{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":40,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":40,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"84a422687fe3","filters":{"venue":"ACM SIGKDD Explorations Newsletter"}},"results":[{"id":"W2011376672","doi":"10.1145/1007730.1007737","title":"Class imbalances versus small disjuncts","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":681,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Class (philosophy); Focus (optics); Classifier (UML); Artificial intelligence; Machine learning","authors":[{"name":"Taeho Jo","is_ca":true},{"name":"Nathalie Japkowicz","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.068174256664078,"gpt":0.279847161916715,"spread":0.211672905252637,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0092356,0.0008459234,0.001393928,0.001573763,0.001320537,0.003126877,0.001240193,0.00156577,0.003487227],"category_scores_gemma":[0.04035357,0.0003554256,0.0005308785,0.002080924,0.00247638,0.005249341,0.002541616,0.003003834,0.0009469661],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000862524,"about_ca_system_score_gemma":0.0005854605,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003409783,"about_ca_topic_score_gemma":0.0005857431,"domain_scores_codex":[0.9904151,0.002924455,0.0007708457,0.002246215,0.003218947,0.000424373],"domain_scores_gemma":[0.9587988,0.0317678,0.00329871,0.003617362,0.001711249,0.00080604],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0030046,0.0007708713,0.03499009,0.0009170283,0.0002916762,0.0005246293,0.0009155563,0.04283617,0.03376814,0.04540804,0.01912977,0.8174434],"study_design_scores_gemma":[0.0004166562,0.002344748,0.03315799,0.0002793549,0.0002936823,0.002319312,0.001582813,0.5031462,0.07975813,0.3345167,0.04200795,0.0001764178],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.368705,0.00465528,0.5936243,0.007358557,0.0009433451,0.0007173375,0.001057946,0.001911701,0.02102649],"genre_scores_gemma":[0.8322751,0.0009214469,0.1607827,0.0008417431,0.0006566821,0.0002880942,0.000735569,0.0002150972,0.003283542],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0092356,"threshold_uncertainty_score":0.04884315,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2040010062","doi":"10.1145/1007730.1007736","title":"Learning from imbalanced data sets with boosting and data generation","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":606,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Boosting (machine learning); Machine learning; Computer science; Artificial intelligence; Benchmarking; Classifier (UML); Training set; Gradient boosting; Ensemble learning; Data mining; Random forest","authors":[{"name":"Hongyu Guo","is_ca":true},{"name":"Herna L. Viktor","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.145876779440151,"gpt":0.3158766600273255,"spread":0.1699998805871745,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01012599,0.001029702,0.002092245,0.002456422,0.000702399,0.001710349,0.002086115,0.001133151,0.0008363509],"category_scores_gemma":[0.0198863,0.0006702857,0.001443948,0.00230897,0.0008285598,0.002178635,0.002025449,0.002533391,0.0007567537],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007716983,"about_ca_system_score_gemma":0.0009543257,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000943973,"about_ca_topic_score_gemma":0.001077159,"domain_scores_codex":[0.9953701,0.001906347,0.00027151,0.0006042019,0.001617236,0.0002307024],"domain_scores_gemma":[0.9905869,0.004327372,0.0005689799,0.002237852,0.002063448,0.0002154235],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004069439,0.0003718267,0.008087236,0.0002144557,0.0002425807,0.000151637,0.000283609,0.3871716,0.007318225,0.01398506,0.008073177,0.5736937],"study_design_scores_gemma":[0.00003896306,0.00008663099,0.0007896099,0.00002477162,0.00002594687,0.00007759319,0.00003348284,0.9731848,0.005327262,0.01701707,0.003371358,0.00002251456],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02402752,0.0005182122,0.9717959,0.0003029614,0.0001826351,0.0002745624,0.0001621339,0.001482045,0.001253995],"genre_scores_gemma":[0.2527709,0.0004439974,0.7436776,0.0003648413,0.0002268879,0.0006070378,0.0008399132,0.0001824163,0.0008863565],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01012599,"threshold_uncertainty_score":0.05355203,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1997102766","doi":"10.1145/1882471.1882478","title":"A brief survey on sequence classification","year":2010,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":552,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Sequence (biology); Feature selection; Artificial intelligence; Feature vector; Task (project management); Feature (linguistics); Machine learning; Pattern recognition (psychology); One-class classification; Support vector machine; Data mining","authors":[{"name":"Zhengzheng Xing","is_ca":true},{"name":"Jian Pei","is_ca":true},{"name":"Eamonn Keogh","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1153428195483115,"gpt":0.3164656778829131,"spread":0.2011228583346016,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002127419,0.001182812,0.002000881,0.007935201,0.0009974279,0.002917623,0.001995204,0.001630756,0.007648352],"category_scores_gemma":[0.007987824,0.0004630765,0.001290355,0.01522422,0.0006779229,0.005571523,0.001210836,0.001669289,0.007023884],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001071644,"about_ca_system_score_gemma":0.001988125,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003623365,"about_ca_topic_score_gemma":0.002026396,"domain_scores_codex":[0.9973118,0.0005059847,0.0003069131,0.0004640861,0.001261459,0.0001497886],"domain_scores_gemma":[0.9958704,0.001991598,0.0002209253,0.0004183938,0.001366528,0.000132053],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00006768613,0.00006738979,0.001249523,0.001072276,0.00004331065,0.00006980303,0.00006038824,0.003780375,0.0008701941,0.009171241,0.02954428,0.9540035],"study_design_scores_gemma":[0.00004122074,0.0003405764,0.004881024,0.001764426,0.0001214381,0.001963628,0.00031133,0.1099377,0.005453723,0.09812224,0.7769269,0.0001358127],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.007639334,0.4490373,0.4988672,0.004766244,0.004532092,0.0004299197,0.002452848,0.002400001,0.02987514],"genre_scores_gemma":[0.0650301,0.5727649,0.310064,0.002997578,0.01037024,0.0008407781,0.0122806,0.0005897784,0.02506206],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.007935201,"threshold_uncertainty_score":0.02558625,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1977843657","doi":"10.1145/1540276.1540279","title":"A brief survey on anonymization techniques for privacy preserving publishing of social network data","year":2008,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":419,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Data publishing; Computer science; Publishing; Social network (sociolinguistics); Information privacy; Data science; k-anonymity; Data anonymization; Graph; Cluster analysis; Internet privacy; Privacy protection; World Wide Web; Data mining; Social media; Artificial intelligence; Theoretical computer science","authors":[{"name":"Bin Zhou","is_ca":true},{"name":"Jian Pei","is_ca":true},{"name":"Wo-Shun Luk","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1723603973409118,"gpt":0.3270708739991915,"spread":0.1547104766582797,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009362618,0.001126807,0.00213124,0.0060283,0.002182735,0.004513108,0.002761124,0.002182632,0.003234979],"category_scores_gemma":[0.0195934,0.0009582146,0.002855928,0.01167465,0.002045332,0.00962874,0.002588589,0.002926613,0.001646852],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001453475,"about_ca_system_score_gemma":0.001972445,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001080636,"about_ca_topic_score_gemma":0.0007225778,"domain_scores_codex":[0.9828653,0.007597637,0.001681548,0.001540377,0.00576733,0.0005477707],"domain_scores_gemma":[0.9783226,0.01328214,0.001199033,0.005110153,0.001917806,0.0001683819],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001662666,0.0002186313,0.002040856,0.004755182,0.0003407922,0.0004219637,0.001215688,0.01616466,0.004533987,0.1776831,0.02048205,0.7719767],"study_design_scores_gemma":[0.0001273578,0.000600755,0.003885292,0.003169048,0.0006194366,0.009467627,0.001501383,0.1010256,0.03207237,0.2128297,0.6343119,0.0003896324],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.006310752,0.09704939,0.8775536,0.003032943,0.0006341385,0.0006919824,0.0005540813,0.0008706865,0.01330238],"genre_scores_gemma":[0.1014268,0.2205497,0.6625041,0.001674603,0.002873887,0.001362123,0.002047841,0.0004426535,0.007118254],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.009362618,"threshold_uncertainty_score":0.04951489,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1984606279","doi":"10.1145/380995.381002","title":"Mining frequent patterns by pattern-growth","year":2000,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":275,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Citation; Computer science; Data science; Library science","authors":[{"name":"Jiawei Han","is_ca":true},{"name":"Jian Pei","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02117662248008691,"gpt":0.2413436073438028,"spread":0.2201669848637159,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003352601,0.001096574,0.001306772,0.006404358,0.0008679016,0.002241037,0.001899173,0.001079605,0.001278633],"category_scores_gemma":[0.01815011,0.0007399113,0.001801517,0.007458903,0.0009778626,0.004002722,0.002000979,0.001695796,0.00146386],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006866153,"about_ca_system_score_gemma":0.001120259,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001756869,"about_ca_topic_score_gemma":0.00199644,"domain_scores_codex":[0.9958681,0.0009225828,0.0004212889,0.0007389449,0.001835931,0.0002131457],"domain_scores_gemma":[0.9881139,0.006898399,0.001114756,0.001394453,0.00225943,0.0002191068],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002769898,0.0003219997,0.02902802,0.001111007,0.0003414162,0.001516754,0.0007633065,0.0817604,0.01380335,0.04190697,0.01428274,0.814887],"study_design_scores_gemma":[0.00005029027,0.0001738079,0.004046618,0.0001739515,0.00009785911,0.001707761,0.0004296358,0.8662623,0.01030192,0.09585565,0.02083694,0.00006329585],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02504306,0.001638829,0.9662091,0.000792207,0.00009321442,0.0005261056,0.001534452,0.001921378,0.002241646],"genre_scores_gemma":[0.1315835,0.001572634,0.860498,0.0001863553,0.00009400722,0.0005125562,0.004078292,0.0001562189,0.001318495],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006404358,"threshold_uncertainty_score":0.01773041,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2101549186","doi":"10.1145/2594473.2594476","title":"Ensembles