca_system_score_codex":0.0017311564,"about_ca_system_score_gemma":0.0030967821,"threshold_uncertainty_score":0.046944797},"labels":[],"label_agreement":null},{"id":"W4404447912","doi":"10.1007/s10115-024-02282-5","title":"Quantum computing and quantum-inspired techniques for feature subset selection: a review","year":2024,"lang":"en","type":"review","venue":"Knowledge and Information Systems","topic":"Quantum Computing Algorithms and Architecture","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 Saskatchewan","funders":"","keywords":"Feature selection; Quantum; Feature (linguistics); Computer science; Quantum computer; Selection (genetic algorithm); Artificial intelligence; Quantum machine learning; Theoretical computer science; Pattern recognition (psychology); Physics; Quantum mechanics","score_opus":0.023840847655537144,"score_gpt":0.31531260652412096,"score_spread":0.29147175886858384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404447912","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":[0.00027569736,0.99585944,0.001994657,0.0003665482,0.00015519443,0.000008251912,0.00002605829,0.000014265276,0.001299805],"genre_scores_gemma":[0.0023401184,0.9951084,0.0016078248,0.0001919598,0.0002788948,0.000013442817,0.000043774427,0.000004194947,0.0004113339],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9998085,0.00003498943,0.00002246956,0.000045692475,0.000073919204,0.000014434929],"domain_scores_gemma":[0.99930584,0.00046094265,0.000053039126,0.000028624689,0.00012664047,0.000024922032],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007047219,0.0007545839,0.0015343305,0.0019475736,0.00031344575,0.0010127568,0.0011970378,0.0009185975,0.0030818311],"category_scores_gemma":[0.0013790709,0.0003421181,0.00062099734,0.0031417126,0.00067702,0.0017194876,0.00068477215,0.0012578106,0.0010227647],"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":[0.000061708284,0.00010648232,0.000297403,0.01626127,0.000182792,0.00012401823,0.000054582462,0.001678193,0.0022428164,0.019330349,0.0198906,0.9397697],"study_design_scores_gemma":[0.000040036666,0.00021401909,0.0012249197,0.0042684563,0.00036060603,0.0012839395,0.0000830975,0.0023703945,0.0026294424,0.023744144,0.96369797,0.00008301575],"about_ca_topic_score_codex":0.0014499451,"about_ca_topic_score_gemma":0.0020298625,"teacher_disagreement_score":0.0030818311,"about_ca_system_score_codex":0.0006254698,"about_ca_system_score_gemma":0.0012452584,"threshold_uncertainty_score":0.010309756},"labels":[],"label_agreement":null},{"id":"W4405700372","doi":"10.1007/s10115-024-02308-y","title":"Building a comprehensive and multi-dimensional information security ontology: elicitation process and OWL implementation","year":2024,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Information and Cyber Security","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":"Computer science; OWL-S; Ontology; Process (computing); Web Ontology Language; Information retrieval; Semantic Web; Semantic Web Stack; Programming language","score_opus":0.013764161713108507,"score_gpt":0.30792314758786826,"score_spread":0.29415898587475975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405700372","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":[0.050417703,0.00011337553,0.92648786,0.0016437838,0.00006967354,0.0021648544,0.0058592805,0.002044725,0.011198792],"genre_scores_gemma":[0.07400801,0.00013730663,0.9172592,0.00019066302,0.000010548926,0.000789074,0.0055057537,0.00028307672,0.0018162623],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9931346,0.0017942801,0.0012984175,0.00072626915,0.0026439675,0.00040237175],"domain_scores_gemma":[0.98789436,0.0045535574,0.0007364993,0.0019255113,0.004488713,0.0004014429],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010480864,0.0007865357,0.000679599,0.004633094,0.002094867,0.0039404538,0.0016560433,0.0010489883,0.0031411378],"category_scores_gemma":[0.02237826,0.00097079243,0.0019205853,0.003235831,0.0011118149,0.0056802863,0.0048672766,0.0023681915,0.0010497521],"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":[0.00032884078,0.0013719731,0.021371871,0.0037765335,0.00037903417,0.0017514717,0.024469584,0.028878298,0.053501666,0.2355609,0.029450199,0.5991596],"study_design_scores_gemma":[0.0001996098,0.00025696488,0.013694586,0.0027241488,0.00075519196,0.0015677417,0.030016463,0.25239706,0.120919324,0.20300668,0.3740954,0.00036678955],"about_ca_topic_score_codex":0.016658837,"about_ca_topic_score_gemma":0.026853945,"teacher_disagreement_score":0.016658837,"about_ca_system_score_codex":0.003192829,"about_ca_system_score_gemma":0.012730371,"threshold_uncertainty_score":0.055428803},"labels":[],"label_agreement":null},{"id":"W4406070280","doi":"10.1007/s10115-024-02321-1","title":"Fake news detection: comparative evaluation of BERT-like models and large language models with generative AI-annotated data","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":43,"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; Vector Institute","funders":"","keywords":"Computer