{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":1100,"total_is_capped":false,"direct_labels_cover":1,"predictions_cover":1100,"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":"6c72437377aa","filters":{"topic":"Data Mining Algorithms and Applications"}},"results":[{"id":"W2064853889","doi":"10.1145/335191.335372","title":"Mining frequent patterns without candidate generation","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":6360,"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; Scalability; Set (abstract data type); Tree (set theory); GSP Algorithm; Trie; Apriori algorithm; Association rule learning; Database transaction; Data structure; Database; Mathematics","authors":[{"name":"Jiawei Han","is_ca":true},{"name":"Jian Pei","is_ca":true},{"name":"Yiwen Yin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02961243143666664,"gpt":0.2688342363591712,"spread":0.2392218049225046,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002773499,0.001056047,0.001519753,0.003317313,0.00100269,0.00206516,0.002329979,0.001365068,0.001859842],"category_scores_gemma":[0.02035071,0.0007856847,0.001484656,0.00490263,0.0007125326,0.004425202,0.001633149,0.001036493,0.001685548],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004008443,"about_ca_system_score_gemma":0.001785138,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001120577,"about_ca_topic_score_gemma":0.001320278,"domain_scores_codex":[0.9961926,0.0007167507,0.000433287,0.0008190486,0.001568536,0.000269712],"domain_scores_gemma":[0.9865072,0.007654811,0.001021671,0.002320056,0.002259916,0.0002362861],"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":[0.0010564,0.000461875,0.01837986,0.0007472817,0.0003290033,0.001876907,0.000418306,0.06016991,0.01936469,0.02295568,0.009558297,0.8646818],"study_design_scores_gemma":[0.0001956015,0.0008190697,0.003836875,0.0001240209,0.0002428607,0.002747715,0.0003290195,0.9142375,0.02069824,0.0440002,0.01270348,0.00006536376],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09213933,0.001131604,0.8995709,0.0004390427,0.0001288635,0.000554885,0.001724529,0.001824547,0.002486367],"genre_scores_gemma":[0.3137254,0.0007244687,0.6758033,0.0002281819,0.0001549204,0.0006648764,0.006173484,0.0001304664,0.00239493],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003317313,"threshold_uncertainty_score":0.01466781,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4252403066","doi":"10.1145/342009.335372","title":"Mining frequent patterns without candidate generation","year":2000,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":3195,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science","authors":[{"name":"Jiawei Han","is_ca":true},{"name":"Jian Pei","is_ca":true},{"name":"Yiwen Yin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02286779170339445,"gpt":0.2596902950618048,"spread":0.2368225033584103,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004382792,0.001247691,0.002073848,0.004333536,0.001130014,0.002441965,0.002905045,0.002042173,0.003817771],"category_scores_gemma":[0.02670068,0.0007844547,0.001815085,0.004712982,0.0007989789,0.003809967,0.001955307,0.001178386,0.002802969],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003137145,"about_ca_system_score_gemma":0.001415975,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006662519,"about_ca_topic_score_gemma":0.0008739092,"domain_scores_codex":[0.9954359,0.001249113,0.0005080149,0.001050329,0.001442901,0.0003138157],"domain_scores_gemma":[0.9812487,0.01245813,0.000897677,0.003136664,0.001986318,0.000272529],"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":[0.001569741,0.000605119,0.01424198,0.0009347532,0.0004545501,0.002022357,0.0003380806,0.03815769,0.01411038,0.02363379,0.01036714,0.8935643],"study_design_scores_gemma":[0.0004492222,0.001388549,0.00489724,0.0001926765,0.0004465717,0.003913676,0.0002759552,0.8844184,0.02549854,0.06096217,0.01745515,0.0001017913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07246657,0.001026843,0.9177967,0.0004429516,0.0002151798,0.0007899947,0.001606556,0.002214215,0.003440989],"genre_scores_gemma":[0.3566686,0.0006150791,0.6301679,0.0003009335,0.000255101,0.0008406,0.007348476,0.0001721125,0.003631213],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004382792,"threshold_uncertainty_score":0.0231787,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2115482638","doi":"10.1023/b:dami.0000005258.31418.83","title":"Mining Frequent Patterns without Candidate Generation: A Frequent-Pattern Tree Approach","year":2003,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":2615,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Data mining; Computer science; Scalability; Tree (set theory); Set (abstract data type); Apriori algorithm; GSP Algorithm; Association rule learning; Trie; Tree structure; Pattern recognition (psychology); Data structure; Artificial intelligence; Database; Mathematics","authors":[{"name":"Jiawei Han","is_ca":false},{"name":"Jian Pei","is_ca":false},{"name":"Yiwen Yin","is_ca":true},{"name":"Runying Mao","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06223377143017436,"gpt":0.2906354910060366,"spread":0.2284017195758623,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005180415,0.001300011,0.003232713,0.004975306,0.001423573,0.003417117,0.00434752,0.002591436,0.002064071],"category_scores_gemma":[0.02203915,0.001061476,0.00262189,0.006072121,0.001022005,0.006168364,0.001920203,0.002250755,0.001881533],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004025765,"about_ca_system_score_gemma":0.001824103,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001007755,"about_ca_topic_score_gemma":0.001539642,"domain_scores_codex":[0.9948336,0.001288063,0.0005897361,0.001057398,0.001971786,0.0002594239],"domain_scores_gemma":[0.9828394,0.01182238,0.0009671408,0.001924336,0.002108966,0.0003377635],"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":[0.001031068,0.001224696,0.01331,0.001352846,0.0007791875,0.002112762,0.0004719143,0.03880866,0.01247221,0.02362365,0.008990197,0.8958228],"study_design_scores_gemma":[0.0002655924,0.0006092436,0.002345708,0.0001688598,0.0007396905,0.003678299,0.0003230342,0.8846376,0.008669051,0.0908212,0.007629762,0.000112],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01554049,0.0007677517,0.9811126,0.0003627418,0.0000735874,0.0002575588,0.0005567197,0.0007541402,0.0005744072],"genre_scores_gemma":[0.09764457,0.0006919973,0.8979678,0.0002048634,0.0001523225,0.000236876,0.001938123,0.0001178523,0.001045749],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005180415,"threshold_uncertainty_score":0.02739698,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2167681385","doi":"10.1109/icdm.2001.989541","title":"CMAR: accurate and efficient classification based on multiple class-association rules","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":1209,"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":"Association rule learning; Associative property; Overfitting; Computer science; Data mining; Class (philosophy); Artificial intelligence; Scalability; Flexibility (engineering); Classification rule; Set (abstract data type); Statistical classification; Decision tree; Tree (set theory); One-class classification; Pattern recognition (psychology); Machine learning; Database; Support vector machine; Mathematics; Artificial neural network; Statistics","authors":[{"name":"Wenmin Li","is_ca":true},{"name":"Jiawei Han","is_ca":true},{"name":"Jian Pei","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03426440388287782,"gpt":0.2436940187716003,"spread":0.2094296148887225,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003774841,0.001162092,0.001680635,0.006523139,0.0009755706,0.002559959,0.002956704,0.00162264,0.00281338],"category_scores_gemma":[0.01235206,0.0005526382,0.001184625,0.00735356,0.0008121165,0.004264732,0.001712257,0.001727023,0.003128091],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004965828,"about_ca_system_score_gemma":0.001300857,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002282555,"about_ca_topic_score_gemma":0.002657159,"domain_scores_codex":[0.9951385,0.000786809,0.0004902168,0.0007456743,0.002544312,0.0002945086],"domain_scores_gemma":[0.9896019,0.004686207,0.00123015,0.001676895,0.002579033,0.0002257967],"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":[0.0002516191,0.0002352833,0.01053027,0.0002527377,0.0001645935,0.0003092932,0.0001372359,0.01621187,0.005394375,0.006606535,0.0142154,0.9456908],"study_design_scores_gemma":[0.00008878131,0.0003514351,0.007936626,0.0001286543,0.0001890012,0.002214228,0.00026143,0.9138779,0.02040935,0.02870301,0.02572107,0.0001185879],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03586447,0.001421316,0.948099,0.0004656134,0.0002519233,0.000319595,0.001264153,0.008378414,0.003935527],"genre_scores_gemma":[0.1673002,0.0006069697,0.8265057,0.0002871864,0.0002294796,0.0003154493,0.001917733,0.0001744098,0.002662919],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006523139,"threshold_uncertainty_score":0.0199635,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2102297485","doi":"10.1145/1132960.1132963","title":"Interestingness measures for data mining","year":2006,"lang":"en","type":"review","venue":"ACM Computing Surveys","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":1125,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Regina","funders":"","keywords":"Computer