for unsupervised outlier detection","year":2014,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":273,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Anomaly detection; Outlier; Cluster analysis; Artificial intelligence; Focus (optics); Machine learning; Unsupervised learning; Core (optical fiber); Ensemble learning; Data mining; Pattern recognition (psychology); Data science","authors":[{"name":"Arthur Zimek","is_ca":false},{"name":"Ricardo J. G. B. Campello","is_ca":false},{"name":"Jörg Sander","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02540533101613996,"gpt":0.2520752218889637,"spread":0.2266698908728237,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005564787,0.00142266,0.002577454,0.003149408,0.001464692,0.002875294,0.002651253,0.001954368,0.00265995],"category_scores_gemma":[0.02456217,0.0008705167,0.002108628,0.002775196,0.001910787,0.004161863,0.004794029,0.004952644,0.001841936],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009753072,"about_ca_system_score_gemma":0.001075466,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001740689,"about_ca_topic_score_gemma":0.002271865,"domain_scores_codex":[0.9940708,0.002141521,0.0003911561,0.001404231,0.001611238,0.0003809703],"domain_scores_gemma":[0.9840063,0.006598711,0.001466586,0.004184335,0.003157094,0.0005868931],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002061435,0.0001537864,0.009718353,0.0003650739,0.0007180205,0.000245107,0.000540842,0.4126281,0.007177214,0.1840091,0.01516169,0.3690766],"study_design_scores_gemma":[0.000008656581,0.00006099247,0.0008924107,0.00005120235,0.0000389214,0.0001303156,0.00006205337,0.8451937,0.002080007,0.1441171,0.007320357,0.00004427313],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004022422,0.0004914585,0.9933347,0.0002686388,0.0001140028,0.00003373204,0.0001166666,0.0006720388,0.0009464174],"genre_scores_gemma":[0.328527,0.001576706,0.6613675,0.0006629539,0.001085126,0.000377689,0.001847781,0.0006825387,0.003872645],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005564787,"threshold_uncertainty_score":0.02942979,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2044306058","doi":"10.1145/568574.568580","title":"Constrained frequent pattern mining","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":133,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Data mining; Sequential Pattern Mining; Constraint (computer-aided design); K-optimal pattern discovery; Context (archaeology); Process (computing); Association rule learning; Mathematics","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Jiawei Han","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06108663403363127,"gpt":0.2512890157389533,"spread":0.190202381705322,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005251919,0.001276512,0.002455785,0.005508731,0.001369036,0.003103514,0.003766886,0.001858197,0.005152642],"category_scores_gemma":[0.03992119,0.0008629601,0.001759415,0.008675211,0.001130665,0.005403168,0.002583422,0.00186896,0.002334146],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006319402,"about_ca_system_score_gemma":0.001881423,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001377088,"about_ca_topic_score_gemma":0.001620722,"domain_scores_codex":[0.9903092,0.002818198,0.00122158,0.001902688,0.003350458,0.0003978335],"domain_scores_gemma":[0.9751687,0.01457816,0.001923212,0.003707842,0.004147016,0.0004749346],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004916203,0.0003649605,0.01206932,0.002616185,0.000818268,0.002889421,0.0005433671,0.1049832,0.006464588,0.175185,0.03474806,0.6588261],"study_design_scores_gemma":[0.0001094705,0.0001384126,0.002121421,0.0003169719,0.0001505343,0.003018994,0.0003053755,0.5641562,0.003941527,0.3787076,0.04694428,0.00008923117],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01172355,0.00223655,0.9773384,0.0007698439,0.0002378818,0.0005609869,0.002301462,0.0006671914,0.004164026],"genre_scores_gemma":[0.1749645,0.003233674,0.8063217,0.0007076478,0.0004133404,0.001094254,0.0091111,0.0002122952,0.003941581],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005508731,"threshold_uncertainty_score":0.02777511,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2069116068","doi":"10.1145/507515.507523","title":"KDD Cup 2001 report","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":110,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Canadian Imperial Bank of Commerce (Canada)","funders":"National Science Foundation","keywords":"Computer science; Artificial intelligence","authors":[{"name":"Jie Cheng","is_ca":true},{"name":"Christos Hatzis","is_ca":false},{"name":"Hisashi Hayashi","is_ca":false},{"name":"Mark-A. Krogel","is_ca":false},{"name":"Shinichi Morishita","is_ca":false},{"name":"David Page","is_ca":false},{"name":"Jun Sese","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03648866402031892,"gpt":0.2723488489952492,"spread":0.2358601849749303,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01427799,0.004349593,0.003919384,0.01407465,0.003423423,0.01518375,0.007351962,0.005119074,0.06238117],"category_scores_gemma":[0.04080032,0.001366317,0.002700991,0.01372903,0.0007897636,0.009432475,0.005072314,0.004595072,0.1541951],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004508577,"about_ca_system_score_gemma":0.01078803,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02085322,"about_ca_topic_score_gemma":0.01241318,"domain_scores_codex":[0.9832467,0.002093966,0.001680836,0.001879014,0.01016904,0.0009303888],"domain_scores_gemma":[0.9593453,0.003876726,0.001049142,0.007337234,0.02550213,0.002889462],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004626579,0.00004531195,0.0003155106,0.000190382,0.00002777871,0.00002789148,0.000009346163,0.000566337,0.00005136554,0.001528973,0.9649695,0.03222144],"study_design_scores_gemma":[0.00008870962,0.00003425711,0.001162278,0.0001766385,0.00004481963,0.0001362808,0.0000494795,0.004788816,0.0007265691,0.004626045,0.9881252,0.00004091719],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.0058059,0.03123854,0.07683369,0.04617757,0.02568327,0.002634465,0.5110664,0.06140276,0.2391573],"genre_scores_gemma":[0.01152322,0.01119728,0.0411831,0.003767368,0.002392936,0.001354031,0.8502707,0.002889219,0.07542215],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.06238117,"threshold_uncertainty_score":0.2086858,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2945319864","doi":"10.1145/3331651.3331653","title":"Urban Human Mobility","year":2019,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":107,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Multidisciplinary approach; Data science; Context (archaeology); Mobility model; Perspective (graphical); Field (mathematics); Global Positioning System; Urban computing; Human dynamics; Human–computer interaction; Artificial intelligence; Telecommunications; Sociology","authors":[{"name":"Jinzhong Wang","is_ca":false},{"name":"Xiangjie Kong","is_ca":false},{"name":"Lijun Sun","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0367557887767918,"gpt":0.3235828484170946,"spread":0.2868270596403028,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002519001,0.0003688185,0.0002175946,0.001098971,0.0006082617,0.001404399,0.0005197416,0.0005150872,0.009492499],"category_scores_gemma":[0.00133685,0.0001695871,0.0003505315,0.002411705,0.0005238373,0.002050304,0.001202847,0.0004736707,0.001712601],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001204019,"about_ca_system_score_gemma":0.0008329767,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00910354,"about_ca_topic_score_gemma":0.01072676,"domain_scores_codex":[0.9997316,0.00007308069,0.00001216038,0.00006772072,0.00006797424,0.0000474917],"domain_scores_gemma":[0.9997645,0.00006662401,0.00003391505,0.00003757349,0.00007662112,0.00002068],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00004255321,0.00003081391,0.01885827,0.0006605436,0.0000761215,0.0003142694,0.001400452,0.03941762,0.001048311,0.6163098,0.07855998,0.2432813],"study_design_scores_gemma":[0.00001097019,0.00005605209,0.02008269,0.0005067254,0.00006362519,0.0007751039,0.003324388,0.0594993,0.0009869995,0.1968902,0.7177368,0.00006724621],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1099405,0.03508145,0.2989787,0.01839554,0.001543793,0.0003538232,0.01508355,0.001487348,0.5191354],"genre_scores_gemma":[0.8824435,0.03045267,0.0278974,0.001085739,0.0004781792,0.0002627802,0.00694996,0.0001306654,0.05029897],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009492499,"threshold_uncertainty_score":0.03175551,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2160036998","doi":"10.1145/1233321.1233324","title":"Utility-based anonymization for privacy preservation with less information loss","year":2006,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":78,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"Chinese University of Hong Kong; National Natural Science Foundation of China","keywords":"Computer science; Data anonymization; Microdata (statistics); Categorical variable; Data mining; Identifier; Information loss; Heuristic; Set (abstract data type); Data quality; Information privacy; Machine learning; Artificial intelligence","authors":[{"name":"Jian Xu","is_ca":false},{"name":"Wei Wang","is_ca":false},{"name":"Jian Pei","is_ca":true},{"name":"Xiaoyuan Wang","is_ca":false},{"name":"Baile Shi","is_ca":false},{"name":"Ada Wai-Chee Fu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04348191414472581,"gpt":0.2574041941641516,"spread":0.2139222800194258,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007342007,0.0009197907,0.001634975,0.00196061,0.001595104,0.003251104,0.002193517,0.001144017,0.001750319],"category_scores_gemma":[0.02508089,0.0005945296,0.001440964,0.004297003,0.002608787,0.007508961,0.004141084,0.002829547,0.0006681969],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001447689,"about_ca_system_score_gemma":0.001911782,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006017312,"about_ca_topic_score_gemma":0.0006784833,"domain_scores_codex":[0.9878126,0.00686149,0.0006029273,0.001547904,0.002681677,0.0004934923],"domain_scores_gemma":[0.9751213,0.008528172,0.001423754,0.01318552,0.001483634,0.0002576655],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004822084,0.0001758619,0.002787054,0.0003093587,0.0002474714,0.0003432738,0.001177113,0.1560175,0.009561687,0.5484378,0.008912064,0.2715486],"study_design_scores_gemma":[0.00007661905,0.0001606835,0.0009779863,0.00008478982,0.000109391,0.0007547281,0.0004026119,0.4291779,0.01653515,0.5307015,0.02094604,0.00007263304],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01084704,0.0003966966,0.9854872,0.0005590122,0.00004686215,0.00008389771,0.0001672727,0.0003658328,0.00204616],"genre_scores_gemma":[0.4798862,0.0009647369,0.5133491,0.0004254837,0.0002706706,0.000408874,0.0008352729,0.0002614351,0.003598257],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007342007,"threshold_uncertainty_score":0.03882873,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1983506100","doi":"10.1145/568574.568581","title":"Exploiting