science; Generative grammar; Language model; Fake news; Artificial intelligence; Natural language processing; Generative model; Information retrieval","score_opus":0.1036317786528191,"score_gpt":0.3879677179702772,"score_spread":0.2843359393174581,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406070280","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":[0.75622374,0.00705089,0.18552965,0.008416146,0.0011039882,0.0009776815,0.006865086,0.018874057,0.014958662],"genre_scores_gemma":[0.93087757,0.0008616416,0.05618494,0.0005697434,0.0002324105,0.000231988,0.007903374,0.00070268824,0.0024357145],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.988739,0.0073841517,0.00069950096,0.001665825,0.0011213087,0.00039024607],"domain_scores_gemma":[0.8040937,0.17951795,0.0024508706,0.006114888,0.00570482,0.002117758],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.023140727,0.0026697882,0.0021198986,0.004082035,0.0016541415,0.0046085375,0.0046598227,0.004596998,0.004127358],"category_scores_gemma":[0.07072268,0.0011824832,0.0018076805,0.0023782176,0.0018486802,0.008757744,0.0026824968,0.004013559,0.0024430533],"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":[0.011297772,0.0043075024,0.026995797,0.0022334831,0.0023696371,0.00057775143,0.001710602,0.6459736,0.0027614457,0.010000179,0.018530698,0.27324155],"study_design_scores_gemma":[0.00012223354,0.00017780889,0.00088712195,0.000039807746,0.00012652637,0.000055698238,0.00012923665,0.9947038,0.0007170009,0.0025367155,0.0004690857,0.000034899134],"about_ca_topic_score_codex":0.03869615,"about_ca_topic_score_gemma":0.03139057,"teacher_disagreement_score":0.03869615,"about_ca_system_score_codex":0.0037355318,"about_ca_system_score_gemma":0.0030254913,"threshold_uncertainty_score":0.12238133},"labels":[],"label_agreement":null},{"id":"W4406711870","doi":"10.1007/s10115-025-02338-0","title":"Unlocking wisdom: enhancing biomedical question answering with domain knowledge","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Topic Modeling","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":"Question answering; Domain (mathematical analysis); Computer science; Information retrieval; Domain knowledge; Data science; Knowledge management; Mathematics","score_opus":0.007620119195518866,"score_gpt":0.24825661733302845,"score_spread":0.2406364981375096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406711870","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":[0.05430708,0.0055256314,0.91614616,0.008659903,0.00047409607,0.00035806975,0.0023113918,0.0053603905,0.0068572927],"genre_scores_gemma":[0.39431584,0.0025881,0.59233534,0.001904283,0.0007463148,0.00024993243,0.0047643883,0.0004832764,0.002612555],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9928377,0.0033607176,0.00058957713,0.0015119177,0.0014720757,0.00022796869],"domain_scores_gemma":[0.95061433,0.03944937,0.0013072796,0.004724205,0.0031408677,0.0007638964],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009022497,0.0012690555,0.0016889955,0.0044372943,0.0011763973,0.004825445,0.0025536388,0.0027782829,0.0036089814],"category_scores_gemma":[0.059460215,0.00075652974,0.0017655015,0.0029734473,0.0012321578,0.01377882,0.005940647,0.0038271116,0.002043953],"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":[0.0006706817,0.0009689598,0.010387246,0.0016118119,0.00047161267,0.00033980847,0.0030128611,0.028899025,0.01082335,0.025925258,0.023621384,0.89326817],"study_design_scores_gemma":[0.00013924374,0.00025613408,0.0029702797,0.0003355075,0.0006850734,0.00042304225,0.0014282042,0.6020393,0.011197248,0.34973451,0.03066989,0.00012160852],"about_ca_topic_score_codex":0.0025941692,"about_ca_topic_score_gemma":0.0048789214,"teacher_disagreement_score":0.009022497,"about_ca_system_score_codex":0.00090443034,"about_ca_system_score_gemma":0.0020635496,"threshold_uncertainty_score":0.04771608},"labels":[],"label_agreement":null},{"id":"W4407052498","doi":"10.1007/s10115-025-02346-0","title":"A novel contrastive multi-view framework for heterogeneous graph embedding","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Graph Neural Networks","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":"Mount Royal University","funders":"","keywords":"Embedding; Computer science; Graph; Theoretical computer science; Artificial intelligence; Mathematics; Natural language