science; Ranking (information retrieval); Data mining; Process (computing); Data science; Measure (data warehouse); Information retrieval","authors":[{"name":"Liqiang Geng","is_ca":true},{"name":"Howard J. Hamilton","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.295248331531593,"gpt":0.4196157088055408,"spread":0.1243673772739478,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01182948,0.002102161,0.00320094,0.01571627,0.001183244,0.005336717,0.003140284,0.002392347,0.004353996],"category_scores_gemma":[0.04178968,0.0007001277,0.002252566,0.01766214,0.003916271,0.0124438,0.002384552,0.004199994,0.002982786],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002354967,"about_ca_system_score_gemma":0.001849337,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007485713,"about_ca_topic_score_gemma":0.0004639476,"domain_scores_codex":[0.9837714,0.004522714,0.001965123,0.002236208,0.007219137,0.0002853557],"domain_scores_gemma":[0.9643875,0.02760815,0.00219629,0.002437723,0.003002548,0.0003678358],"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":[0.00009031595,0.0001143399,0.003197697,0.005828315,0.0003505194,0.0002774609,0.0004287259,0.009012241,0.001398263,0.4541155,0.03452443,0.4906621],"study_design_scores_gemma":[0.00004658729,0.0001428691,0.002716347,0.001786901,0.0001605591,0.001502742,0.0002782932,0.02324318,0.001417747,0.761216,0.20737,0.000118846],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.004497429,0.259448,0.6963258,0.005523076,0.001767422,0.0008034714,0.002440838,0.001188548,0.02800545],"genre_scores_gemma":[0.1069207,0.1901569,0.6730881,0.003736886,0.009052588,0.003551429,0.005359794,0.0005243137,0.007609311],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.01571627,"threshold_uncertainty_score":0.06256104,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2151953639","doi":"10.1109/tkde.2005.166","title":"Fast algorithms for frequent itemset mining using FP-trees","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Knowledge and Data Engineering","topic":"Data Mining Algorithms and Applications","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":"Concordia University","funders":"","keywords":"Computer science; Data mining; Traverse; Association rule learning; Algorithm; Tree (set theory); Data structure; Tree structure; Trie; Prefix; Binary tree; Mathematics","authors":[{"name":"Gösta Grahne","is_ca":true},{"name":"Jun Zhu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05517782314487753,"gpt":0.3067297556925058,"spread":0.2515519325476283,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006331121,0.002710182,0.003150566,0.007442361,0.002156128,0.002559494,0.003521581,0.002636136,0.004388689],"category_scores_gemma":[0.02853196,0.001748009,0.002424892,0.01038756,0.000772853,0.006881258,0.001551706,0.002630675,0.005217967],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001013653,"about_ca_system_score_gemma":0.002183775,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002423296,"about_ca_topic_score_gemma":0.002118937,"domain_scores_codex":[0.9952855,0.001128476,0.0007426538,0.0007659665,0.001780671,0.000296862],"domain_scores_gemma":[0.98015,0.0130163,0.001323591,0.001565065,0.003760589,0.0001844281],"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":[0.0004819868,0.0002638352,0.003242886,0.0008821056,0.0003762087,0.0003620489,0.0003072555,0.0569064,0.004409283,0.01765049,0.01712701,0.8979905],"study_design_scores_gemma":[0.000334072,0.0003799578,0.001654827,0.00027508,0.0002647925,0.001877789,0.0002277501,0.8398507,0.01417568,0.1123861,0.02840991,0.0001634568],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004194064,0.001256618,0.990008,0.0002194829,0.0001194287,0.000227195,0.0004799875,0.002760461,0.0007347819],"genre_scores_gemma":[0.02491221,0.0008922489,0.9714082,0.00008709006,0.0001118427,0.00055577,0.001323147,0.0001381519,0.0005714179],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007442361,"threshold_uncertainty_score":0.03348255,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W80315153","doi":"10.1137/1.9781611972740.51","title":"A Foundational Approach to Mining Itemset Utilities from Databases","year":2004,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","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":"University of Regina","funders":"","keywords":"Database transaction; Association rule learning; Computer science; Data mining; Database; Information retrieval","authors":[{"name":"Hong Yao","is_ca":false},{"name":"Howard J. Hamilton","is_ca":false},{"name":"Cory J. Butz","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0640315864945909,"gpt":0.2867239123360464,"spread":0.2226923258414555,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0061926,0.001200956,0.001638022,0.006577279,0.001670126,0.004600517,0.005217737,0.00173157,0.00276521],"category_scores_gemma":[0.04051817,0.001002374,0.002395754,0.008819396,0.002314292,0.00933635,0.003814054,0.00456713,0.001196132],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001042095,"about_ca_system_score_gemma":0.001796648,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001209101,"about_ca_topic_score_gemma":0.001236367,"domain_scores_codex":[0.9922919,0.002138781,0.00058464,0.001052062,0.003610287,0.0003223996],"domain_scores_gemma":[0.9822238,0.01206422,0.0007564473,0.00294079,0.001753274,0.0002614197],"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":[0.0001415139,0.0003398492,0.004182547,0.0006354713,0.0002426446,0.0003776102,0.0005095645,0.04776768,0.001352989,0.6369063,0.008712232,0.2988316],"study_design_scores_gemma":[0.0000254241,0.00006815109,0.0006636133,0.0001298795,0.00006075089,0.0004992475,0.00007156505,0.2322106,0.001344793,0.7579581,0.006934163,0.00003375082],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003860387,0.0004372615,0.9930905,0.0004330973,0.00003546718,0.0001294395,0.000303426,0.0002468966,0.001463543],"genre_scores_gemma":[0.1867706,0.00215617,0.8051761,0.0005577474,0.0006871358,0.0006952624,0.001765158,0.0001471064,0.002044701],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006577279,"threshold_uncertainty_score":0.03274995,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2507635762","doi":"10.1007/978-3-319-46131-1_8","title":"The SPMF Open-Source Data Mining Library Version 2","year":2016,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":551,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Moncton","funders":"","keywords":"Computer science; Implementation; Authorship attribution; Data mining; Plug-in; Open source; Visualization; Information retrieval; Interface (matter); Software; Software engineering; Artificial intelligence; Programming language; Operating system","authors":[{"name":"Philippe Fournier‐Viger","is_ca":false},{"name":"Jerry Chun‐Wei Lin","is_ca":false},{"name":"Antonio Gomariz","is_ca":false},{"name":"Ted Gueniche","is_ca":true},{"name":"Azadeh Soltani","is_ca":false},{"name":"Zhi‐Hong Deng","is_ca":false},{"name":"Hoang Thanh Lam","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03201302824163588,"gpt":0.2657663356650473,"spread":0.2337533074234114,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001940271,0.002244086,0.001624025,0.008104548,0.000984663,0.003663898,0.00314273,0.001021941,0.09998918],"category_scores_gemma":[0.01106983,0.001664031,0.002213541,0.007032293,0.0004069157,0.004346961,0.003115716,0.001905128,0.1602121],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006831475,"about_ca_system_score_gemma":0.002828294,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002711548,"about_ca_topic_score_gemma":0.003462576,"domain_scores_codex":[0.998166,0.0001286488,0.0002354661,0.0004107615,0.000917638,0.0001414117],"domain_scores_gemma":[0.9966611,0.001063037,0.0002236631,0.0009538312,0.0009402927,0.0001580994],"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":[0.000149189,0.00006270984,0.001431012,0.0009142794,0.00009253359,0.0000940828,0.00008878981,0.0008449109,0.002495168,0.002471667,0.7680002,0.2233555],"study_design_scores_gemma":[0.0001075437,0.00004833894,0.003290851,0.0003356924,0.00007862146,0.0007284876,0.0000574234,0.007047871,0.01054855,0.01708001,0.9605722,0.0001044529],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"software","genre_scores_codex":[0.002605223,0.00208896,0.2444228,0.0008653002,0.0005747835,0.0005880137,0.3725758,0.3479229,0.02835625],"genre_scores_gemma":[0.009871546,0.00196949,0.3045815,0.0008224488,0.0003022922,0.001375211,0.5844759,0.05820962,0.03839192],"genre_candidate":"software","genre_consensus":null,"teacher_disagreement_score":0.09998918,"threshold_uncertainty_score":0.3344971,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1569729202","doi":"10.1007/3-540-45571-x_47","title":"Mining Access Patterns Efficiently from Web Logs","year":2000,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":498,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Web mining; Data mining; Web mapping; Tree (set theory); Web modeling; World Wide Web; Information retrieval; Web service","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Jiawei Han","is_ca":true},{"name":"Behzad Mortazavi-Asl","is_ca":true},{"name":"Zhu