succinct constraints using FP-trees","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Pruning; Focus (optics); Succinctness; Set (abstract data type); Constraint (computer-aided design); Tree (set theory); Data mining; Search tree; Theoretical computer science; Algorithm; Search algorithm; Programming language; Mathematics","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Raymond T. Ng","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1028758920195478,"gpt":0.27695920445099,"spread":0.1740833124314422,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004690676,0.001673721,0.001810835,0.004312415,0.001274286,0.002801389,0.002469158,0.001675728,0.002337971],"category_scores_gemma":[0.03180594,0.0009439907,0.001597816,0.005396755,0.00101359,0.006794352,0.002222359,0.001891945,0.001008427],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008563449,"about_ca_system_score_gemma":0.002400971,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00357419,"about_ca_topic_score_gemma":0.003539474,"domain_scores_codex":[0.9940457,0.001790674,0.0007010201,0.0006920156,0.002479104,0.000291416],"domain_scores_gemma":[0.9775859,0.01651764,0.00151171,0.001782625,0.002368566,0.0002334471],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000500514,0.0002163918,0.005002375,0.0007267724,0.0002196055,0.001230601,0.0004723508,0.1828321,0.006213338,0.07130087,0.009889776,0.7213953],"study_design_scores_gemma":[0.00007073287,0.00009748056,0.0003325344,0.0001270546,0.00006486352,0.0007152213,0.0001077212,0.8250488,0.007830206,0.1574425,0.008111482,0.00005131341],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00737023,0.0003073197,0.989903,0.0002476,0.00002936311,0.0001333499,0.0003866487,0.0008555032,0.0007670429],"genre_scores_gemma":[0.08521721,0.0004970855,0.9115506,0.0001632062,0.00005934118,0.0002926781,0.001543693,0.0001285799,0.0005476437],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004690676,"threshold_uncertainty_score":0.02480698,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1972132022","doi":"10.1145/1046456.1046468","title":"Subspace clustering for high dimensional categorical data","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":73,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Computer science; Clustering high-dimensional data; Categorical variable; Correlation clustering; CURE data clustering algorithm; Data mining; Subspace topology; Canopy clustering algorithm; Data stream clustering; Focus (optics); Algorithm; Pattern recognition (psychology); Artificial intelligence; Machine learning","authors":[{"name":"Guojun Gan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1123055424811128,"gpt":0.3448912619355348,"spread":0.232585719454422,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003667024,0.0007325431,0.0013952,0.003133809,0.001615953,0.001524821,0.001362129,0.001086037,0.001365948],"category_scores_gemma":[0.01125367,0.0003534379,0.001386236,0.004813941,0.001364131,0.001934973,0.00186994,0.001586164,0.000713491],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001175548,"about_ca_system_score_gemma":0.001714046,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003713113,"about_ca_topic_score_gemma":0.004320384,"domain_scores_codex":[0.9958442,0.001840004,0.0002245165,0.0005922827,0.00133916,0.000159775],"domain_scores_gemma":[0.9945497,0.00251713,0.0004290871,0.0009376222,0.001399775,0.0001666127],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001577147,0.00009320082,0.003772436,0.0005072683,0.0001755037,0.0002328453,0.0008529113,0.3639867,0.009032694,0.1766486,0.007019468,0.4375206],"study_design_scores_gemma":[0.000009494041,0.00005123314,0.0009633853,0.00002273448,0.00001552045,0.0001524013,0.0001588011,0.8882079,0.003188365,0.101581,0.005601386,0.00004775618],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002977177,0.0001891443,0.9962111,0.00008878906,0.00001316635,0.0000216117,0.00006211713,0.0001968504,0.0002400615],"genre_scores_gemma":[0.08437975,0.00045641,0.9132391,0.00006756825,0.00006028299,0.0001528903,0.0006502906,0.0001009036,0.0008928911],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003713113,"threshold_uncertainty_score":0.01939332,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3165211483","doi":"10.1145/3468507.3468511","title":"On the Applicability of Machine Learning Fairness Notions","year":2021,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":70,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"Agence Nationale de la Recherche","keywords":"Subjectivity; Set (abstract data type); Computer science; Artificial intelligence; Machine learning; Fairness measure; Epistemology","authors":[{"name":"Karima Makhlouf","is_ca":true},{"name":"Sami Zhioua","is_ca":false},{"name":"Catuscia Palamidessi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08874008202605167,"gpt":0.3517598061059928,"spread":0.2630197240799412,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04046299,0.001112392,0.001358533,0.003191134,0.003987904,0.006649208,0.002376873,0.003349282,0.004439565],"category_scores_gemma":[0.1189348,0.0005449083,0.001596481,0.002520363,0.01391042,0.01280024,0.005647481,0.007344269,0.0005563529],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004570568,"about_ca_system_score_gemma":0.003215683,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002863401,"about_ca_topic_score_gemma":0.001292878,"domain_scores_codex":[0.9651904,0.02075117,0.001511089,0.004062342,0.006583966,0.001901029],"domain_scores_gemma":[0.8317467,0.1450293,0.006539567,0.008170796,0.006534627,0.001979101],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00007646711,0.00004104744,0.001500467,0.00007125486,0.00002010748,0.00005189357,0.0003968925,0.01139375,0.000167106,0.9676468,0.0009345767,0.01769955],"study_design_scores_gemma":[0.00001616356,0.00004706016,0.0006826743,0.0001177595,0.00001196109,0.00006520068,0.0001595122,0.03639375,0.0002778718,0.9586048,0.003597314,0.0000258814],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06663588,0.006639157,0.8209251,0.02175014,0.000777714,0.000239966,0.0002746684,0.0002235898,0.08253382],"genre_scores_gemma":[0.8743927,0.00239099,0.1151718,0.002527869,0.001320004,0.0003399654,0.0001048635,0.0001249119,0.003626858],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.04046299,"threshold_uncertainty_score":0.2139913,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1993415883","doi":"10.1145/1809400.1809407","title":"FpVAT","year":2010,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":68,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Computer science; Process (computing); Knowledge extraction; Representation (politics); Data mining; Simple (philosophy); Visualization; Information retrieval; Visual analytics; Machine learning; Artificial intelligence","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Christopher L. Carmichael","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03376557419937707,"gpt":0.2960755426787169,"spread":0.2623099684793399,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001796819,0.001488022,0.00084176,0.0026379,0.0009174957,0.004232883,0.003864375,0.00180378,0.07432938],"category_scores_gemma":[0.01180415,0.0008645819,0.001927128,0.002036151,0.0006863873,0.005178381,0.003895302,0.002279335,0.04115142],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007818714,"about_ca_system_score_gemma":0.001272172,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002652686,"about_ca_topic_score_gemma":0.001966863,"domain_scores_codex":[0.9982285,0.0002658416,0.0001456486,0.0003589769,0.0007840805,0.0002170149],"domain_scores_gemma":[0.9967331,0.001200014,0.0001631766,0.000835768,0.0008630971,0.0002049071],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000739422,0.0001344196,0.001715311,0.0007669899,0.0001051411,0.0006059158,0.0005423429,0.004749477,0.005261746,0.04161418,0.4569957,0.4867694],"study_design_scores_gemma":[0.0002057069,0.0001408419,0.001060646,0.000259526,0.00006447329,0.00168928,0.0001557069,0.05405937,0.01161866,0.07385423,0.8567603,0.0001312288],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.005421641,0.001057066,0.626004,0.001216864,0.001025812,0.0005327042,0.01721902,0.2744164,0.07310648],"genre_scores_gemma":[0.1067368,0.001825817,0.7053497,0.001599934,0.0004968328,0.002032246,0.06073688,0.05176189,0.06945979],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.07432938,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1968064480","doi":"10.1145/2031331.2031336","title":"Trajectory anonymity in publishing personal mobility data","year":2011,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":68,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Data publishing; Identifier; Anonymity; Global Positioning System; TRACE (psycholinguistics); Location-based service; Computer security; Object (grammar); World Wide Web; GSM; Mobile device; Location data; Personally identifiable information; k-anonymity; Internet privacy; Data science; Publishing; Computer network; Telecommunications","authors":[{"name":"Francesco Bonchi","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Hui