processing","score_opus":0.0197432400307529,"score_gpt":0.30193892668569033,"score_spread":0.28219568665493744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407052498","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":[0.0033019911,0.00023734814,0.9950001,0.00011525471,0.000032449083,0.000026443615,0.00016426537,0.00032884904,0.00079329294],"genre_scores_gemma":[0.23372571,0.0006286226,0.7583791,0.00031989536,0.00017711172,0.00019436821,0.0014957583,0.0004125117,0.004666918],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99925977,0.00022246336,0.00003065414,0.00023199472,0.00019428469,0.00006088786],"domain_scores_gemma":[0.9990631,0.00033297174,0.00008392607,0.00022207452,0.00021523771,0.000082762526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010229041,0.00108125,0.0011816955,0.0015783154,0.0004760426,0.001704453,0.002648952,0.001606342,0.004007639],"category_scores_gemma":[0.0030070809,0.00055818487,0.0011267466,0.001595019,0.00065524934,0.0034595185,0.0027178586,0.002284261,0.001074315],"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":[0.00048503594,0.00029579102,0.0014017858,0.00044690378,0.0003483632,0.00041448447,0.00031729508,0.14807141,0.036835033,0.24307156,0.015565171,0.55274725],"study_design_scores_gemma":[0.000012907668,0.000052302832,0.00019860694,0.000015788914,0.000032369942,0.0001021952,0.000030151275,0.94477457,0.0025155977,0.048644334,0.0036049879,0.000016156773],"about_ca_topic_score_codex":0.0037282542,"about_ca_topic_score_gemma":0.0072044316,"teacher_disagreement_score":0.004007639,"about_ca_system_score_codex":0.0006591571,"about_ca_system_score_gemma":0.00065385574,"threshold_uncertainty_score":0.013406873},"labels":[],"label_agreement":null},{"id":"W4408468809","doi":"10.1007/s10115-025-02373-x","title":"Integrated deep learning and blockchain-based framework for cloud manufacturing with improved customer satisfaction","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Digital Transformation in Industry","field":"Engineering","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":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Blockchain; Customer satisfaction; Cloud computing; Deep learning; Computer science; Artificial intelligence; Cloud manufacturing; Data science; Business; Marketing; Computer security","score_opus":0.0062192971478256314,"score_gpt":0.21846607383437658,"score_spread":0.21224677668655095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408468809","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":[0.22867657,0.0008806698,0.7542017,0.0011347162,0.00018326197,0.000111207766,0.00053165894,0.0018080547,0.012472135],"genre_scores_gemma":[0.9699545,0.00013890641,0.025995431,0.00009509034,0.000021304184,0.00003506475,0.00024297576,0.000025751133,0.0034909416],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997166,0.000059453872,0.000011510631,0.000052589843,0.00008231069,0.00007739058],"domain_scores_gemma":[0.9996731,0.00012127669,0.000026281406,0.000035465535,0.00010736242,0.000036379293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051821925,0.00036701775,0.0006404091,0.00032430905,0.0003797938,0.00073333835,0.00084920437,0.00077204837,0.0048848866],"category_scores_gemma":[0.00075146137,0.00017683835,0.00035101155,0.00051076326,0.00030870427,0.0010455616,0.0009925247,0.00073743064,0.000451769],"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":[0.00038075913,0.0003503343,0.0028410302,0.000090385365,0.000049598842,0.00015085483,0.000042404317,0.8547973,0.004966734,0.0059864842,0.0027283477,0.12761584],"study_design_scores_gemma":[0.0000054565653,0.00001999621,0.00012653286,0.0000016655207,0.0000037821005,0.000004810766,0.0000055941096,0.99759346,0.0005339989,0.0015124105,0.0001898592,0.0000024048863],"about_ca_topic_score_codex":0.009584733,"about_ca_topic_score_gemma":0.015052615,"teacher_disagreement_score":0.009584733,"about_ca_system_score_codex":0.0007579846,"about_ca_system_score_gemma":0.0017632112,"threshold_uncertainty_score":0.01905787},"labels":[],"label_agreement":null},{"id":"W4409393577","doi":"10.1007/s10115-025-02389-3","title":"Atom search optimization: a systematic review of current variants and applications","year":2025,"lang":"en","type":"review","venue":"Knowledge and Information Systems","topic":"Metaheuristic Optimization 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 Chicoutimi","funders":"","keywords":"Current (fluid); Computer science; Atom (system on chip); Algorithm; Theoretical computer science; Mathematics; Engineering; Parallel computing; Electrical engineering","score_opus":0.04351094629363648,"score_gpt":0.3644960439534498,"score_spread":0.3209850976598133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409393577","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":[0.00019871643,0.9983719,0.0008279579,0.00013498231,0.0000522063,0.000023314436,0.00007751989,0.000010621962,0.0003029041],"genre_scores_gemma":[0.0026039367,0.9933791,0.0033526574,0.00023489151,0.000087466295,0.000049973478,0.00011929906,0.0000133307285,0.00015929838],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9987863,0.00031508642,0.0003061752,0.00020863672,0.00033482097,0.000049027007],"domain_scores_gemma":[0.99533147,0.003713437,0.00040749658,0.00011090937,0.00036905296,0.00006767635],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029622642,0.0015415282,0.0036800103,0.003953484,0.00032402814,0.0020285018,0.0017587083,0.0013636728,0.005474031],"category_scores_gemma":[0.009594556,0.00059354777,0.003168311,0.0064108544,0.00058663,0.0015601707,0.001032442,0.0012971164,0.0009005861],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031565718,0.000116859344,0.00061631453,0.17834122,0.0021935622,0.00009278375,0.000072622744,0.0028302318,0.0006320779,0.004098843,0.012166999,0.79852295],"study_design_scores_gemma":[0.00074597134,0.0012374355,0.0052270973,0.1631771,0.02162509,0.0017707638,0.00033898314,0.0061837067,0.0022303981,0.016161978,0.78100497,0.00029651218],"about_ca_topic_score_codex":0.0031749667,"about_ca_topic_score_gemma":0.0064829076,"teacher_disagreement_score":0.005474031,"about_ca_system_score_codex":0.00079744734,"about_ca_system_score_gemma":0.0037699367,"threshold_uncertainty_score":0.018312454},"labels":[],"label_agreement":null},{"id":"W4409602824","doi":"10.1007/s10115-025-02408-3","title":"Convolution