Hua","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02328515983586727,"gpt":0.2684583184193423,"spread":0.245173158583475,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005351744,0.001053995,0.001590372,0.004864707,0.0006777162,0.002148615,0.001223549,0.0008037309,0.001478329],"category_scores_gemma":[0.004710914,0.0006749138,0.00110389,0.005454469,0.0003179325,0.003828865,0.0009420459,0.001060279,0.002080331],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003989917,"about_ca_system_score_gemma":0.001307729,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004068955,"about_ca_topic_score_gemma":0.01093539,"domain_scores_codex":[0.998943,0.00009156775,0.0001348597,0.0002143665,0.0005014354,0.0001147333],"domain_scores_gemma":[0.9969393,0.001596038,0.0003211331,0.000519199,0.000496306,0.0001280244],"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":[0.0004688528,0.0007371667,0.04447253,0.0005716355,0.0002296992,0.0007373516,0.0003405875,0.01785458,0.03391417,0.003742243,0.0189341,0.8779971],"study_design_scores_gemma":[0.0001133604,0.0002564374,0.0160463,0.000093706,0.0002887953,0.002309353,0.0007513207,0.8928275,0.02917481,0.04410286,0.01397659,0.00005894445],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.3542314,0.002303479,0.6025456,0.001109383,0.0001547531,0.0006563626,0.01221171,0.02241161,0.004375703],"genre_scores_gemma":[0.5739632,0.001372387,0.3953209,0.0001764135,0.0002441577,0.0003279865,0.02128657,0.0004650252,0.006843396],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004864707,"threshold_uncertainty_score":0.008090496,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1546703457","doi":"10.1137/1.9781611972733.6","title":"Hierarchical Document Clustering Using Frequent Itemsets","year":2003,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":469,"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":"Cluster analysis; Computer science; Document clustering; Information retrieval; Hierarchical clustering; Brown clustering; Intuition; Vocabulary; Artificial intelligence; Library science; Fuzzy clustering; Canopy clustering algorithm; Psychology","authors":[{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Ke Wang","is_ca":true},{"name":"Martin Ester","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03053123436791092,"gpt":0.2888526014938362,"spread":0.2583213671259252,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002736442,0.001462162,0.002356941,0.01299791,0.001591599,0.002668737,0.001951745,0.001275058,0.001738829],"category_scores_gemma":[0.01408443,0.0007416634,0.001795711,0.01063476,0.0005776742,0.002631754,0.001308696,0.001131503,0.002090389],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001029582,"about_ca_system_score_gemma":0.001428309,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004226096,"about_ca_topic_score_gemma":0.004023103,"domain_scores_codex":[0.9958973,0.001197961,0.000452586,0.0009429297,0.001291705,0.0002175849],"domain_scores_gemma":[0.9937807,0.002857612,0.0007371727,0.0008447085,0.001635085,0.0001446466],"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":[0.0003007356,0.0002485574,0.00747498,0.000778018,0.0005266748,0.000515512,0.0005063927,0.1441382,0.00565214,0.01683213,0.01441965,0.8086071],"study_design_scores_gemma":[0.00005526491,0.0001098917,0.002207426,0.0001126937,0.0001512354,0.0006399642,0.000269226,0.9228911,0.004730357,0.05813573,0.01061628,0.00008077128],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01236632,0.001065179,0.9816865,0.0001966724,0.00008757701,0.0003779845,0.001035102,0.001976786,0.001207935],"genre_scores_gemma":[0.1103045,0.0007689231,0.8834099,0.0000669574,0.0001263706,0.0004533988,0.003752519,0.000123304,0.0009941801],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01299791,"threshold_uncertainty_score":0.01447189,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1906561224","doi":"10.1109/pdis.1996.568665","title":"A fast distributed algorithm for mining association rules","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":439,"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":"Association rule learning; Computer science; Scalability; Database transaction; Data mining; Distributed database; Distributed algorithm; Association (psychology); Algorithm design; Algorithm; Distributed computing; Database","authors":[{"name":"David W. Cheung","is_ca":false},{"name":"Jiawei Han","is_ca":true},{"name":"Vincent Ng","is_ca":false},{"name":"A.W.-C. Fu","is_ca":false},{"name":"Yongjian Fu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02193598178644724,"gpt":0.2439799088221452,"spread":0.222043927035698,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004592292,0.001559391,0.002124554,0.004026335,0.001767337,0.001948495,0.00304531,0.001823224,0.003491309],"category_scores_gemma":[0.01417434,0.0008990695,0.001462838,0.005680232,0.0009520062,0.002976439,0.002289332,0.002121191,0.002496214],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006767472,"about_ca_system_score_gemma":0.002767029,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002257898,"about_ca_topic_score_gemma":0.002890581,"domain_scores_codex":[0.9962805,0.0008770883,0.000385674,0.0009908822,0.001296471,0.0001695043],"domain_scores_gemma":[0.9949414,0.002803612,0.0002995447,0.0008568087,0.0009655636,0.0001331264],"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":[0.0003441313,0.0001974819,0.001540146,0.0003179038,0.0002302127,0.0002295338,0.0001557027,0.05233393,0.004912802,0.01544628,0.01122237,0.9130694],"study_design_scores_gemma":[0.0005706827,0.0002913939,0.0007950916,0.00007460299,0.0001687651,0.001517933,0.0001175807,0.8765871,0.006412079,0.08524678,0.02814674,0.00007129946],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.001550103,0.0003075896,0.9964199,0.0001360091,0.00007356898,0.0001276368,0.00009399965,0.0009207347,0.0003704253],"genre_scores_gemma":[0.02123502,0.0002258497,0.9764371,0.0001117193,0.00007374202,0.0004306303,0.000412576,0.00005978741,0.001013734],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004592292,"threshold_uncertainty_score":0.02428663,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2126046032","doi":"","title":"SPMF: a Java open-source pattern mining library","year":2014,"lang":"en","type":"article","venue":"Journal of Machine Learning Research","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":417,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Moncton","funders":"","keywords":"Computer science; Java; Source code; License; Interface (matter); MIT License; Open source; Documentation; Database transaction; Implementation; Data mining; Database; Programming language; Information retrieval; World Wide Web; Operating system; Software","authors":[{"name":"Philippe Fournier‐Viger","is_ca":true},{"name":"Antonio Gomariz","is_ca":false},{"name":"Ted Gueniche","is_ca":false},{"name":"Azadeh Soltani","is_ca":false},{"name":"Chengwei Wu","is_ca":false},{"name":"Vincent S. Tseng","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05115801282898343,"gpt":0.3570069714406174,"spread":0.305848958611634,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001503976,0.002422423,0.001575958,0.00640036,0.0009745544,0.002754886,0.004373354,0.001179383,0.02992554],"category_scores_gemma":[0.008411132,0.001722056,0.002727928,0.005530511,0.0004051624,0.004108985,0.002154084,0.001644868,0.03174977],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006285683,"about_ca_system_score_gemma":0.002723038,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003037419,"about_ca_topic_score_gemma":0.004064546,"domain_scores_codex":[0.9980674,0.0001474851,0.0003440146,0.0005070489,0.0007903407,0.0001437532],"domain_scores_gemma":[0.9972995,0.001116809,0.0002865419,0.000420327,0.0007168239,0.0001600886],"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":[0.000375494,0.000244651,0.005469434,0.002468527,0.0003552765,0.0005632711,0.0002402311,0.003866631,0.00868571,0.005863612,0.3871024,0.5847647],"study_design_scores_gemma":[0.0003385453,0.0001891149,0.008706733,0.0005702608,0.000269141,0.003096116,0.0001794463,0.07574019,0.03102589,0.03267725,0.8469191,0.0002882296],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"software","genre_scores_codex":[0.00291021,0.0008301216,0.649344,0.0003484307,0.0002033044,0.0004698611,0.0550875,0.2852024,0.005604147],"genre_scores_gemma":[0.01163088,0.0009362017,0.8173952,0.0004749344,0.0001185974,0.00131935,0.1384892,0.01946582,0.0101698],"genre_candidate":"software","genre_consensus":null,"teacher_disagreement_score":0.02992554,"threshold_uncertainty_score":0.1001109,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1498871448","doi":"10.1109/icdm.2001.989550","title":"H-mine: hyper-structure mining of frequent patterns in large databases","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":408,"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; Scalability; Overhead (engineering); Process (computing); Set (abstract data type); Database; GSP Algorithm; Association rule learning; Apriori algorithm","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Jiawei Han","is_ca":true},{"name":"Hongjun Lü","is_ca":false},{"name":"Shojiro Nishio","is_ca":false},{"name":"Shiwei