Wang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2671506687550872,"gpt":0.3419145920846799,"spread":0.07476392332959275,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01821933,0.0006490466,0.001833009,0.003221767,0.00388941,0.008955074,0.003697078,0.002358919,0.002228262],"category_scores_gemma":[0.06870109,0.0009008547,0.001336283,0.009061621,0.003936021,0.02051424,0.008629141,0.003220022,0.001555885],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002392814,"about_ca_system_score_gemma":0.003512637,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001328973,"about_ca_topic_score_gemma":0.0007051249,"domain_scores_codex":[0.9710475,0.01363624,0.003250269,0.0023148,0.008446764,0.001304571],"domain_scores_gemma":[0.8948246,0.0371167,0.007809563,0.05078919,0.008062039,0.001397982],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001022019,0.0001999134,0.01163599,0.0008084534,0.0002132504,0.0008957203,0.00493176,0.04349284,0.006287502,0.6039342,0.01092174,0.3156567],"study_design_scores_gemma":[0.0001293124,0.0004723945,0.002919971,0.0004853025,0.0002875659,0.003940495,0.003851346,0.2416645,0.04948181,0.556316,0.1401921,0.0002593622],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0414396,0.002766902,0.9369996,0.003923298,0.0004612191,0.0002825004,0.0009068308,0.001275149,0.01194487],"genre_scores_gemma":[0.7040697,0.005570541,0.2755642,0.000849315,0.001211254,0.0005882494,0.001696071,0.0003486509,0.01010201],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01821933,"threshold_uncertainty_score":0.09635413,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2039552226","doi":"10.1145/2207243.2207248","title":"Mapping question items to skills with non-negative matrix factorization","year":2012,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Intelligent Tutoring Systems and Adaptive Learning","field":"Computer Science","cited_by":67,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Computer science; Non-negative matrix factorization; Matrix decomposition; Task (project management); Variance (accounting); Machine learning; Process (computing); Matrix (chemical analysis); Artificial intelligence; Automation; Data science","authors":[{"name":"Michel C. Desmarais","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02240098912766123,"gpt":0.2745934823823906,"spread":0.2521924932547293,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007474936,0.002060758,0.001011324,0.001906229,0.0005919309,0.001467536,0.0011451,0.001858723,0.002815559],"category_scores_gemma":[0.03929055,0.0007632956,0.002457541,0.001715986,0.001418752,0.001554636,0.001183552,0.00300831,0.001069585],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001117407,"about_ca_system_score_gemma":0.001551792,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009094439,"about_ca_topic_score_gemma":0.00693582,"domain_scores_codex":[0.9954235,0.00244532,0.000215058,0.001200402,0.000491961,0.000223837],"domain_scores_gemma":[0.9726719,0.02299178,0.001262225,0.00147108,0.001403088,0.0001998755],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006203665,0.000852518,0.01733246,0.0008651331,0.0005994539,0.0004034456,0.001286469,0.4605456,0.01021635,0.02235271,0.006480274,0.4784452],"study_design_scores_gemma":[0.00005582652,0.0001371239,0.005374594,0.00004156988,0.00003078984,0.00007116574,0.00008927819,0.9586881,0.001391768,0.03250131,0.001576199,0.00004229552],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02926227,0.0001783994,0.968937,0.0002343523,0.0000563459,0.0002577236,0.0003075694,0.0003045694,0.0004618172],"genre_scores_gemma":[0.3110607,0.0002152216,0.6846852,0.0001965315,0.00009333855,0.0009674554,0.001520421,0.0000926575,0.001168454],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009094439,"threshold_uncertainty_score":0.03953165,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2128899929","doi":"10.1145/2207243.2207247","title":"Social network analysis and mining to support the assessment of on-line student participation","year":2012,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Automatic summarization; Computer science; Toolbox; Social network analysis; Identification (biology); Visualization; Data science; World Wide Web; Exploit; Social media; Online discussion; Information retrieval; Data mining","authors":[{"name":"Reihaneh Rabbany k.","is_ca":true},{"name":"Mansoureh Takaffoli","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.076759501794539,"gpt":0.4074878988997948,"spread":0.3307283971052558,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003451931,0.0007747635,0.0005087633,0.00840867,0.0005463275,0.001097539,0.0008988763,0.0006392974,0.002034631],"category_scores_gemma":[0.01566686,0.000228976,0.0004933882,0.004083227,0.0002934235,0.001742421,0.001298564,0.0005982706,0.0007183402],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005031094,"about_ca_system_score_gemma":0.0005884621,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002032433,"about_ca_topic_score_gemma":0.003736563,"domain_scores_codex":[0.9973477,0.001286623,0.0002473943,0.0003682208,0.0006621255,0.00008784772],"domain_scores_gemma":[0.9850154,0.009319642,0.002234639,0.001310874,0.001568147,0.0005512285],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.000511169,0.001127446,0.17254,0.0006236427,0.0005179383,0.0005192722,0.002883145,0.05422796,0.02013442,0.01201097,0.0121467,0.7227573],"study_design_scores_gemma":[0.00004461997,0.0002611475,0.06711384,0.00009345093,0.00007684124,0.0004265814,0.001404785,0.884607,0.01147069,0.01974942,0.01468364,0.00006788074],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3007649,0.000399261,0.6794255,0.000872199,0.00009263673,0.0007819736,0.005055937,0.005899176,0.00670833],"genre_scores_gemma":[0.6293612,0.0002195218,0.3643605,0.0000552939,0.00006315135,0.0005083874,0.003547214,0.0001383548,0.001746396],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00840867,"threshold_uncertainty_score":0.01825577,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2034607275","doi":"10.1145/1046456.1046459","title":"Learning important models for web page blocks based on layout and content analysis","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Web page; Block (permutation group theory); Information retrieval; Partition (number theory); Page view; Static web page; Data mining; World Wide Web; Web navigation","authors":[{"name":"Ruihua Song","is_ca":false},{"name":"Haifeng Liu","is_ca":true},{"name":"Ji-Rong Wen","is_ca":false},{"name":"Wei‐Ying Ma","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07156513468895788,"gpt":0.263723749372729,"spread":0.1921586146837712,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00182862,0.001169123,0.001306003,0.004589635,0.0004590129,0.001424245,0.001739314,0.001763806,0.001474455],"category_scores_gemma":[0.00860346,0.0008015702,0.001885037,0.001680219,0.0008838507,0.003108758,0.0007338889,0.002000357,0.0009379449],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001595592,"about_ca_system_score_gemma":0.0007554874,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006761526,"about_ca_topic_score_gemma":0.01195777,"domain_scores_codex":[0.9991167,0.0002434251,0.00006878802,0.0002612631,0.0002004821,0.000109279],"domain_scores_gemma":[0.9950523,0.003272102,0.0004518704,0.0003802472,0.0006836073,0.0001599412],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001016153,0.0008458124,0.02730521,0.0003148658,0.0002784122,0.0003444196,0.0004048986,0.621131,0.0125158,0.01028108,0.005888219,0.3196743],"study_design_scores_gemma":[0.00000810925,0.0000224061,0.0008096609,0.000006287091,0.00001491555,0.0000302837,0.000010084,0.9954496,0.0007566248,0.002700908,0.0001859514,0.000005177646],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1488709,0.0007031263,0.8453839,0.0003921349,0.00006289181,0.0002692192,0.0005604704,0.002684837,0.001072639],"genre_scores_gemma":[0.7508843,0.0004855798,0.2420981,0.000170843,0.0001511158,0.0004258105,0.00228132,0.0002894324,0.00321359],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006761526,"threshold_uncertainty_score":0.0134443,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2155709205","doi":"10.1145/846183.846209","title":"WEBKDD'99","year":2000,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"York University; DePaul University","keywords":"Computer science; Profiling (computer programming); Knowledge extraction; Data science; World Wide Web; Information retrieval; Data mining","authors":[{"name":"Brij Masand","is_ca":false},{"name":"Myra Spiliopoulou","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02750424736256481,"gpt":0.2548875557424921,"spread":0.2273833083799272,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01242737,0.002564441,0.004983031,0.01268518,0.003636714,0.01476485,0.007351781,0.004203286,0.0344637],"category_scores_gemma":[0.03654352,0.002892656,0.004261038,0.00886555,0.00119494,0.006689289,0.007341637,0.006464954,0.07641745],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00230556,"about_ca_system_score_gemma":0.008189315,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01263386,"about_ca_topic_score_gemma":0.01669731,"domain_scores_codex":[0.9862482,0.002469296,0.001514615,0.001763685,0.007462013,0.0005420889],"domain_scores_gemma":[0.9826435,0.004912077,0.0005288849,0.006073744,0.004557343,0.001284409],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000105752,0.00009964802,0.0006273994,0.0005657977,0.000191962,0.00009563291,0.00005632605,0.002290422,0.0001741904,0.006618697,0.9157326,0.07344157],"study_design_scores_gemma":[0.0002251111,0.00002655597,0.0007179374,0.0001788008,0.0000748536,0.0002712271,0.00005099546,0.01895625,0.001605171,0.03288684,0.9449236,0.00008252347],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.003584557,0.01619981,0.2606511,0.01805268,0.01018537,0.001577161,0.3695178,0.2479245,0.07230699],"genre_scores_gemma":[0.01615906,0.01170835,0.1969053,0.004780677,0.0009781393,0.002696942,0.7118547,0.01478229,0.04013462],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.0344637,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2736013904","doi":"10.1145/1809400.1809425","title":"Efficient algorithms for the mining of constrained frequent patterns from uncertain data","year":2010,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Computer science; Association rule learning; Data mining; K-optimal pattern discovery; Database transaction; Uncertain data; Data stream mining; Knowledge extraction; Geodetic datum; Data science; Database","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Dale A. Brajczuk","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1036399997101661,"gpt":0.321903110243897,"spread":0.2182631105337309,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00378502,0.001490798,0.002570947,0.003616424,0.001296984,0.002364431,0.003620428,0.001874689,0.001545415],"category_scores_gemma":[0.01959967,0.001142373,0.001602991,0.006404051,0.0009693201,0.00493014,0.002852098,0.00217554,0.0009719052],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007637398,"about_ca_system_score_gemma":0.001761341,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001768512,"about_ca_topic_score_gemma":0.002557092,"domain_scores_codex":[0.9967042,0.0007951105,0.0004878687,0.0006295961,0.001174269,0.0002090059],"domain_scores_gemma":[0.9891756,0.007793809,0.0008052234,0.00100861,0.001055772,0.0001609616],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002913763,0.0003083281,0.002659361,0.0006080969,0.0002565883,0.0004054123,0.000387399,0.2517956,0.002639309,0.02984852,0.007195931,0.703604],"study_design_scores_gemma":[0.00009680348,0.00009077478,0.0004763338,0.00005803023,0.00004814102,0.0004053732,0.0001182334,0.9227229,0.001342403,0.07061002,0.004003501,0.00002757037],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00493444,0.0008094885,0.9929756,0.0001850233,0.00003604019,0.0001531037,0.0001907698,0.0003481932,0.0003672648],"genre_scores_gemma":[0.04114702,0.0007305404,0.9560835,0.00009062207,0.00009723192,0.0004542032,0.0009206558,0.00005033215,0.0004257792],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00378502,"threshold_uncertainty_score":0.02001733,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1965269329","doi":"10.1145/1147234.1147235","title":"Introduction