neural network-AlexNet with gazelle optimization algorithm-based software defect prediction","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software Engineering Research","field":"Computer Science","cited_by":4,"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; Artificial neural network; Software; Convolution (computer science); Convolutional neural network; Algorithm; Artificial intelligence; Pattern recognition (psychology); Data mining; Machine learning","score_opus":0.006693307644077392,"score_gpt":0.21677554290158868,"score_spread":0.2100822352575113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409602824","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":[0.30647585,0.001892792,0.66660196,0.00074899045,0.00040523085,0.00012916191,0.0016675317,0.014669637,0.0074088597],"genre_scores_gemma":[0.9058656,0.000408354,0.08226555,0.00020391365,0.00007206571,0.000094864736,0.0027425922,0.00019123423,0.008155865],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99962866,0.000028641965,0.00001977341,0.000135748,0.000107961365,0.00007920995],"domain_scores_gemma":[0.9995906,0.00008033214,0.000042958523,0.00005483596,0.00020286163,0.000028446026],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005170647,0.0010035048,0.0010175756,0.0011278678,0.00036063849,0.0006342893,0.0016804906,0.0011025586,0.002330731],"category_scores_gemma":[0.0010649464,0.00036514853,0.000779052,0.0010072341,0.00031412815,0.0011958344,0.0007744034,0.0009932304,0.0009317607],"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":[0.0004506439,0.0004781147,0.009612527,0.00016232769,0.00020029467,0.00023095036,0.00005768349,0.35885626,0.014056915,0.0027566166,0.018866494,0.5942712],"study_design_scores_gemma":[0.0000050064214,0.000021064458,0.00059049705,0.000004652705,0.000012232809,0.00001618633,0.0000037134273,0.99674004,0.0019579355,0.00042059828,0.00022386218,0.0000042395486],"about_ca_topic_score_codex":0.023861652,"about_ca_topic_score_gemma":0.02133195,"teacher_disagreement_score":0.023861652,"about_ca_system_score_codex":0.00088068633,"about_ca_system_score_gemma":0.0013871111,"threshold_uncertainty_score":0.047445536},"labels":[],"label_agreement":null},{"id":"W4409677710","doi":"10.1007/s10115-025-02381-x","title":"A Bayesian hierarchical model for orthogonal Tucker decomposition with oblivious tensor compression","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Tensor decomposition and applications","field":"Mathematics","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 Alberta","funders":"","keywords":"Tucker decomposition; Tensor decomposition; Decomposition; Compression (physics); Tensor (intrinsic definition); Bayesian probability; Mathematics; Computer science; Artificial intelligence; Algorithm; Pattern recognition (psychology); Pure mathematics; Chemistry; Physics","score_opus":0.022339431200952747,"score_gpt":0.32378291151621913,"score_spread":0.30144348031526635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409677710","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":[0.0048411526,0.00020159593,0.9929107,0.00035888923,0.000030241106,0.000030874373,0.00024180625,0.00023477314,0.0011498696],"genre_scores_gemma":[0.41943315,0.0012846753,0.5598404,0.0005334445,0.00030719166,0.000540211,0.0017117811,0.0004987083,0.015850475],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99820125,0.0007805835,0.00009387297,0.00033150884,0.0003682292,0.00022460209],"domain_scores_gemma":[0.99522805,0.002611038,0.0004714307,0.00081099063,0.00065549626,0.00022301753],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032148534,0.0009899555,0.0019744253,0.0011756512,0.00080006267,0.0020561481,0.002769689,0.002213102,0.0053952145],"category_scores_gemma":[0.0119917765,0.0010802838,0.0011799515,0.0023075666,0.002003685,0.0047786455,0.0022909162,0.003087255,0.0016075746],"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":[0.00020120025,0.000084732346,0.00038168978,0.000112298054,0.00007101379,0.00008333433,0.00012260467,0.55538166,0.0011473014,0.38658214,0.0056107333,0.050221305],"study_design_scores_gemma":[0.000008653277,0.0000111686195,0.00004663361,0.000009002585,0.000007098665,0.000012489736,0.000006055652,0.9119976,0.00016061177,0.087234795,0.0004934979,0.0000123608015],"about_ca_topic_score_codex":0.0106872115,"about_ca_topic_score_gemma":0.011075132,"teacher_disagreement_score":0.0106872115,"about_ca_system_score_codex":0.0017048579,"about_ca_system_score_gemma":0.0028681937,"threshold_uncertainty_score":0.02125001},"labels":[],"label_agreement":null},{"id":"W4410692212","doi":"10.1007/s10115-025-02417-2","title":"Providing