Tang","is_ca":false},{"name":"D. Yang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03236253180025632,"gpt":0.2603427974947213,"spread":0.227980265694465,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002300002,0.0007583228,0.001072077,0.003917022,0.0009124177,0.001337354,0.001907329,0.0009996117,0.001490064],"category_scores_gemma":[0.00993496,0.0006545185,0.0009156875,0.003922861,0.0007306199,0.003275242,0.001384123,0.00090609,0.00101484],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003596254,"about_ca_system_score_gemma":0.001393257,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008472004,"about_ca_topic_score_gemma":0.001281781,"domain_scores_codex":[0.9986552,0.0003572844,0.000155704,0.0002630511,0.0004951757,0.00007365629],"domain_scores_gemma":[0.9944004,0.003337364,0.0006854302,0.0008262018,0.0005716637,0.0001789639],"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":[0.0006749117,0.0004596098,0.01845374,0.0008847728,0.0004228011,0.000641374,0.000637886,0.04912608,0.01303533,0.01875453,0.0167599,0.8801489],"study_design_scores_gemma":[0.0003250731,0.0004867192,0.005945705,0.0001317067,0.0001493176,0.001966256,0.0003853873,0.8569765,0.02555637,0.08838577,0.01959812,0.0000930728],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0392856,0.0009436585,0.9517953,0.0004984605,0.00008460147,0.0003459182,0.001701741,0.004464449,0.0008802628],"genre_scores_gemma":[0.1105738,0.0004629631,0.8845419,0.0001762013,0.0001066678,0.0003357707,0.002796361,0.0001537377,0.0008526726],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003917022,"threshold_uncertainty_score":0.01216376,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2146606092","doi":"10.1145/1824795.1824798","title":"A taxonomy of sequential pattern mining algorithms","year":2010,"lang":"en","type":"review","venue":"ACM Computing Surveys","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":396,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Windsor","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Taxonomy (biology); Tree traversal; Sequential Pattern Mining; Key (lock); Data mining; Web mining; Information retrieval; Artificial intelligence; Machine learning; Algorithm; Web page; World Wide Web","authors":[{"name":"Nizar R. Mabroukeh","is_ca":true},{"name":"C. I. Ezeife","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1231731861239806,"gpt":0.3469654630301945,"spread":0.2237922769062139,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005904639,0.002009987,0.002254194,0.01127989,0.001123441,0.004337171,0.004012777,0.002146513,0.003380337],"category_scores_gemma":[0.01569924,0.0009610943,0.00175648,0.01993494,0.001280025,0.008406404,0.001347549,0.002531636,0.004808785],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001550476,"about_ca_system_score_gemma":0.00377114,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002244065,"about_ca_topic_score_gemma":0.001684907,"domain_scores_codex":[0.9944044,0.001118697,0.0009690289,0.001020817,0.002322363,0.0001645939],"domain_scores_gemma":[0.9894288,0.005581796,0.0005956767,0.0007946815,0.003428592,0.0001704738],"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":[0.00006199003,0.000150519,0.002503383,0.003979444,0.0001368254,0.0001965273,0.0002445334,0.007004847,0.001119951,0.07572141,0.0167041,0.8921766],"study_design_scores_gemma":[0.00008593519,0.0004653084,0.003154749,0.004073422,0.0002493643,0.004602333,0.0004663457,0.1040231,0.004623978,0.3354673,0.5425833,0.0002047006],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.003680165,0.1911608,0.7796493,0.004159501,0.0009110859,0.000892119,0.001177805,0.001440661,0.01692847],"genre_scores_gemma":[0.02225502,0.1663861,0.8009118,0.001076167,0.001027957,0.0009242094,0.002398322,0.0001531492,0.004867184],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.01127989,"threshold_uncertainty_score":0.03122711,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1599136575","doi":"10.1109/adl.1998.670376","title":"Discovering Web access patterns and trends by applying OLAP and data mining technology on Web logs","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":396,"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; Web mining; World Wide Web; Clickstream; Data Web; The Internet; Web page; Web development; Web intelligence; Online analytical processing; Web API; Database; Data warehouse","authors":[{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Man Xin","is_ca":true},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05271091983902785,"gpt":0.3046453902207973,"spread":0.2519344703817695,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002464975,0.001109592,0.00104184,0.01541914,0.000825274,0.003222097,0.001133476,0.0006806756,0.0008710705],"category_scores_gemma":[0.01097112,0.0005199921,0.0009882668,0.009561462,0.000422372,0.00377558,0.00105477,0.00117534,0.0009255581],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007253101,"about_ca_system_score_gemma":0.00145253,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005884929,"about_ca_topic_score_gemma":0.009898797,"domain_scores_codex":[0.9978446,0.0003677043,0.0004040168,0.0003773504,0.0008975642,0.0001088405],"domain_scores_gemma":[0.9909277,0.004726763,0.001405777,0.001145075,0.001509894,0.0002849116],"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":[0.0003403196,0.001018867,0.158522,0.001159101,0.000331889,0.0008751893,0.00148214,0.02577516,0.01224351,0.004625603,0.006914358,0.7867119],"study_design_scores_gemma":[0.00009083089,0.0006612669,0.1068372,0.0005479518,0.0004526331,0.002283764,0.00399781,0.7519039,0.04162776,0.05133466,0.03996887,0.0002933447],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3039468,0.00185392,0.6458239,0.001875415,0.0001366389,0.001438113,0.02285529,0.01584917,0.006220762],"genre_scores_gemma":[0.3910933,0.001586887,0.5842049,0.0001982356,0.0001152683,0.000819077,0.02022056,0.0002986738,0.001463067],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01541914,"threshold_uncertainty_score":0.01303619,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2028470267","doi":"10.1017/s0269888906000737","title":"A survey of Knowledge Discovery and Data Mining process models","year":2006,"lang":"en","type":"article","venue":"The Knowledge Engineering Review","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":388,"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; Process (computing); Business process discovery; Interoperability; Knowledge extraction; Process mining; Process modeling; Automation; Knowledge management; Data mining; Work in process; Engineering; Business process; Business process management; World Wide Web; Business process modeling","authors":[{"name":"Lukasz Kurgan","is_ca":true},{"name":"Petr Musı́lek","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06022485778747027,"gpt":0.3127235079404043,"spread":0.252498650152934,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005959783,0.001680907,0.00220777,0.007423976,0.0009979659,0.006085434,0.003872105,0.002549596,0.00338526],"category_scores_gemma":[0.01196047,0.0008168241,0.001744613,0.01593905,0.00144442,0.008557822,0.001622123,0.002264064,0.002902926],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002884157,"about_ca_system_score_gemma":0.004416508,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003623604,"about_ca_topic_score_gemma":0.002399644,"domain_scores_codex":[0.9946496,0.001217069,0.0006933624,0.0008353121,0.002415668,0.0001889498],"domain_scores_gemma":[0.9929304,0.00496399,0.0003909752,0.0005340478,0.001071167,0.0001094415],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"systematic_review","study_design_scores_codex":[0.0001088359,0.0001990566,0.002887725,0.005439402,0.0001850524,0.0003781015,0.0004350243,0.02548851,0.0007138635,0.3943774,0.02513954,0.5446475],"study_design_scores_gemma":[0.00003369687,0.0001371097,0.001314571,0.002487131,0.0002207918,0.001461523,0.0003806889,0.1024414,0.001653251,0.3979549,0.491796,0.0001189846],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.005718871,0.2684447,0.6773576,0.01048176,0.0008264425,0.000398426,0.001876486,0.001059862,0.03383582],"genre_scores_gemma":[0.06520494,0.4908822,0.4250058,0.002500199,0.001591125,0.0009789361,0.003995808,0.0002517941,0.009589125],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.007423976,"threshold_uncertainty_score":0.0315187,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2140617045","doi":"","title":"Application of data mining techniques for medical image classification","year":2001,"lang":"en","type":"article","venue":"ACM Multimedia","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":311,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Association rule learning; Computer science; Categorization; Mammography; Data mining; Breast cancer; Artificial intelligence; Artificial neural network; Anomaly detection; Contextual image classification; Pattern recognition (psychology); Image (mathematics); Machine learning; Cancer; Medicine","authors":[{"name":"Maria-Luiza Antonie","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Alexandru