to the special issue on successful real-world data mining applications","year":2006,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Data science; Field (mathematics); Database transaction; Transaction data; Knowledge extraction; Data discovery; Data stream mining; Data mining; World Wide Web; Database; Metadata","authors":[{"name":"Gabor Melli","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Brendan Kitts","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03975452893360564,"gpt":0.2972189779695197,"spread":0.257464449035914,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004038095,0.002147446,0.00216389,0.005425847,0.0018374,0.008494624,0.00192076,0.00311956,0.1333086],"category_scores_gemma":[0.01359921,0.0007724827,0.001364082,0.005363408,0.001058865,0.007381915,0.003215316,0.003862701,0.0996301],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009681179,"about_ca_system_score_gemma":0.001558425,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003454486,"about_ca_topic_score_gemma":0.001080792,"domain_scores_codex":[0.9961204,0.0005123159,0.000396575,0.0005080903,0.00228125,0.0001813088],"domain_scores_gemma":[0.9813572,0.007568348,0.001109687,0.001450196,0.00590077,0.002613738],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000009217181,0.00002209169,0.00006741856,0.0001677842,0.000006405303,0.00002823305,0.000014487,0.00004456528,0.0001145068,0.0006016301,0.9650856,0.03383804],"study_design_scores_gemma":[0.000005994305,0.00003102175,0.00024642,0.0001570598,0.000008654823,0.0001368899,0.00002444053,0.0001348304,0.00008125543,0.00154577,0.9976192,0.000008440879],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"editorial","genre_gemma":"editorial","genre_scores_codex":[0.001002881,0.07082165,0.02091778,0.07909087,0.6991462,0.0007269252,0.003813791,0.002623769,0.1218562],"genre_scores_gemma":[0.003426979,0.05357733,0.01030057,0.02150859,0.6522149,0.0003956907,0.004738772,0.001940027,0.2518971],"genre_candidate":"editorial","genre_consensus":"editorial","teacher_disagreement_score":0.1333086,"threshold_uncertainty_score":0.4459618,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3034271075","doi":"10.1145/3262579","title":"Session details: Special issue on learning from imbalanced datasets","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; Canadian Imperial Bank of Commerce (Canada)","funders":"","keywords":"Computer science; Session (web analytics); Machine learning; Data science; Artificial intelligence; World Wide Web","authors":[{"name":"Nitesh V. Chawla","is_ca":true},{"name":"Nathalie Japkowicz","is_ca":true},{"name":"Aleksander Kołcz","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03792720065423506,"gpt":0.2847760692288678,"spread":0.2468488685746328,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.003181774,0.002045957,0.003099862,0.002117321,0.001941879,0.006545901,0.001616411,0.003306417,0.5244928],"category_scores_gemma":[0.006777999,0.0006045218,0.001405022,0.003198714,0.0004625297,0.004947979,0.002577284,0.003764578,0.3605936],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006748054,"about_ca_system_score_gemma":0.001423357,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007136277,"about_ca_topic_score_gemma":0.002192729,"domain_scores_codex":[0.9988193,0.000186639,0.00007176059,0.0003143375,0.0004737676,0.0001341054],"domain_scores_gemma":[0.9933069,0.001732783,0.0002323425,0.0007236471,0.00211855,0.00188583],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004990388,0.00003129237,0.0000417029,0.00008005273,0.000006453767,0.00001520971,0.000005183033,0.00003176309,0.0002725272,0.0002055576,0.9820238,0.01723673],"study_design_scores_gemma":[0.00009478854,0.0001426906,0.000977741,0.0001468146,0.00003161931,0.0001440572,0.00004581004,0.0007412262,0.0008753337,0.002964033,0.9938048,0.00003103454],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"editorial","genre_gemma":"editorial","genre_scores_codex":[0.004856523,0.02573356,0.05487802,0.08053809,0.4036677,0.003263497,0.0472072,0.01762188,0.3622335],"genre_scores_gemma":[0.008774329,0.011766,0.01002053,0.007562592,0.1155018,0.000975743,0.03022941,0.004211358,0.8109583],"genre_candidate":"editorial","genre_consensus":"editorial","teacher_disagreement_score":0.5244928,"threshold_uncertainty_score":0.6782533,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2085995514","doi":"10.1145/772862.772888","title":"WEBKDD 2002","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Conjunction (astronomy); Theme (computing); Knowledge extraction; Data science; Web mining; Data mining; Information retrieval; World Wide Web; Web service","authors":[{"name":"Brij Masand","is_ca":false},{"name":"Myra Spiliopoulou","is_ca":false},{"name":"Jaideep Srivastava","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06108137065648202,"gpt":0.2461904553370433,"spread":0.1851090846805613,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009158089,0.002114617,0.00431655,0.008350158,0.002323132,0.01012818,0.005940833,0.003657967,0.02785235],"category_scores_gemma":[0.03009435,0.00271893,0.003165991,0.007292266,0.0007663778,0.005103022,0.003918403,0.005329791,0.05528815],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002155068,"about_ca_system_score_gemma":0.008490472,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01189358,"about_ca_topic_score_gemma":0.01614295,"domain_scores_codex":[0.9908513,0.001707069,0.001269858,0.001132372,0.004686547,0.0003527501],"domain_scores_gemma":[0.9868537,0.003757763,0.0005820954,0.003962881,0.00382758,0.001015962],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001203183,0.00007616068,0.0006886384,0.0007788946,0.0001958773,0.00009822108,0.00003449061,0.002502848,0.0001749274,0.004592247,0.9365782,0.05415916],"study_design_scores_gemma":[0.0002411882,0.00002360891,0.0007189346,0.000184618,0.00008690448,0.0003671507,0.00002975757,0.01470752,0.001595574,0.01860962,0.9633626,0.00007251176],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.003637423,0.0158106,0.1920556,0.02092851,0.008032216,0.001163515,0.5552667,0.1601573,0.04294815],"genre_scores_gemma":[0.01225516,0.00993473,0.1293986,0.004701413,0.0007423991,0.001813486,0.8075586,0.007987105,0.02560854],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.9721476,"threshold_uncertainty_score":0.09317541,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2112993176","doi":"10.1145/1294301.1294305","title":"Introduction to the special issue on data mining for health informatics","year":2007,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":18,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"","keywords":"Personalized medicine; Computer science; Profiling (computer programming); Computational biology; Informatics; Metabolite profiling; Biomedicine; Data science; Oligonucleotide; Bioinformatics; Data mining; DNA; Biology; Metabolomics; Genetics","authors":[{"name":"Raymond T. Ng","is_ca":true},{"name":"Jian Pei","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05493152489629841,"gpt":0.3437162114384144,"spread":0.288784686542116,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006042525,0.002315043,0.003541037,0.006679961,0.001574657,0.008875033,0.002482741,0.004387085,0.09214429],"category_scores_gemma":[0.01951091,0.001149926,0.002407814,0.00620027,0.001500689,0.009598449,0.004117718,0.008608382,0.06871255],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001188677,"about_ca_system_score_gemma":0.001592281,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007252396,"about_ca_topic_score_gemma":0.001579819,"domain_scores_codex":[0.9959289,0.0009555925,0.0005478277,0.0007899681,0.001618359,0.000159361],"domain_scores_gemma":[0.9759517,0.01389591,0.0008299267,0.001854375,0.005071844,0.002396293],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000121722,0.0000210324,0.00006489298,0.000192887,0.00001756339,0.00003240387,0.00001496529,0.00008782917,0.0001285615,0.001372055,0.9506359,0.04741964],"study_design_scores_gemma":[0.00001243219,0.00002462443,0.0002496753,0.0002341357,0.0000125058,0.0001926598,0.00002283388,0.0005052029,0.00007816074,0.005784146,0.9928656,0.00001809184],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"editorial","genre_gemma":"editorial","genre_scores_codex":[0.0004011689,0.08150359,0.08007386,0.1364696,0.6471421,0.0006250517,0.003915236,0.004621704,0.04524778],"genre_scores_gemma":[0.001851833,0.0542386,0.03101238,0.05941332,0.7044128,0.0005518081,0.00607046,0.00265989,0.1397889],"genre_candidate":"editorial","genre_consensus":"editorial","teacher_disagreement_score":0.09214429,"threshold_uncertainty_score":0.3082533,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2122949122","doi":"10.1145/507515.507524","title":"MDM/KDD","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Data science; Work (physics); Information retrieval; World Wide Web","authors":[{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Simeon Simoff","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0716650625786421,"gpt":0.2493575275794745,"spread":0.1776924650008325,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004280315,0.00136471,0.002630912,0.008692591,0.001884302,0.00835224,0.004285601,0.002396629,0.06591492],"category_scores_gemma":[0.01683063,0.000991069,0.001399182,0.006152758,0.0007946876,0.005146033,0.005635313,0.001919479,0.07017294],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001061165,"about_ca_system_score_gemma":0.002035378,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002137613,"about_ca_topic_score_gemma":0.002999631,"domain_scores_codex":[0.9930457,0.0009516705,0.0006179529,0.001050482,0.003977829,0.000356396],"domain_scores_gemma":[0.9930131,0.001037075,0.0002387689,0.003425978,0.001853718,0.0004313618],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002000347,0.0001137823,0.001570188,0.0009662665,0.0001606297,0.0001974617,0.00008883083,0.002583215,0.001075985,0.01759438,0.6694021,0.3060472],"study_design_scores_gemma":[0.00008239333,0.00003427221,0.001247162,0.0001776456,0.00004491883,0.0006399453,0.0001699353,0.01906053,0.003562905,0.03440709,0.9405256,0.00004752208],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.009459023,0.01625342,0.4069296,0.02064378,0.01227831,0.001708729,0.1546187,0.1154664,0.262642],"genre_scores_gemma":[0.103515,0.01608327,0.5283687,0.005693034,0.003774759,0.002329188,0.2256157,0.009914721,0.1047057],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.06591492,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2067575620","doi":"10.1145/772862.772886","title":"MDM/KDD2002","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Conjunction (astronomy); Knowledge extraction; Data science; Series (stratigraphy); Information retrieval; Data mining; Geology","authors":[{"name":"Simeon Simoff","is_ca":false},{"name":"Chabane Djeraba","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07236190396419816,"gpt":0.2474651348445299,"spread":0.1751032308803317,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005616736,0.002311852,0.00350434,0.009831926,0.002148738,0.007460354,0.005816385,0.003082678,0.02823519],"category_scores_gemma":[0.01753741,0.001441848,0.002019803,0.007913313,0.0006011191,0.0044316,0.004614318,0.002766304,0.04392592],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001551625,"about_ca_system_score_gemma":0.003116563,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007160733,"about_ca_topic_score_gemma":0.007915754,"domain_scores_codex":[0.9932119,0.0009729071,0.0006590746,0.0007721998,0.004033761,0.0003500833],"domain_scores_gemma":[0.9925712,0.001066382,0.0001944889,0.0030609,0.002610338,0.0004966406],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001288439,0.00009661775,0.000518774,0.000438071,0.0001244148,0.0001127965,0.00004041995,0.002436752,0.0003751926,0.004833978,0.8421062,0.1487879],"study_design_scores_gemma":[0.0001327462,0.00003842268,0.0009465482,0.0001370609,0.00005762353,0.0004189527,0.00007865416,0.02519308,0.002918023,0.02040398,0.9496248,0.00005014633],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.009912218,0.02000324,0.3355142,0.02069417,0.0148027,0.002056382,0.30064,0.1743626,0.1220144],"genre_scores_gemma":[0.03283739,0.01052741,0.3438961,0.003487657,0.001768438,0.00205232,0.545409,0.007939018,0.05208261],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.02823519,"threshold_uncertainty_score":0.09445614,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2024539262","doi":"10.1145/1294301.1294307","title":"Novel approaches for small biomolecule classification and structural similarity search","year":2007,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Nearest neighbor search; Classifier (UML); Minkowski distance; Pruning; Similarity (geometry); Artificial intelligence; Data mining; Pattern recognition (psychology); Binary classification; k-nearest neighbors algorithm; Machine learning; Euclidean distance; Support vector machine","authors":[{"name":"Emre Karakoç","is_ca":true},{"name":"Artem Cherkasov","is_ca":false},{"name":"S. Cenk Şahinalp","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2537296575736704,"gpt":0.3552435413062274,"spread":0.101513883732557,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000973081,0.0007591496,0.001625753,0.002292727,0.0005619989,0.001520207,0.002178154,0.001387024,0.002425392],"category_scores_gemma":[0.002324812,0.0004843497,0.001263856,0.002636962,0.0009136142,0.002064306,0.001396073,0.00134822,0.001372974],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001041079,"about_ca_system_score_gemma":0.001063423,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001547434,"about_ca_topic_score_gemma":0.001788121,"domain_scores_codex":[0.998718,0.0001491555,0.0001090109,0.0002868761,0.0006678045,0.00006914711],"domain_scores_gemma":[0.9990526,0.0003024024,0.0001331543,0.0001890461,0.0002816392,0.00004110203],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001386807,0.0002353808,0.001214688,0.0005388155,0.0001149176,0.0001429185,0.0001200314,0.1285105,0.03283333,0.09315337,0.004377791,0.7386196],"study_design_scores_gemma":[0.00002117802,0.00006833074,0.0004464665,0.00002901313,0.00002249931,0.0001468983,0.00002797952,0.926623,0.008486012,0.05299591,0.01111011,0.00002259455],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003774489,0.0006022376,0.9937596,0.00009531013,0.00004454398,0.00005871624,0.00005585181,0.0004177853,0.001191635],"genre_scores_gemma":[0.05684989,0.000765164,0.9381091,0.0001556606,0.000107559,0.0002510637,0.0004422839,0.0001135869,0.00320565],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002425392,"threshold_uncertainty_score":0.008113801,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2627710665","doi":"10.1145/772862.772887","title":"Multi-relational data mining","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Computer science; Knowledge extraction; Data science; Relational database; Data mining; Information retrieval","authors":[{"name":"Sašo Džeroski","is_ca":false},{"name":"Luc De Raedt","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2564192870307092,"gpt":0.3081369896490975,"spread":0.05171770261838832,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006857553,0.001218159,0.001647424,0.005641037,0.001064944,0.005854344,0.003925705,0.001098868,0.005143336],"category_scores_gemma":[0.02124682,0.0006189688,0.001897475,0.009295062,0.0007344011,0.009080381,0.003884666,0.002009049,0.00588246],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008066392,"about_ca_system_score_gemma":0.001640217,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009818592,"about_ca_topic_score_gemma":0.001252073,"domain_scores_codex":[0.9904523,0.001955686,0.001170302,0.002036065,0.004117121,0.0002684682],"domain_scores_gemma":[0.9893662,0.00481165,0.0008931778,0.002504222,0.002101863,0.0003228285],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001303465,0.0001803224,0.007866713,0.002610154,0.0005474271,0.001015412,0.0003367455,0.007883908,0.003798897,0.07670861,0.04730609,0.8516154],"study_design_scores_gemma":[0.00005584606,0.0002371599,0.004233899,0.001054388,0.0005908032,0.007599789,0.0007719019,0.1293656,0.01538532,0.2542057,0.5863222,0.0001773451],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003711064,0.02772629,0.9485956,0.003167174,0.0007881157,0.0005365558,0.003351207,0.00255208,0.009572045],"genre_scores_gemma":[0.06739901,0.02847978,0.8844896,0.001813847,0.001195457,0.0004931945,0.01004433,0.0003927262,0.00569206],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006857553,"threshold_uncertainty_score":0.03626668,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2028572284","doi":"10.1145/846183.846208","title":"Discovering geographic knowledge in data rich environments","year":2000,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"City, University of London; University of Utah; University of Florida; Florida State University","keywords":"Data science; Identification (biology); Computer science; Geographic information system; Knowledge extraction; Data mining; Geography; Cartography","authors":[{"name":"Harvey J. Miller","is_ca":false},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05390222068445348,"gpt":0.2862366298714703,"spread":0.2323344091870168,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001199852,0.0003976022,0.000814627,0.003248507,0.001487177,0.003877477,0.00107569,0.001293856,0.004283152],"category_scores_gemma":[0.006962259,0.0005834575,0.0006649494,0.004981956,0.001336185,0.005466469,0.004656455,0.001232277,0.001412812],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001081499,"about_ca_system_score_gemma":0.0008717801,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004602766,"about_ca_topic_score_gemma":0.01386598,"domain_scores_codex":[0.9989304,0.0002774399,0.00006953018,0.0002622918,0.0003460106,0.0001142323],"domain_scores_gemma":[0.9979961,0.001169249,0.0001426311,0.0002838821,0.000290106,0.0001181228],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005830548,0.00021165,0.02017882,0.001362105,0.0002173122,0.003168229,0.00504635,0.08456948,0.004602861,0.1940073,0.09891484,0.5871381],"study_design_scores_gemma":[0.0000683147,0.0001017977,0.009708973,0.0004198844,0.0001285509,0.0009594487,0.007106631,0.1215325,0.006282609,0.2685934,0.5849556,0.0001423363],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2103834,0.01494717,0.6741806,0.0212938,0.001726138,0.0005451472,0.03311635,0.004486203,0.03932123],"genre_scores_gemma":[0.5976043,0.01032767,0.342886,0.001383553,0.000411323,0.0002570696,0.03137835,0.0002170008,0.0155348],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004602766,"threshold_uncertainty_score":0.0143286,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2082225138","doi":"10.1145/1046456.1046458","title":"Extracting relational data from HTML repositories","year":2004,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Set (abstract data type); Information retrieval; Data mining; Information extraction; Quality (philosophy)","authors":[{"name":"Ruth Yuee Zhang","is_ca":true},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Ruben H. Zamar","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1297204511692792,"gpt":0.3005306323875561,"spread":0.170810181218277,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002320453,0.0009021745,0.001054459,0.01087782,0.0007095666,0.003061582,0.001582568,0.0009650076,0.001158987],"category_scores_gemma":[0.01986454,0.0007102127,0.001321234,0.009080175,0.0004567526,0.003221844,0.001894418,0.0009342164,0.001922132],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005577767,"about_ca_system_score_gemma":0.001329678,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002656304,"about_ca_topic_score_gemma":0.004033797,"domain_scores_codex":[0.9969796,0.0004648877,0.000426762,0.0003990147,0.001579481,0.0001502828],"domain_scores_gemma":[0.9853367,0.00596384,0.001358155,0.004004272,0.003000018,0.000336957],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003661885,0.0003859211,0.04202877,0.001449787,0.0002699603,0.001771433,0.001875047,0.02556974,0.04613278,0.008904273,0.01632897,0.8549171],"study_design_scores_gemma":[0.0001323311,0.0004619152,0.03943773,0.0004592144,0.000741255,0.00374834,0.002586659,0.5392147,0.2862467,0.04468074,0.08199382,0.0002966518],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1676576,0.001911842,0.7773164,0.000998979,0.00007092505,0.0006377461,0.02084624,0.0251,0.005460272],"genre_scores_gemma":[0.2414712,0.001131223,0.7131192,0.0001314196,0.00007759978,0.0002576243,0.0415202,0.0008820247,0.001409501],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01087782,"threshold_uncertainty_score":0.01227188,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4406939872","doi":"10.1145/3715073.3715081","title":"Graph