and evaluating a model for big data anonymization streams by using in-memory processing","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","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":"Regional Municipality of Niagara","funders":"","keywords":"Computer science; Big data; Data stream mining; Data mining; STREAMS; Data anonymization; Data science; Artificial intelligence; Information privacy; Computer security","score_opus":0.10863723861002288,"score_gpt":0.3474931435090034,"score_spread":0.2388559048989805,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410692212","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":[0.034733977,0.00019259212,0.9599126,0.00079864456,0.00012489446,0.00021667512,0.0002829995,0.0011739639,0.00256368],"genre_scores_gemma":[0.5500241,0.0005094272,0.4444589,0.0002672859,0.00014050378,0.00039972665,0.0010362318,0.0002188607,0.0029449314],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99754333,0.0008302453,0.00014436804,0.00053996197,0.0006739047,0.000268261],"domain_scores_gemma":[0.994991,0.0023911514,0.00041593783,0.0009950255,0.0009996226,0.0002072179],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004200311,0.0010327793,0.0008916215,0.0010861141,0.0012155054,0.0034065668,0.0024921077,0.0014591167,0.0017379565],"category_scores_gemma":[0.011157952,0.0004105819,0.00093487266,0.0013559374,0.0012031851,0.0049072355,0.0015331483,0.0018619549,0.0006400116],"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":[0.00034590895,0.00019444803,0.0021720782,0.000093828094,0.000049218015,0.00011170034,0.00018826913,0.91681683,0.0022866044,0.035754073,0.0033523003,0.038634684],"study_design_scores_gemma":[0.0000053671315,0.000014686201,0.000052527746,0.0000028091779,0.000003614197,0.000016516015,0.000025196814,0.9929912,0.00081405457,0.0055732885,0.00049632794,0.000004417443],"about_ca_topic_score_codex":0.008318335,"about_ca_topic_score_gemma":0.007460958,"teacher_disagreement_score":0.008318335,"about_ca_system_score_codex":0.0028000455,"about_ca_system_score_gemma":0.002715582,"threshold_uncertainty_score":0.022213638},"labels":[],"label_agreement":null},{"id":"W4411073803","doi":"10.1007/s10115-025-02479-2","title":"A new privacy-preserving approach for publishing periodical reporting systems data","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","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":"Nipissing University","funders":"National Natural Science Foundation of China","keywords":"Data publishing; Computer science; Publishing; Information privacy; Data science; Information retrieval; Internet privacy; Data mining; World Wide Web; Political science","score_opus":0.07142405149752376,"score_gpt":0.31387815178908424,"score_spread":0.24245410029156048,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411073803","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":[0.0035618823,0.00039694895,0.9859513,0.002069891,0.00042548368,0.00018445235,0.0003907561,0.0014675303,0.0055515743],"genre_scores_gemma":[0.1648076,0.0013008483,0.80583596,0.0016726495,0.0021921848,0.00046425528,0.0013170914,0.00054531376,0.021864025],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97343725,0.007171664,0.0029633446,0.00368462,0.011657863,0.0010852611],"domain_scores_gemma":[0.9507025,0.010624407,0.0025354226,0.029438483,0.005671646,0.0010275823],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013948292,0.000773623,0.0017669257,0.004531141,0.0032600204,0.0131896455,0.0053502093,0.0034538994,0.006619264],"category_scores_gemma":[0.0325782,0.0011784278,0.0026666236,0.006446609,0.0038754812,0.019529184,0.0075629796,0.0063687256,0.0040687076],"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":[0.00043285594,0.00029760806,0.0010211476,0.00027526505,0.00018392237,0.0005676668,0.00096727617,0.00918229,0.011144329,0.77109355,0.017018834,0.18781525],"study_design_scores_gemma":[0.00016936388,0.00029437835,0.00052119454,0.0001374996,0.00036910572,0.0023164458,0.00034853866,0.16680853,0.033235617,0.6448339,0.15077919,0.00018610811],"about_ca_topic_score_codex":0.0011330043,"about_ca_topic_score_gemma":0.0009740931,"teacher_disagreement_score":0.013948292,"about_ca_system_score_codex":0.002152871,"about_ca_system_score_gemma":0.005250215,"threshold_uncertainty_score":0.07376647},"labels":[],"label_agreement":null},{"id":"W4413243437","doi":"10.1007/s10115-025-02553-9","title":"Light spectral-based graph convolutional networks: optimizing ranking and longtail awareness for e-commerce recommendation systems","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Recommender Systems and Techniques","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":"Computer science; Recommender system; Graph; Ranking (information retrieval); Information retrieval; Theoretical