Coman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06693477581833332,"gpt":0.3635870522762567,"spread":0.2966522764579234,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00246669,0.0006650204,0.000994207,0.005502011,0.0006377185,0.002182742,0.001287781,0.001127631,0.001796026],"category_scores_gemma":[0.008841845,0.0003450267,0.001142432,0.00454287,0.0005483444,0.001156868,0.0008992943,0.001402356,0.001075799],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004961986,"about_ca_system_score_gemma":0.0008512942,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001293311,"about_ca_topic_score_gemma":0.001454806,"domain_scores_codex":[0.9983589,0.0002889301,0.000256691,0.000199502,0.0008323081,0.00006372383],"domain_scores_gemma":[0.9952055,0.002642074,0.0003887445,0.0004581421,0.001193502,0.0001119947],"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":[0.0002914564,0.0003323599,0.01125978,0.0007634738,0.0002924147,0.0004234319,0.0001856722,0.009678897,0.01670489,0.004005877,0.006832369,0.9492294],"study_design_scores_gemma":[0.0001671997,0.0006328303,0.01993801,0.0005774511,0.0008605029,0.005875756,0.0006005303,0.7408517,0.1042798,0.05108702,0.07499322,0.0001359676],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04780476,0.009684786,0.9288068,0.003136897,0.0005855324,0.0005056557,0.001537624,0.002584477,0.005353482],"genre_scores_gemma":[0.3086364,0.005845502,0.6792709,0.0005727809,0.0004220039,0.0003314746,0.00199592,0.0001244586,0.002800589],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005502011,"threshold_uncertainty_score":0.01304525,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2105564121","doi":"","title":"Privacy preserving frequent itemset mining","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":286,"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 mining; Database transaction; Set (abstract data type); Process (computing); Database","authors":[{"name":"Stanley Robson de Medeiros Oliveira","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04667822289339513,"gpt":0.2506300199518214,"spread":0.2039517970584263,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01257219,0.0007452705,0.002247817,0.002669935,0.001601629,0.004362104,0.002976813,0.00181461,0.001026399],"category_scores_gemma":[0.06270926,0.0006689395,0.001539628,0.004486921,0.001729834,0.00691947,0.002832348,0.002080668,0.001128792],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007526493,"about_ca_system_score_gemma":0.002284506,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006114056,"about_ca_topic_score_gemma":0.0005940087,"domain_scores_codex":[0.9834824,0.005934305,0.002271809,0.002366958,0.005097992,0.0008466653],"domain_scores_gemma":[0.915818,0.03293405,0.006341335,0.03793606,0.006328288,0.0006422987],"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":[0.001792892,0.000634059,0.03126199,0.0006884516,0.0006358859,0.0009761049,0.001048435,0.120192,0.04093894,0.06741104,0.009743424,0.7246768],"study_design_scores_gemma":[0.0001294906,0.0005516213,0.003791015,0.0001103037,0.0002579427,0.002661139,0.000515132,0.6811715,0.08099169,0.2143537,0.01534705,0.0001193851],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05636937,0.000745524,0.9370478,0.001017825,0.00007558829,0.0001914208,0.0008899931,0.001559992,0.002102499],"genre_scores_gemma":[0.5411311,0.0008001415,0.4528724,0.0004135014,0.0002409553,0.0002273855,0.002336557,0.0001590362,0.001818986],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01257219,"threshold_uncertainty_score":0.06648886,"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":"W2112971308","doi":"10.1007/978-3-319-06608-0_4","title":"Fast Vertical Mining of Sequential Patterns Using Co-occurrence Information","year":2014,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":242,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Moncton","funders":"European Regional Development Fund; Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Pruning; Bottleneck; Data mining; Representation (politics); Sequential Pattern Mining; Information bottleneck method; State (computer science); Pattern recognition (psychology); Artificial intelligence; Algorithm; Cluster analysis","authors":[{"name":"Philippe Fournier‐Viger","is_ca":true},{"name":"Antonio Gomariz","is_ca":false},{"name":"Manuel Campos","is_ca":false},{"name":"Rincy Thomas","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02807506608825088,"gpt":0.2794808825966699,"spread":0.251405816508419,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005802197,0.001039975,0.001231359,0.00585938,0.0008274781,0.001511339,0.001221115,0.0006289563,0.002607624],"category_scores_gemma":[0.003180935,0.0005523649,0.001201645,0.007420786,0.0002622421,0.001790701,0.001058442,0.0008338204,0.002060868],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003315832,"about_ca_system_score_gemma":0.001248969,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0041108,"about_ca_topic_score_gemma":0.007662056,"domain_scores_codex":[0.9989889,0.00007849128,0.0001280769,0.0002274318,0.0004574138,0.000119721],"domain_scores_gemma":[0.9974541,0.001143644,0.0002956646,0.0003091532,0.0006483599,0.0001489875],"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":[0.0005458505,0.0002932385,0.01468603,0.0006352427,0.0002438781,0.0005987593,0.0002001172,0.008186978,0.02760908,0.003859074,0.01201194,0.9311298],"study_design_scores_gemma":[0.0001481118,0.0005987884,0.01709593,0.0002398156,0.000460945,0.002996227,0.0005386145,0.8784603,0.03786397,0.03629705,0.02520213,0.00009808243],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1411406,0.003559109,0.8349794,0.0003431131,0.0002556377,0.0003657185,0.008043376,0.006679707,0.004633316],"genre_scores_gemma":[0.3204465,0.001284734,0.654882,0.00008269391,0.0001632905,0.0002152056,0.01773678,0.000292523,0.004896363],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00585938,"threshold_uncertainty_score":0.008723319,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2513061279","doi":"10.1007/s10115-016-0986-0","title":"EFIM: a fast and memory efficient algorithm for high-utility itemset mining","year":2016,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":239,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Moncton","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Computer science; Key (lock); Data mining; Projection (relational algebra); Database transaction; Task (project management); Tree (set theory); Algorithm; High memory; Database; Mathematics; Parallel computing","authors":[{"name":"Souleymane Zida","is_ca":true},{"name":"Philippe Fournier‐Viger","is_ca":false},{"name":"Jerry Chun‐Wei Lin","is_ca":false},{"name":"Chengwei Wu","is_ca":false},{"name":"Vincent S. Tseng","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01286533636610831,"gpt":0.2381222319819108,"spread":0.2252568956158025,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002128504,0.001848496,0.002000174,0.005484755,0.001302347,0.001932655,0.004423071,0.001950544,0.008165998],"category_scores_gemma":[0.01012878,0.0009413406,0.001424512,0.007221487,0.0004804058,0.003943192,0.002838069,0.001691584,0.004507513],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007999422,"about_ca_system_score_gemma":0.002295863,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003248723,"about_ca_topic_score_gemma":0.005698227,"domain_scores_codex":[0.9985874,0.0002644389,0.0002229209,0.0002816058,0.0004789818,0.0001646483],"domain_scores_gemma":[0.9972329,0.001373022,0.0001545134,0.0005448204,0.0005814935,0.0001133296],"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":[0.0008522374,0.0003128728,0.002929007,0.0004685944,0.0001886996,0.0002351143,0.0001192448,0.01661964,0.003935123,0.004598603,0.03019193,0.939549],"study_design_scores_gemma":[0.0004705612,0.0004178163,0.002324692,0.0002041119,0.0001981563,0.001326218,0.0002057489,0.893647,0.01904894,0.04372665,0.03832556,0.0001045901],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01899075,0.001886892,0.9525266,0.0004182133,0.0002427152,0.0004443645,0.003413398,0.01990284,0.002174239],"genre_scores_gemma":[0.04341999,0.0003351131,0.948034,0.0001868907,0.00007432727,0.000337082,0.004750748,0.0004298192,0.002431891],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008165998,"threshold_uncertainty_score":0.02731794,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2052496634","doi":"10.1007/s10844-006-0006-z","title":"Constraint-based sequential pattern mining: the pattern-growth methods","year":2007,"lang":"en","type":"article","venue":"Journal of Intelligent Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":233,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Simon Fraser University; National Science Foundation","keywords":"Sequential Pattern Mining; Computer science; Constraint (computer-aided design); Data mining; Point (geometry); Artificial intelligence; Mathematics","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Jiawei Han","is_ca":false},{"name":"Wei Wang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04073344562841342,"gpt":0.3292330749489349,"spread":0.2884996293205215,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004593479,0.001543776,0.002678273,0.004161199,0.0009722202,0.002356693,0.004278783,0.001356875,0.003021284],"category_scores_gemma":[0.02172062,0.0008862916,0.002053272,0.008666292,0.001209194,0.004052677,0.001782616,0.002951365,0.001336409],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008071624,"about_ca_system_score_gemma":0.003552192,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006587581,"about_ca_topic_score_gemma":0.008069252,"domain_scores_codex":[0.9963036,0.0009912582,0.0003343472,0.0007489662,0.001469998,0.0001518419],"domain_scores_gemma":[0.9845999,0.009689311,0.001042564,0.001765534,0.002563521,0.0003391607],"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":[0.0004930506,0.0004632476,0.008692825,0.001365225,0.000530222,0.0004030888,0.00020497,0.1048208,0.004852985,0.03952989,0.01292851,0.8257152],"study_design_scores_gemma":[0.00006378006,0.0001117592,0.001041217,0.0000766235,0.0001115441,0.000368294,0.00005718877,0.9401612,0.002650973,0.04912446,0.00619279,0.0000401099],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00747499,0.001794766,0.987452,0.0006003351,0.00009173618,0.0002421883,0.0006912326,0.0006777401,0.0009750847],"genre_scores_gemma":[0.09248449,0.002475891,0.9001809,0.0002999555,0.0002456716,0.0005136831,0.001640873,0.0002431965,0.001915409],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006587581,"threshold_uncertainty_score":0.02429295,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2744510879","doi":"10.1016/j.ins.2017.08.031","title":"Information