Fairness via Authentic Counterfactuals: Tackling Structural and Causal Challenges","year":2025,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Counterfactual conditional; Computer science; Psychology; Counterfactual thinking; Social psychology","authors":[{"name":"Zichong Wang","is_ca":false},{"name":"Zhipeng Yin","is_ca":false},{"name":"Wenbin Zhang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05891837359566172,"gpt":0.3637102994061234,"spread":0.3047919258104617,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.08245675,0.001623039,0.002142012,0.00300114,0.002235653,0.004892112,0.004457165,0.003354891,0.003295373],"category_scores_gemma":[0.2617899,0.001299294,0.001830786,0.002524874,0.008828229,0.00852803,0.007319254,0.005969099,0.0003535529],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003166756,"about_ca_system_score_gemma":0.005092664,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00235988,"about_ca_topic_score_gemma":0.002360548,"domain_scores_codex":[0.931234,0.05669592,0.00124119,0.005804467,0.004237184,0.0007872095],"domain_scores_gemma":[0.6355671,0.3125336,0.01220413,0.0321883,0.005926295,0.001580522],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0004140274,0.0001731234,0.007240899,0.0003937298,0.0002950126,0.0004008133,0.001452419,0.1815886,0.0007726151,0.7136476,0.001902508,0.09171874],"study_design_scores_gemma":[0.00006443323,0.00005552564,0.0003171825,0.00005939062,0.00004507948,0.00006045887,0.00008224523,0.2882572,0.0006748924,0.7087042,0.001658986,0.00002045121],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01517421,0.0002738213,0.981253,0.001414918,0.00007178869,0.000256797,0.0001037998,0.0001832524,0.001268439],"genre_scores_gemma":[0.5662485,0.0004239454,0.4295262,0.0009566079,0.0002879954,0.0009703605,0.0002707197,0.0001622836,0.001153427],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.08245675,"threshold_uncertainty_score":0.4360782,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3112675244","doi":"10.1145/3447556.3447563","title":"A Report on the First Workshop on Document Intelligence (DI) at NeurIPS 2019","year":2021,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Computer science; Library science; Intelligence analysis; Data science; World Wide Web; Computer security","authors":[{"name":"Hamid Motahari","is_ca":false},{"name":"Nigel Duffy","is_ca":false},{"name":"Paul N. Bennett","is_ca":false},{"name":"Tania Bedrax-Weiss","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05144086556164734,"gpt":0.2786518561287233,"spread":0.2272109905670759,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01424333,0.00176705,0.001271056,0.00365521,0.003021276,0.01260311,0.002404891,0.003095025,0.1061409],"category_scores_gemma":[0.01497179,0.0007164805,0.001613607,0.003529062,0.001021443,0.0075982,0.007569919,0.005346602,0.08640075],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004309299,"about_ca_system_score_gemma":0.01165435,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01395496,"about_ca_topic_score_gemma":0.01981775,"domain_scores_codex":[0.9882081,0.002386011,0.0004642483,0.001387515,0.00591398,0.001640099],"domain_scores_gemma":[0.977434,0.002334068,0.0003844204,0.001272003,0.01118753,0.00738808],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002511496,0.0001751448,0.0004981045,0.0002844417,0.00002017323,0.0001120812,0.000308093,0.0002483977,0.002287738,0.003005009,0.9077805,0.08502915],"study_design_scores_gemma":[0.00002993066,0.00008617953,0.0006569277,0.0001358339,0.00001466792,0.00006385956,0.0003720102,0.0003112793,0.001646847,0.001358222,0.9952968,0.00002753048],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.02140272,0.04027476,0.0886662,0.1402938,0.2195899,0.004308079,0.03743915,0.007245475,0.44078],"genre_scores_gemma":[0.02335778,0.0134505,0.03590702,0.008208128,0.01172727,0.001079551,0.04172464,0.003098549,0.8614466],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.1061409,"threshold_uncertainty_score":0.3550768,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2039760999","doi":"10.1145/380995.381054","title":"Report on MDM/KDD2000","year":2000,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Citation; Library science; Computer science; World Wide Web","authors":[{"name":"Simeon Simoff","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03370207834165126,"gpt":0.2762263369441018,"spread":0.2425242586024506,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008331763,0.001751476,0.001526762,0.006836616,0.001576617,0.005989876,0.003133589,0.001757236,0.1530601],"category_scores_gemma":[0.01894994,0.000706418,0.0009306782,0.005960707,0.0004801991,0.004888448,0.00358117,0.002004829,0.1890418],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002183932,"about_ca_system_score_gemma":0.003771511,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009454688,"about_ca_topic_score_gemma":0.008069629,"domain_scores_codex":[0.9935681,0.0007583573,0.0003978781,0.0008117711,0.004075146,0.0003888296],"domain_scores_gemma":[0.9820747,0.00165507,0.0003162716,0.00369694,0.01015488,0.002102148],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001297546,0.0001003864,0.0002517113,0.0001036708,0.00001402778,0.00003027178,0.00001871726,0.0002022399,0.0004784981,0.001616446,0.9182311,0.07882307],"study_design_scores_gemma":[0.00004906712,0.00004190286,0.0006857741,0.00005802642,0.00001292616,0.00005234157,0.00003374923,0.0008921779,0.001689017,0.001654089,0.9948178,0.00001301547],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.01065248,0.02142335,0.09096522,0.04937772,0.05367302,0.003515792,0.2750656,0.03650503,0.4588218],"genre_scores_gemma":[0.01625372,0.01106213,0.06689351,0.005005975,0.007335693,0.001424988,0.4198779,0.007038139,0.4651079],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.1530601,"threshold_uncertainty_score":0.512037,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2110788923","doi":"10.1145/1117454.1117475","title":"WebKDD 2005","year":2005,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Theme (computing); Conjunction (astronomy); Web mining; World Wide Web; Data science; Knowledge extraction; Data mining; Web service","authors":[{"name":"Olfa Nasraoui","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Myra Spiliopoulou","is_ca":false},{"name":"Bamshad Mobasher","is_ca":false},{"name":"Brij Masand","is_ca":false},{"name":"Philip S. Yu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03629811663647502,"gpt":0.2671368829012802,"spread":0.2308387662648052,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01473106,0.002775596,0.005192119,0.01057587,0.003311045,0.01200205,0.007702077,0.004525787,0.02967346],"category_scores_gemma":[0.04387472,0.002849277,0.004745157,0.007557062,0.0009537194,0.005477368,0.004991178,0.005654415,0.05764106],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003261767,"about_ca_system_score_gemma":0.01184397,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02172909,"about_ca_topic_score_gemma":0.02996112,"domain_scores_codex":[0.9877388,0.00268544,0.001719936,0.001834497,0.00546423,0.000557113],"domain_scores_gemma":[0.9831117,0.004926772,0.0007002512,0.005750616,0.004451186,0.001059479],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001439183,0.00009195577,0.00141141,0.001004928,0.0004224712,0.0001070821,0.00005353027,0.003692633,0.0001551154,0.005339997,0.9305846,0.05699222],"study_design_scores_gemma":[0.0003315652,0.00004225452,0.001046788,0.0002981366,0.0001811699,0.0003966957,0.00006077576,0.02198163,0.001606897,0.02428911,0.949665,0.00009994803],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.00413389,0.01712402,0.1136337,0.01546675,0.006916166,0.001372855,0.6918054,0.1171061,0.03244108],"genre_scores_gemma":[0.01168171,0.007800427,0.1037497,0.003622262,0.0005778375,0.001811538,0.8527166,0.003620089,0.01441974],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.02967346,"threshold_uncertainty_score":0.0992676,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4311079975","doi":"10.1145/3575637.3575654","title":"Experiences with Contrastive Predictive Coding in Industrial Time-Series Classification","year":2022,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"","keywords":"Computer science; Artificial intelligence; Classifier (UML); Machine learning; Coding (social sciences); Data mining; Multivariate statistics; Pattern recognition (psychology); Artificial neural network; Time series; Statistics","authors":[{"name":"Sunanda Gamage","is_ca":true},{"name":"Benjamin Klöpper","is_ca":false},{"name":"Jagath Samarabandu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04476407521292988,"gpt":0.2294792200046118,"spread":0.1847151447916819,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006951128,0.00070778,0.000602707,0.0007749548,0.0004686418,0.001017187,0.001554307,0.001053642,0.0006812192],"category_scores_gemma":[0.0169574,0.0002837024,0.0005170659,0.001216989,0.00155272,0.001948675,0.001009874,0.003094657,0.0002253583],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001405947,"about_ca_system_score_gemma":0.0006932846,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01145355,"about_ca_topic_score_gemma":0.006603139,"domain_scores_codex":[0.9976481,0.001268742,0.00009405665,0.0003042615,0.0005831781,0.0001016459],"domain_scores_gemma":[0.9872737,0.00934541,0.0002406172,0.001388169,0.001554959,0.000197171],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0005549998,0.0005216944,0.006881026,0.0001519247,0.0001007542,0.0001920497,0.001250069,0.3923897,0.01050216,0.01688013,0.005473001,0.5651025],"study_design_scores_gemma":[0.00001691978,0.0001142256,0.0008592178,0.00001341565,0.000008419917,0.00005394838,0.00006682951,0.9807867,0.0085315,0.007568834,0.001959352,0.00002062019],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.21576,0.002547436,0.769778,0.002935998,0.0002071992,0.0001497438,0.0001671373,0.00215993,0.006294702],"genre_scores_gemma":[0.7748652,0.0007044714,0.2220449,0.0002826446,0.0001299123,0.0000662657,0.0003191827,0.0001006671,0.001486786],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01145355,"threshold_uncertainty_score":0.03676152,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2048976003","doi":"10.1145/1540276.1540299","title":"WebKDD 2008","year":2008,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Knowledge extraction; Las vegas; Conjunction (astronomy); Data science; Library science; World Wide Web; Data mining; History; Archaeology; Tourism","authors":[{"name":"Olfa Nasraoui","is_ca":false},{"name":"Myra Spiliopoulou","is_ca":false},{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Jaideep Srivastava","is_ca":false},{"name":"Bamshad Mobasher","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05751956576975635,"gpt":0.2554444167508769,"spread":0.1979248509811206,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009461475,0.002605064,0.00433431,0.01091421,0.002458075,0.01193603,0.006865692,0.004240846,0.0514448],"category_scores_gemma":[0.03179611,0.002801312,0.004313322,0.007380276,0.0008714417,0.006915428,0.005391442,0.006646803,0.1180333],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00265016,"about_ca_system_score_gemma":0.008745913,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01206965,"about_ca_topic_score_gemma":0.01672895,"domain_scores_codex":[0.9903261,0.001672439,0.001059519,0.001378845,0.005096935,0.0004660473],"domain_scores_gemma":[0.9889035,0.002732135,0.0005253476,0.003478671,0.003530724,0.0008296949],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005105793,0.00003984022,0.0004196902,0.0003814396,0.0001114248,0.00003832971,0.00002461632,0.001100564,0.00008340739,0.003579611,0.9624169,0.0317531],"study_design_scores_gemma":[0.0001240936,0.00001550044,0.0004522918,0.000182699,0.00005613153,0.0002067209,0.00002404669,0.00771473,0.0007527834,0.01942636,0.9709877,0.00005700118],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.002065438,0.01455987,0.1098321,0.01667681,0.005772306,0.0008737277,0.6128939,0.1760378,0.06128797],"genre_scores_gemma":[0.006576434,0.008628371,0.0863389,0.00368892,0.0006053533,0.001292923,0.8587785,0.008541122,0.02554945],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.9485552,"threshold_uncertainty_score":0.1721,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2246590112","doi":"10.1145/2783702.2783708","title":"New Research Directions in Knowledge Discovery and Allied Spheres","year":2015,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Systems, Applications & Products in Data Processing (Canada)","funders":"Agence Nationale de la Recherche","keywords":"Computer science; Knowledge extraction; Data science; Schema (genetic algorithms); Viewpoints; Domain knowledge; Realm; Point (geometry); Information retrieval; World Wide Web; Knowledge management; Data mining","authors":[{"name":"Anisoara Nica","is_ca":true},{"name":"Fabian M. Suchanek","is_ca":false},{"name":"Aparna S. Varde","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1693548473193507,"gpt":0.3730839166959473,"spread":0.2037290693765966,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02499502,0.001212536,0.002878319,0.00869273,0.003673486,0.01785903,0.003853273,0.007708369,0.008291723],"category_scores_gemma":[0.02853077,0.001004806,0.002322207,0.012552,0.02812273,0.05967896,0.007604721,0.01203755,0.001952498],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006219712,"about_ca_system_score_gemma":0.006550976,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002929097,"about_ca_topic_score_gemma":0.00189987,"domain_scores_codex":[0.986639,0.006915032,0.0007385156,0.002096522,0.00295122,0.0006597089],"domain_scores_gemma":[0.8842483,0.09885149,0.001993168,0.006030641,0.005607362,0.003269078],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003523683,0.00006969159,0.0004108705,0.0007593469,0.00003244476,0.00008270062,0.0008407016,0.0004793203,0.00009636361,0.9386621,0.006765353,0.05176594],"study_design_scores_gemma":[0.00001579258,0.00002526778,0.0001934253,0.0004508302,0.00001438222,0.0001166538,0.0009895836,0.002491033,0.00009517717,0.9366944,0.0588868,0.00002663897],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"commentary","genre_scores_codex":[0.006816556,0.5727245,0.1481304,0.2058488,0.005019105,0.00008644663,0.0002321893,0.0002716352,0.06087048],"genre_scores_gemma":[0.2022746,0.5810171,0.1510269,0.02385276,0.02774064,0.000377851,0.0005170411,0.0001962268,0.01299686],"genre_candidate":"commentary","genre_consensus":null,"teacher_disagreement_score":0.02499502,"threshold_uncertainty_score":0.1321878,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4311079932","doi":"10.1145/3575637.3575645","title":"Finding Multidimensional Simpson's Paradox","year":2022,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University; Burnaby Hospital","funders":"","keywords":"Computer science; Curse of dimensionality; Multidimensional data; Data science; Phenomenon; Data mining; Machine learning; Epistemology","authors":[{"name":"Jay Xu","is_ca":true},{"name":"Jian Pei","is_ca":false},{"name":"Zicun Cong","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04358511030504669,"gpt":0.2740753601844798,"spread":0.2304902498794331,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007770575,0.0006511525,0.002227119,0.004647752,0.00303803,0.004691786,0.001790905,0.001941205,0.002969849],"category_scores_gemma":[0.07232878,0.0006749765,0.00142313,0.00434157,0.004138089,0.008352136,0.004317204,0.00368032,0.0002700079],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001755473,"about_ca_system_score_gemma":0.001618492,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001140653,"about_ca_topic_score_gemma":0.0009145209,"domain_scores_codex":[0.9941192,0.001885486,0.0005717825,0.00144619,0.001596989,0.0003803752],"domain_scores_gemma":[0.9500843,0.03647042,0.005053728,0.004186904,0.002745613,0.001459058],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0003250371,0.0001528306,0.05793236,0.0007496676,0.0004233066,0.001973165,0.002978988,0.01959277,0.002734529,0.7993779,0.01272624,0.1010333],"study_design_scores_gemma":[0.00004030241,0.00003330301,0.003905992,0.00007644729,0.00003841929,0.0008158768,0.0006440897,0.06881768,0.0007424405,0.9223539,0.002459259,0.00007230094],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.4857565,0.002958874,0.4887296,0.006838712,0.0005146546,0.00016499,0.001006474,0.0003984214,0.0136317],"genre_scores_gemma":[0.8972443,0.0008312762,0.09845445,0.0009243557,0.0002959352,0.000154304,0.0006311869,0.00007477406,0.001389375],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007770575,"threshold_uncertainty_score":0.0410952,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1978091735","doi":"10.1145/1809400.1809419","title":"Summary of the first ACM SIGKDD workshop on knowledge discovery from uncertain data (U'09)","year":2010,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Data science; Knowledge extraction; Information retrieval; Data mining","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Lise Getoor","is_ca":false},{"name":"Ander de Keijzer","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07323755533522192,"gpt":0.3036676604865192,"spread":0.2304301051512972,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005195947,0.001423151,0.00175675,0.00496995,0.001415042,0.005537197,0.001864083,0.001374484,0.09000779],"category_scores_gemma":[0.008758026,0.0006088722,0.001236055,0.005481289,0.0005320875,0.003903991,0.002149763,0.002130377,0.05761694],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001889409,"about_ca_system_score_gemma":0.003057474,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01137274,"about_ca_topic_score_gemma":0.02277276,"domain_scores_codex":[0.9980003,0.0004329083,0.000162336,0.0003133408,0.000924507,0.0001664382],"domain_scores_gemma":[0.988488,0.001923712,0.0002618359,0.0008106786,0.006671367,0.001844425],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001031638,0.00004323981,0.0001388119,0.0002887473,0.00001554451,0.00003212491,0.00002297294,0.0002372655,0.0002558925,0.0006057452,0.9040582,0.09419821],"study_design_scores_gemma":[0.00002961595,0.00007976595,0.0009775781,0.0003188518,0.00004165627,0.00008250884,0.00006426586,0.000820344,0.0008982691,0.002064766,0.9945912,0.00003119253],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"editorial","genre_gemma":"commentary","genre_scores_codex":[0.006606672,0.2406987,0.1192925,0.088022,0.3016812,0.001924323,0.05206734,0.008009085,0.1816982],"genre_scores_gemma":[0.02243642,0.1623983,0.04778581,0.01039471,0.05506768,0.0009781147,0.1077574,0.003716509,0.5894651],"genre_candidate":"commentary","genre_consensus":null,"teacher_disagreement_score":0.09000779,"threshold_uncertainty_score":0.301106,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1967519701","doi":"10.1145/772862.772885","title":"KDD-2002 workshop report fractals and self-similarity in data mining","year":2002,"lang":"en","type":"article","venue":"ACM SIGKDD Explorations Newsletter","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"Carnegie Mellon University; University of Southern California; National Science Foundation","keywords":"Computer science; Similarity (geometry); Data mining; Conjunction (astronomy); Fractal; Data science; Self-similarity; Information retrieval; Artificial intelligence; Mathematics","authors":[{"name":"Jafar Adibi","is_ca":false},{"name":"Christos Faloutsos","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1189229411204401,"gpt":0.2997637057302101,"spread":0.1808407646097699,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01274225,0.001220201,0.002306295,0.004295015,0.001161644,0.006795173,0.002514632,0.002629062,0.009911507],"category_scores_gemma":[0.02355784,0.0008080897,0.001276698,0.004272582,0.0008576661,0.004709425,0.003033153,0.003625528,0.009018932],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001703676,"about_ca_system_score_gemma":0.003893578,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003146335,"about_ca_topic_score_gemma":0.0023301,"domain_scores_codex":[0.9915894,0.002276985,0.000800215,0.001079015,0.003888059,0.0003663473],"domain_scores_gemma":[0.9846577,0.004657278,0.0003612047,0.00222306,0.006263889,0.001836906],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002728339,0.0002517638,0.001500057,0.0008629645,0.0001589015,0.0002494677,0.000204253,0.005771636,0.001364503,0.0208231,0.6899593,0.2785811],"study_design_scores_gemma":[0.0001941493,0.0002284515,0.002131399,0.0004762207,0.0001374366,0.0006072592,0.0002690688,0.04328719,0.00638449,0.05071599,0.8954658,0.0001025346],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.0168566,0.08913877,0.6975074,0.05587592,0.05035581,0.001303714,0.01466802,0.007713255,0.06658049],"genre_scores_gemma":[0.1212166,0.08215702,0.5237071,0.01017184,0.02583126,0.001990843,0.0760776,0.00411619,0.1547315],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.01274225,"threshold_uncertainty_score":0.0673883,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}