computer science; Artificial intelligence; Data mining","score_opus":0.016379129920057144,"score_gpt":0.2629572443075103,"score_spread":0.24657811438745314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413243437","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":[0.09916506,0.0011929411,0.8925282,0.0007261812,0.00014217633,0.000060753544,0.0003520153,0.0024293957,0.0034034245],"genre_scores_gemma":[0.889461,0.00049261336,0.101066165,0.0004106515,0.00010903921,0.000060934897,0.0005177885,0.00018785599,0.0076939655],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99968207,0.0000669787,0.000014542337,0.00009639773,0.00006643116,0.00007367193],"domain_scores_gemma":[0.99877983,0.00062458444,0.000102410646,0.00013785656,0.00028211382,0.00007323497],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080611854,0.000859588,0.0010885809,0.00087577855,0.00057608495,0.0010382477,0.0017912093,0.0015673088,0.0024860366],"category_scores_gemma":[0.0037606696,0.00059221894,0.000512743,0.0011122322,0.0006592777,0.0020052104,0.00094068795,0.0013593672,0.0008144342],"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":[0.00045982402,0.00041600215,0.0024133965,0.00015464684,0.0001260354,0.00008191954,0.00010626785,0.6357599,0.010062391,0.009933681,0.008181375,0.33230457],"study_design_scores_gemma":[0.0000033379742,0.000013064847,0.00010504254,0.000002760788,0.0000074390655,0.0000047596263,0.0000040375353,0.9971727,0.0005020026,0.0020677694,0.00011411408,0.0000029417768],"about_ca_topic_score_codex":0.026112834,"about_ca_topic_score_gemma":0.045968037,"teacher_disagreement_score":0.026112834,"about_ca_system_score_codex":0.0014376225,"about_ca_system_score_gemma":0.0011480756,"threshold_uncertainty_score":0.051921666},"labels":[],"label_agreement":null},{"id":"W4413982584","doi":"10.1007/s10115-025-02581-5","title":"A knowledge-based approach for pseudo-relevance feedback by exploiting semantic relevance","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Information Retrieval and Search Behavior","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":"York University","funders":"","keywords":"Relevance (law); Relevance feedback; Computer science; Information retrieval; Artificial intelligence; Image retrieval","score_opus":0.019883780679341264,"score_gpt":0.2714622971701461,"score_spread":0.2515785164908048,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413982584","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":[0.0061076335,0.0003299564,0.98965144,0.00027912544,0.00009584036,0.00015373489,0.00009989125,0.00090808445,0.002374254],"genre_scores_gemma":[0.41957793,0.0003454488,0.5713725,0.000358886,0.00028960925,0.00037089834,0.00046493046,0.00026302988,0.006956795],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9940546,0.0020891412,0.0003204862,0.0007820238,0.002471671,0.00028201763],"domain_scores_gemma":[0.9901529,0.00482697,0.00031667395,0.0016872879,0.0027777492,0.00023848268],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004035518,0.00090910704,0.0015074343,0.0029399376,0.001197161,0.0021947434,0.0033270828,0.0022414406,0.0063588414],"category_scores_gemma":[0.018596716,0.00060299,0.0011115946,0.0020910217,0.0012177264,0.005986033,0.0025573994,0.0018368294,0.002164808],"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":[0.0010033278,0.0010337429,0.0012641356,0.000672198,0.00022586204,0.00035853055,0.00082075974,0.063813135,0.03595321,0.11213849,0.014335052,0.76838154],"study_design_scores_gemma":[0.00007013977,0.00016258635,0.0003875243,0.00003690717,0.00008076789,0.00020534325,0.00006746743,0.9124602,0.009435197,0.071983024,0.0050378456,0.00007292388],"about_ca_topic_score_codex":0.004033792,"about_ca_topic_score_gemma":0.006513511,"teacher_disagreement_score":0.0063588414,"about_ca_system_score_codex":0.001060511,"about_ca_system_score_gemma":0.0021551433,"threshold_uncertainty_score":0.021342099},"labels":[],"label_agreement":null},{"id":"W4414874910","doi":"10.1007/s10115-025-02552-w","title":"LINDAS: a learned approach to index algorithm selection","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Management and Algorithms","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":"York University","funders":"National Natural Science Foundation of China","keywords":"Search engine indexing; Index (typography); Selection (genetic algorithm); Hash function; Index selection; Range (aeronautics)","score_opus":0.012550522390299163,"score_gpt":0.2511002347204886,"score_spread":0.23854971233018943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414874910","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":[0.0017816199,0.00022593993,0.9875897,0.00028280416,0.00012492198,0.00010137833,0.00035544357,0.007880293,0.0016578018],"genre_scores_gemma":[0.0567,0.00027872648,0.9315127,0.00046542834,0.00028962796,0.00035894246,0.0015765207,0.0015895925,0.00722845],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99390596,0.0020644378,0.0004585481,0.00081069645,0.0024495448,0.00031088258],"domain_scores_gemma":[0.9916781,0.0039876658,0.00032507494,0.0017877871,0.0018759116,0.00034542521],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004811449,0.0014720255,0.0017403814,0.0037861518,0.0015787015,0.0039827013,0.0060766926,0.0024419308,0.01629964],"category_scores_gemma":[0.028140556,0.0011551275,0.0018109856,0.0030072036,0.0012211122,0.005321223,0.0045650713,0.004039236,0.007183031],"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":[0.00066339143,0.0004009618,0.0026738117,0.00037211858,0.00035008817,0.00017037935,0.00021712855,0.060399126,0.0030520954,0.044137973,0.047823884,0.83973914],"study_design_scores_gemma":[0.000115525385,0.00010476259,0.00026002436,0.00006237984,0.0000717003,0.00016254616,0.00005573384,0.90885067,0.0040465286,0.06895257,0.01727892,0.000038755767],"about_ca_topic_score_codex":0.0034688425,"about_ca_topic_score_gemma":0.00963736,"teacher_disagreement_score":0.01629964,"about_ca_system_score_codex":0.0014749538,"about_ca_system_score_gemma":0.0034539865,"threshold_uncertainty_score":0.0545277},"labels":[],"label_agreement":null},{"id":"W4415139805","doi":"10.1007/s10115-025-02601-4","title":"Learning to rank under uncertainty: a robust neural approach","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Machine Learning and Algorithms","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":"ENCODE; Probabilistic logic; Relevance (law); Generalizability theory; Ambiguity; Divergence-from-randomness model; Rank (graph theory); Space (punctuation); Learning to rank","score_opus":0.011889342082284602,"score_gpt":0.2471812885091502,"score_spread":0.2352919464268656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415139805","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":[0.0067036697,0.0006665194,0.99041754,0.0005320853,0.000083439845,0.000027722357,0.00009434563,0.0002742593,0.0012003455],"genre_scores_gemma":[0.64415485,0.0017188677,0.34229273,0.00076862174,0.0012084096,0.00024333433,0.00072786084,0.00027634387,0.008608928],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996405,0.0013744821,0.00023914114,0.00077054923,0.0009109923,0.00029977603],"domain_scores_gemma":[0.98621446,0.009069901,0.0012462233,0.0013553407,0.0018373303,0.00027678534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0062491074,0.0013695915,0.0029353302,0.0026730301,0.00091738516,0.0031894648,0.0034902308,0.0036039113,0.0035628087],"category_scores_gemma":[0.032782543,0.0010511671,0.0012207406,0.0022593623,0.0028881375,0.0053334907,0.0029805892,0.0031081217,0.0008564648],"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":[0.00022422559,0.00012090826,0.00081812154,0.00021868055,0.00023537687,0.00010852948,0.00008668525,0.7401189,0.001457297,0.07555168,0.0046612695,0.17639834],"study_design_scores_gemma":[0.000009788282,0.000027894781,0.00009504177,0.00001346052,0.000019411154,0.000020538666,0.0000070335404,0.9599388,0.00031572732,0.039236564,0.00030054623,0.0000151455915],"about_ca_topic_score_codex":0.0047296574,"about_ca_topic_score_gemma":0.0030934012,"teacher_disagreement_score":0.0062491074,"about_ca_system_score_codex":0.0011749123,"about_ca_system_score_gemma":0.0016639988,"threshold_uncertainty_score":0.03304887},"labels":[],"label_agreement":null},{"id":"W4415500225","doi":"10.1007/s10115-025-02604-1","title":"X-HEART: eXplainable heterogeneous log anomaly detection using robust transformers","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software System Performance and Reliability","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 Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; IBM Canada; Vector Institute","keywords":"Interpretability; Anomaly detection; Raw data; Anomaly (physics); Transparency (behavior); Transformer","score_opus":0.012437552983026049,"score_gpt":0.23720021654328957,"score_spread":0.22476266356026353,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415500225","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":[0.01030505,0.00006297203,0.9626764,0.00008981708,0.000031696174,0.000038126705,0.0004192946,0.025881175,0.0004954712],"genre_scores_gemma":[0.50353444,0.00015996322,0.4890385,0.00014521752,0.00007610048,0.00008672591,0.0016298549,0.002312573,0.0030167203],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994149,0.00008447211,0.000027862961,0.00021317045,0.00019900639,0.000060660306],"domain_scores_gemma":[0.99844164,0.00068669405,0.00016844692,0.00050046964,0.00014776907,0.000054930802],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00089259224,0.0011579269,0.0007513934,0.0013887692,0.00034827457,0.0013393563,0.0021363269,0.00093704887,0.0055465074],"category_scores_gemma":[0.004628624,0.0004862009,0.0008603664,0.0007106345,0.0006767585,0.00263236,0.0021778543,0.0012629518,0.001409006],"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":[0.0014884132,0.00027605443,0.011174416,0.00033251522,0.0002896667,0.0012015023,0.0003512436,0.14719827,0.034147467,0.041468095,0.0227347,0.7393377],"study_design_scores_gemma":[0.000048998038,0.00006460985,0.00062678714,0.000012722992,0.00003435037,0.0001999647,0.00003159541,0.9498964,0.010711249,0.035357483,0.0029947148,0.000021136228],"about_ca_topic_score_codex":0.0021458697,"about_ca_topic_score_gemma":0.0026446988,"teacher_disagreement_score":0.0055465074,"about_ca_system_score_codex":0.0003902235,"about_ca_system_score_gemma":0.0007902804,"threshold_uncertainty_score":0.018554926},"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":"W646176396","doi":"10.1007/s10115-015-0892-x","title":"Continuous