sciences 1968–2016: A retrospective analysis with text mining and bibliometric","year":2017,"lang":"en","type":"article","venue":"Information Sciences","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":227,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"Natural Science Foundation of Zhejiang Province; National Natural Science Foundation of China","keywords":"Citation; Data science; Computer science; Construct (python library); Salient; Citation analysis; Bibliometrics; Web of science; Information retrieval; Key (lock); Library science; Political science; MEDLINE","authors":[{"name":"Dejian Yu","is_ca":false},{"name":"Zeshui Xu","is_ca":false},{"name":"Witold Pedrycz","is_ca":true},{"name":"Wanru Wang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02740014933177197,"gpt":0.3024425598335855,"spread":0.2750424105018136,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.003320504,0.0005663913,0.0006806977,0.02974497,0.0008845802,0.002808987,0.0007704235,0.0006930016,0.003149277],"category_scores_gemma":[0.02692447,0.0003409297,0.0008009623,0.05115952,0.0008044082,0.002734912,0.001893073,0.0009323588,0.003628584],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002049643,"about_ca_system_score_gemma":0.004162979,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03083964,"about_ca_topic_score_gemma":0.03813508,"domain_scores_codex":[0.9966582,0.000274707,0.0007139881,0.0004145872,0.001622919,0.000315608],"domain_scores_gemma":[0.9684734,0.006230276,0.01012631,0.003182163,0.01066099,0.001326861],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001179357,0.0002111312,0.7333357,0.002234352,0.0006445932,0.0009057139,0.002034845,0.001316035,0.001709437,0.005827375,0.06403238,0.1865691],"study_design_scores_gemma":[0.00001441713,0.0001008877,0.7770672,0.0007768211,0.0003539589,0.0007797325,0.0007649763,0.0003675613,0.002536909,0.0009261089,0.2162664,0.00004505106],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6079813,0.03806808,0.004145121,0.004146655,0.0009783671,0.0002267422,0.2919109,0.000447113,0.05209574],"genre_scores_gemma":[0.7794808,0.02348904,0.00430505,0.0007615463,0.001440855,0.0002323668,0.1694072,0.0003023369,0.02058087],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.970255,"threshold_uncertainty_score":0.0613203,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1794915560","doi":"10.1007/978-3-540-68125-0_61","title":"A Tree-Based Approach for Frequent Pattern Mining from Uncertain Data","year":2008,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":224,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Computer science; Database transaction; Data mining; Tree (set theory); Transaction data; Database; Mathematics","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Mark Anthony F. Mateo","is_ca":true},{"name":"Dale A. Brajczuk","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06928100922740438,"gpt":0.280066816554506,"spread":0.2107858073271017,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002232265,0.0008951187,0.001780954,0.002977605,0.0009786004,0.001915566,0.002350358,0.00167942,0.002998027],"category_scores_gemma":[0.01127941,0.000628568,0.001917247,0.006337795,0.0004891509,0.003409761,0.001430065,0.001766289,0.001939154],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00043116,"about_ca_system_score_gemma":0.001263593,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002819096,"about_ca_topic_score_gemma":0.003836194,"domain_scores_codex":[0.9973277,0.0005595571,0.0003378844,0.000386112,0.001285236,0.0001035715],"domain_scores_gemma":[0.9941757,0.003731587,0.0002119632,0.0004803358,0.001256821,0.0001436016],"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":[0.000326477,0.0002774832,0.001485008,0.0007249972,0.000359861,0.0006608374,0.0003449412,0.059506,0.009388476,0.02726307,0.01488152,0.8847813],"study_design_scores_gemma":[0.00006684085,0.0001912446,0.000636648,0.0001176105,0.0001937641,0.001082635,0.0001174632,0.9045362,0.003698254,0.07488728,0.01440624,0.00006585241],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00223084,0.0005532297,0.995732,0.0001303148,0.00006299346,0.000088666,0.0003143845,0.0005032094,0.0003843311],"genre_scores_gemma":[0.02726466,0.0007710917,0.969488,0.0001260727,0.0001107321,0.0001790468,0.001080526,0.00007444665,0.0009055603],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.002998027,"threshold_uncertainty_score":0.01180553,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2117423052","doi":"10.1109/icdm.2002.1183881","title":"Text document categorization by term association","year":2003,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":224,"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; Classifier (UML); Categorization; Artificial intelligence; Text categorization; Association rule learning; Machine learning; Natural language processing; Data mining; Information retrieval; Pattern recognition (psychology)","authors":[{"name":"Maria-Luiza Antonie","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.005463032048834978,"gpt":0.2254333012677956,"spread":0.2199702692189607,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001965918,0.001288859,0.002104377,0.01887806,0.001359844,0.003091806,0.001630575,0.001345324,0.003847195],"category_scores_gemma":[0.008328383,0.0003848431,0.001952354,0.01496377,0.0009172916,0.003989653,0.001470869,0.001343699,0.00651803],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008152172,"about_ca_system_score_gemma":0.001645116,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002096726,"about_ca_topic_score_gemma":0.002086104,"domain_scores_codex":[0.9969261,0.0004566119,0.000495104,0.0008055436,0.00115198,0.0001646471],"domain_scores_gemma":[0.996585,0.001324837,0.0004781901,0.0005127912,0.000969644,0.0001295363],"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":[0.0001850968,0.0002342459,0.002929035,0.0008393743,0.0001679322,0.0001979212,0.0002277342,0.003810916,0.01687346,0.008394107,0.01969717,0.946443],"study_design_scores_gemma":[0.0002499652,0.000923178,0.01644896,0.0007967952,0.000677807,0.005677599,0.001193084,0.4947587,0.07439812,0.1688602,0.2355898,0.0004257906],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02531536,0.004801477,0.947479,0.0006406003,0.0005637011,0.002040956,0.005908788,0.007241863,0.006008335],"genre_scores_gemma":[0.04351626,0.001363272,0.9436159,0.0001143968,0.0002554178,0.0009065357,0.006849216,0.0002173185,0.003161681],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01887806,"threshold_uncertainty_score":0.01287013,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2169474320","doi":"10.1109/ideas.2003.1214917","title":"Incremental mining of frequent patterns without candidate generation or support constraint","year":2003,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":216,"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; Constraint (computer-aided design); Data mining; Tree (set theory); Tree structure; Binary tree; Algorithm; Engineering; Mathematics","authors":[{"name":"Wendy Cheung","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0403410905032692,"gpt":0.284965270308043,"spread":0.2446241798047738,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001432076,0.0004723834,0.001019957,0.002246292,0.0005970101,0.001182154,0.002126653,0.0006924628,0.001255584],"category_scores_gemma":[0.01186994,0.0004740657,0.0008609847,0.002672651,0.0003781389,0.003276031,0.001127635,0.0007380983,0.0006882698],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002678097,"about_ca_system_score_gemma":0.0012262,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001081538,"about_ca_topic_score_gemma":0.001963685,"domain_scores_codex":[0.9985321,0.0002527769,0.0001639465,0.0002348237,0.0007153237,0.0001011474],"domain_scores_gemma":[0.9909397,0.005519658,0.0005312313,0.001122007,0.001689669,0.0001976885],"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":[0.0008434816,0.0004110382,0.0147226,0.0006692452,0.0002667585,0.001667637,0.0003613333,0.02927555,0.02599532,0.01215893,0.00674951,0.9068785],"study_design_scores_gemma":[0.0002490384,0.0009551772,0.006432623,0.0001454426,0.0004595139,0.003579644,0.000386375,0.8815486,0.04092842,0.04427118,0.02094423,0.00009987576],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1101384,0.0008930831,0.8838767,0.0003322701,0.0001057463,0.0003339888,0.0009521928,0.001546583,0.001821116],"genre_scores_gemma":[0.2848112,0.0004471237,0.7094833,0.0001644032,0.0001124696,0.0003062897,0.003328128,0.00009485344,0.001252346],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002246292,"threshold_uncertainty_score":0.007573605,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1581468894","doi":"10.1007/978-3-540-30116-5_6","title":"Mining Positive and Negative Association Rules: An Approach for Confined Rules","year":2004,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":197,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Association rule learning; Association (psychology); Associative property; Computer science; Process (computing); Data mining; Correlation; Complement (music); Affinity analysis; Space (punctuation); Artificial intelligence; Mathematics; Psychology","authors":[{"name":"Maria-Luiza Antonie","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01876755616653307,"gpt":0.2550179653593767,"spread":0.2362504091928436,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008058853,0.001468557,0.004790636,0.006365721,0.002459797,0.004656602,0.005972693,0.002563363,0.002641395],"category_scores_gemma":[0.03196687,0.001518497,0.004310071,0.006808148,0.002277877,0.006290877,0.00616732,0.003333452,0.00125485],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006402401,"about_ca_system_score_gemma":0.002309177,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002278725,"about_ca_topic_score_gemma":0.004185705,"domain_scores_codex":[0.9900087,0.001606697,0.001517173,0.002155354,0.004379222,0.0003329274],"domain_scores_gemma":[0.9742613,0.01684671,0.001294399,0.004173562,0.002753085,0.0006708881],"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":[0.0009668858,0.0008562336,0.01471082,0.001998806,0.001095533,0.002288864,0.001994393,0.01915748,0.01705073,0.03625412,0.008521005,0.8951052],"study_design_scores_gemma":[0.0002882215,0.0008048496,0.006561678,0.0008094026,0.00184874,0.006441029,0.001777904,0.6216221,0.01842772,0.3068484,0.03428097,0.0002890594],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02110611,0.001456475,0.9730309,0.0003195213,0.00007942766,0.0004466411,0.0007816427,0.001569555,0.001209733],"genre_scores_gemma":[0.1029219,0.0006795523,0.8908889,0.0004308728,0.000169558,0.0005434233,0.00217965,0.0001964586,0.001989719],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008058853,"threshold_uncertainty_score":0.04261982,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1972870286","doi":"10.1145/347090.347166","title":"Can we push more constraints into frequent pattern mining?","year":2000,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":191,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Citation; Computer science; Data science; Library science; World Wide Web","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01379721733865673,"gpt":0.2539804103553299,"spread":0.2401831930166732,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02113976,0.002075373,0.003078308,0.00491877,0.001469285,0.006734315,0.003083233,0.003284102,0.01515075],"category_scores_gemma":[0.180309,0.001846397,0.00285942,0.006646126,0.002649268,0.03217742,0.005961687,0.006994996,0.007451232],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000871788,"about_ca_system_score_gemma":0.002949145,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002967972,"about_ca_topic_score_gemma":0.003725439,"domain_scores_codex":[0.9815974,0.007908004,0.002371131,0.0028332,0.004755954,0.0005343353],"domain_scores_gemma":[0.8777519,0.09165931,0.005038159,0.01253419,0.01075011,0.002266339],"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":[0.00130784,0.0004128552,0.009905577,0.002585056,0.0008626205,0.0005429131,0.0009582473,0.02153503,0.003496741,0.05187677,0.05314139,0.8533749],"study_design_scores_gemma":[0.0003658469,0.0003849583,0.002906358,0.001499934,0.0004765161,0.0006752766,0.001092741,0.1777503,0.003188428,0.7194772,0.09194032,0.0002422085],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01300264,0.009926141,0.9029835,0.05968752,0.00260114,0.0003501363,0.002642248,0.003010062,0.005796563],"genre_scores_gemma":[0.1310818,0.01076523,0.8320078,0.01141707,0.004094565,0.0006477325,0.004121387,0.001134454,0.004729962],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02113976,"threshold_uncertainty_score":0.1117991,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2128723273","doi":"10.1145/775047.775081","title":"Mining frequent item sets by opportunistic projection","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":189,"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; Scalability; Set (abstract data type); Tree (set theory); Data mining; Database transaction; Projection (relational algebra); Algorithm; Database; Mathematics","authors":[{"name":"Junqiang Liu","is_ca":false},{"name":"Yunhe Pan","is_ca":false},{"name":"Ke Wang","is_ca":true},{"name":"Jiawei Han","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05014063780212529,"gpt":0.2591399793286207,"spread":0.2089993415264954,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00144702,0.000784869,0.001171749,0.002552874,0.0007563852,0.001532141,0.001527681,0.0006214076,0.001214316],"category_scores_gemma":[0.008033784,0.000706648,0.0008868196,0.003049658,0.0005443969,0.002765302,0.001843231,0.0007406863,0.0008376313],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002485614,"about_ca_system_score_gemma":0.001010115,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008059439,"about_ca_topic_score_gemma":0.001113496,"domain_scores_codex":[0.9978113,0.0006080067,0.0001986202,0.0003865321,0.000868568,0.0001270816],"domain_scores_gemma":[0.9958759,0.002420501,0.0003201005,0.0006836969,0.0005971739,0.0001026916],"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":[0.0004208209,0.000294087,0.01217409,0.0003885608,0.0002651931,0.0006453535,0.0004937842,0.06641503,0.01149153,0.01082515,0.004272688,0.8923138],"study_design_scores_gemma":[0.00007319939,0.0003480939,0.003224506,0.00008153407,0.0001044332,0.001937826,0.0003702157,0.926689,0.01410211,0.04212935,0.0108787,0.00006099368],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03188948,0.0003071237,0.9649445,0.0001123036,0.00001906182,0.0001890848,0.0003587597,0.001433694,0.0007458855],"genre_scores_gemma":[0.2099542,0.0003075005,0.7863613,0.00008661348,0.00004133215,0.000499147,0.001909515,0.0000866862,0.0007536418],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.002552874,"threshold_uncertainty_score":0.0076527,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2617643752","doi":"10.1038/nmeth.4299","title":"Clustering","year":2017,"lang":"en","type":"article","venue":"Nature Methods","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":186,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Canada's Michael Smith Genome Sciences Centre","funders":"","keywords":"Cluster analysis; Computational biology; Computer science; Biology; Artificial intelligence","authors":[{"name":"Naomi Altman","is_ca":false},{"name":"Martin Krzywinski","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03710119200032522,"gpt":0.4468142170302719,"spread":0.4097130250299466,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009026948,0.001530131,0.001090223,0.004837763,0.00208774,0.003326081,0.002533961,0.001561665,0.04904741],"category_scores_gemma":[0.003448035,0.0006300681,0.001686299,0.004865369,0.0005926371,0.001930103,0.001767952,0.001253854,0.04320178],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001143944,"about_ca_system_score_gemma":0.001949271,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003760311,"about_ca_topic_score_gemma":0.005618083,"domain_scores_codex":[0.99773,0.0002360727,0.0001064231,0.0007782408,0.0009569162,0.0001924442],"domain_scores_gemma":[0.9984125,0.0001512657,0.0000568298,0.0005726009,0.0007455741,0.00006126078],"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":[0.0003169349,0.0001616235,0.001857624,0.0005471633,0.0001619393,0.0001513692,0.0002556545,0.01902731,0.023214,0.05896748,0.1288031,0.7665358],"study_design_scores_gemma":[0.00009507867,0.0002272822,0.005765708,0.0002185595,0.0001868627,0.001107992,0.0007112143,0.242278,0.07042643,0.1062504,0.5725403,0.0001921924],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01110317,0.0009787902,0.848076,0.0006996649,0.0009543774,0.0009255686,0.007591069,0.01260721,0.1170642],"genre_scores_gemma":[0.1249799,0.001343413,0.6559362,0.0008397775,0.0004134817,0.001080238,0.03096812,0.003153373,0.1812855],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.04904741,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2157169143","doi":"10.1109/icdm.2012.20","title":"Direct Discovery of High Utility Itemsets without Candidate Generation","year":2012,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":179,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University; Simon Fraser University","funders":"","keywords":"Computer science; Scalability; Data mining; Pruning; Bounding overwatch; Property (philosophy); Artificial intelligence; Database","authors":[{"name":"Junqiang Liu","is_ca":false},{"name":"Ke