similarity search for evolving queries","year":2015,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Advanced Image and Video Retrieval Techniques","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":"Simon Fraser University; British Columbia Institute of Technology","funders":"","keywords":"Multiset; Nearest neighbor search; Computer science; Set (abstract data type); Similarity (geometry); Data stream; Data mining; Hash function; Result set; Theoretical computer science; Information retrieval; Mathematics; Artificial intelligence","score_opus":0.03910859700177923,"score_gpt":0.31033596522634793,"score_spread":0.2712273682245687,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W646176396","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":[0.21547222,0.0037174774,0.7717161,0.0005665836,0.00014233013,0.00030906784,0.00079673127,0.0019377631,0.0053417445],"genre_scores_gemma":[0.75913817,0.0007786587,0.23295681,0.000176724,0.00018018637,0.00017597862,0.0018883613,0.00018822312,0.004516892],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99723685,0.00064781954,0.00022714422,0.00058097637,0.0011067956,0.00020039787],"domain_scores_gemma":[0.99239326,0.004738639,0.0004527863,0.0011110433,0.0010050323,0.00029936168],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030062415,0.00054023037,0.002267526,0.0030672213,0.0007576893,0.0022747016,0.003105583,0.0023862624,0.0048951237],"category_scores_gemma":[0.020014133,0.0006062967,0.0008168883,0.0043587717,0.0009674365,0.0058208895,0.0020607088,0.0016275398,0.0010864524],"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":[0.002293932,0.0011611361,0.006833219,0.0008974142,0.00024804432,0.00086019543,0.0009404044,0.17097487,0.029192723,0.062171053,0.01321137,0.7112157],"study_design_scores_gemma":[0.00011254578,0.00029536991,0.000923669,0.000019574298,0.000045698016,0.00035629995,0.00014810888,0.9660037,0.0035546946,0.026448734,0.0020680574,0.00002355349],"about_ca_topic_score_codex":0.004264256,"about_ca_topic_score_gemma":0.0029466755,"teacher_disagreement_score":0.0048951237,"about_ca_system_score_codex":0.0013392027,"about_ca_system_score_gemma":0.0010329123,"threshold_uncertainty_score":0.01637584},"labels":[],"label_agreement":null},{"id":"W7117313584","doi":"10.1007/s10115-025-02657-2","title":"Enhanced software defect prediction using edge feature and self-attention GAN with pelican optimization","year":2025,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Software Engineering 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":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Feature (linguistics); Convolutional neural network; Software; Enhanced Data Rates for GSM Evolution; Artificial neural network; Software bug; Pattern recognition (psychology); Software system; Deep learning","score_opus":0.006036464635271571,"score_gpt":0.22966269469882605,"score_spread":0.22362623006355448,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117313584","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":[0.06730061,0.00052305387,0.9268516,0.00021073956,0.000071748545,0.000034635457,0.00021144778,0.0019008863,0.0028953215],"genre_scores_gemma":[0.8349361,0.00021697501,0.15669139,0.00032483647,0.000072138144,0.00008263584,0.00090149086,0.0002559264,0.0065184105],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997658,0.000038823444,0.000009168431,0.00007573621,0.000075538774,0.00003494032],"domain_scores_gemma":[0.99952364,0.00021869525,0.00003759529,0.000055568173,0.00014138043,0.000023014198],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004572316,0.00078506046,0.0009948313,0.0007615689,0.00018366579,0.00046096413,0.001097354,0.0007684284,0.0016513655],"category_scores_gemma":[0.0010756797,0.00027206147,0.00069003925,0.00054112554,0.00024940036,0.0008731938,0.0004940538,0.00072980725,0.00040648898],"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":[0.0002726526,0.00023217621,0.0034309935,0.00009502385,0.00010625576,0.00012017525,0.000039438517,0.560045,0.019679483,0.0037568954,0.007045634,0.40517634],"study_design_scores_gemma":[0.0000022982451,0.000011431807,0.00019221428,0.0000012602161,0.000005204764,0.000008841039,0.0000013535279,0.9986823,0.0006320789,0.00036154623,0.00009973648,0.0000017745562],"about_ca_topic_score_codex":0.004411426,"about_ca_topic_score_gemma":0.008136703,"teacher_disagreement_score":0.004411426,"about_ca_system_score_codex":0.00039236684,"about_ca_system_score_gemma":0.0005976309,"threshold_uncertainty_score":0.008771479},"labels":[],"label_agreement":null}]}