Wang","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02712208151631306,"gpt":0.2701453798814657,"spread":0.2430232983651526,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003857282,0.0009950185,0.002255677,0.003457578,0.0008924085,0.001626152,0.00290541,0.001352299,0.001476542],"category_scores_gemma":[0.01880302,0.0006861231,0.00143551,0.004280138,0.0008839745,0.002946792,0.001921192,0.00139474,0.0009645682],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000532463,"about_ca_system_score_gemma":0.001331796,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000892253,"about_ca_topic_score_gemma":0.001927675,"domain_scores_codex":[0.9978101,0.0006567533,0.0001835442,0.0004041546,0.0007587887,0.0001866812],"domain_scores_gemma":[0.9826348,0.01266474,0.001053575,0.002069143,0.001289612,0.0002881175],"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":[0.00084117,0.000845897,0.02396401,0.0006575735,0.0002436475,0.001005921,0.0005525041,0.1174271,0.01265085,0.03124073,0.004893419,0.8056771],"study_design_scores_gemma":[0.0001177705,0.0003395166,0.002494989,0.00008563579,0.0001033235,0.0008598744,0.0001817209,0.9443722,0.009081717,0.03915498,0.003172971,0.00003522848],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07352252,0.0005431063,0.9225376,0.0002578428,0.00003485299,0.0004583084,0.000290837,0.001038494,0.001316487],"genre_scores_gemma":[0.3580469,0.0003348658,0.6377784,0.0001765934,0.00006663569,0.0004146486,0.001301735,0.00009319644,0.00178703],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003857282,"threshold_uncertainty_score":0.02039951,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2103986443","doi":"10.1145/564376.564412","title":"Document clustering with committees","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":174,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Cluster analysis; Information retrieval; Artificial intelligence","authors":[{"name":"Patrick Pantel","is_ca":true},{"name":"Dekang Lin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01839922667157008,"gpt":0.2197457956381992,"spread":0.2013465689666291,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01101068,0.001411087,0.001850168,0.00524215,0.001829945,0.00392261,0.002203913,0.001941019,0.005017842],"category_scores_gemma":[0.04172497,0.0008146841,0.002098172,0.007240537,0.001158686,0.00413127,0.002873637,0.001880964,0.004063326],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001181089,"about_ca_system_score_gemma":0.001550262,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002179129,"about_ca_topic_score_gemma":0.002101592,"domain_scores_codex":[0.9838816,0.008081938,0.001405824,0.002474466,0.003717137,0.0004390365],"domain_scores_gemma":[0.9791988,0.008494368,0.00129603,0.005504679,0.005045365,0.0004606616],"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":[0.0006019738,0.0001560754,0.00273231,0.0006054219,0.0004244772,0.0001321603,0.0005595175,0.1352077,0.006054404,0.06102528,0.02604021,0.7664605],"study_design_scores_gemma":[0.0001395248,0.0003340204,0.001819139,0.0001172049,0.0001968932,0.0005211554,0.0002258051,0.8035221,0.0189684,0.1157939,0.05825046,0.0001115238],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.005525586,0.0005929355,0.9872634,0.0001423815,0.0001015973,0.0003529771,0.0004202763,0.002164887,0.003435979],"genre_scores_gemma":[0.07907792,0.000468123,0.9120487,0.0001350357,0.00023216,0.0006482719,0.0023068,0.000647557,0.00443537],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01101068,"threshold_uncertainty_score":0.0582307,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2005499394","doi":"10.14778/2168651.2168658","title":"Dense subgraph maintenance under streaming edge weight updates for real-time story identification","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":172,"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; Enhanced Data Rates for GSM Evolution; Identification (biology); Social media; Globe; Scale (ratio); Point (geometry); Data science; Edge device; Range (aeronautics); Social network (sociolinguistics); World Wide Web; Artificial intelligence; Geography; Mathematics; Engineering","authors":[{"name":"Albert Angel","is_ca":true},{"name":"Nikos Sarkas","is_ca":true},{"name":"Nick Koudas","is_ca":true},{"name":"Divesh Srivastava","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0132090247020189,"gpt":0.2384151435654986,"spread":0.2252061188634797,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001517123,0.0009236693,0.001450579,0.002850521,0.0007931802,0.001218835,0.003024932,0.001241139,0.00103636],"category_scores_gemma":[0.01736637,0.0007298542,0.0006752348,0.003270765,0.0008244656,0.00416486,0.001646877,0.001088803,0.0005256925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009957345,"about_ca_system_score_gemma":0.001067874,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006720096,"about_ca_topic_score_gemma":0.01021115,"domain_scores_codex":[0.9988111,0.0002821934,0.0001037186,0.0003674589,0.0003314287,0.0001040963],"domain_scores_gemma":[0.991293,0.004996122,0.001146479,0.001515558,0.0007943123,0.0002544668],"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":[0.0007717399,0.0003907945,0.0152247,0.0003176146,0.0001963684,0.0004400513,0.0006669966,0.3944598,0.01724005,0.008579812,0.008912885,0.5527992],"study_design_scores_gemma":[0.00002887312,0.0000539496,0.001239196,0.000008827959,0.00002933911,0.0001529856,0.00008766979,0.9860271,0.00385686,0.007543304,0.0009612164,0.00001063115],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1518908,0.0008149408,0.8400802,0.000516991,0.00005622228,0.0002299817,0.001092897,0.004168876,0.0011491],"genre_scores_gemma":[0.604866,0.0003570198,0.3890345,0.0001449346,0.0001144436,0.0002642142,0.003451779,0.0003094155,0.001457663],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006720096,"threshold_uncertainty_score":0.01336199,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2105347536","doi":"10.1109/icdm.2006.62","title":"DSTree: A Tree Structure for the Mining of Frequent Sets from Data Streams","year":2006,"lang":"en","type":"article","venue":"Proceedings","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":165,"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":"Data stream mining; Computer science; STREAMS; Data mining; Tree (set theory); Flood myth; Data stream; Tree structure; Data structure; Computer network; Mathematics","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Quamrul Khan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03155954862587509,"gpt":0.269129729241148,"spread":0.2375701806152729,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002685164,0.0009806827,0.00144402,0.006015873,0.001387768,0.002009381,0.002058256,0.001252216,0.001988641],"category_scores_gemma":[0.01471681,0.0007398162,0.001585821,0.00785032,0.0006764212,0.004718186,0.001829569,0.001570987,0.001436887],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000700775,"about_ca_system_score_gemma":0.002020203,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002746751,"about_ca_topic_score_gemma":0.005083958,"domain_scores_codex":[0.9980633,0.0004579292,0.0003688622,0.0003331144,0.0006794645,0.00009740916],"domain_scores_gemma":[0.9945767,0.002556811,0.0006312231,0.0007811853,0.001182568,0.0002715788],"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":[0.0008231459,0.0003228838,0.01227521,0.002203172,0.000616367,0.001192764,0.001733235,0.05621209,0.01353893,0.0576793,0.05990762,0.7934954],"study_design_scores_gemma":[0.0002843834,0.0005395743,0.004157865,0.0005510325,0.0002845155,0.002443035,0.0006499967,0.6300499,0.01218926,0.2155492,0.1330932,0.0002079877],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008527115,0.001080357,0.9759679,0.0004829606,0.0001262941,0.0004311415,0.008415407,0.004146759,0.0008219958],"genre_scores_gemma":[0.04499452,0.0009979475,0.9390578,0.0002020817,0.00009359829,0.0006254062,0.01320089,0.0002611717,0.0005665286],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006015873,"threshold_uncertainty_score":0.01420063,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1888464276","doi":"10.1109/icde.2000.839450","title":"Efficient mining of constrained correlated sets","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":161,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Set (abstract data type); Computer science; Meaning (existential); Data mining; Mathematics; Algorithm; Theoretical computer science; Programming language","authors":[{"name":"Gösta Grahne","is_ca":true},{"name":"Laks V. S. Lakshmanan","is_ca":false},{"name":"X. Wang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02343558730825698,"gpt":0.2382634183594904,"spread":0.2148278310512334,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006759698,0.002056451,0.003975282,0.00562707,0.001627437,0.003259034,0.003837269,0.00208623,0.001290092],"category_scores_gemma":[0.05630355,0.00141868,0.002055787,0.01091332,0.001828582,0.005760157,0.003084636,0.002163082,0.0006826252],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001282263,"about_ca_system_score_gemma":0.002365236,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002564427,"about_ca_topic_score_gemma":0.003237038,"domain_scores_codex":[0.9865069,0.005013568,0.001255025,0.003199997,0.003375424,0.0006490421],"domain_scores_gemma":[0.9568281,0.02927061,0.004744997,0.005575967,0.003089018,0.000491355],"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":[0.0008783665,0.001070017,0.04608493,0.002034237,0.001557879,0.001525144,0.001026569,0.4267737,0.008986099,0.09817309,0.01114541,0.4007445],"study_design_scores_gemma":[0.00005363292,0.0002008464,0.003090726,0.0001346611,0.0001004383,0.0007648024,0.0002906381,0.8843546,0.005179349,0.1028433,0.002917478,0.00006955638],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07381917,0.001182573,0.9215789,0.0005087919,0.00004689373,0.0003190778,0.001004935,0.0003479626,0.001191854],"genre_scores_gemma":[0.4224741,0.00102732,0.570931,0.0003431505,0.0001632283,0.000549275,0.003801488,0.0001131294,0.0005972342],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006759698,"threshold_uncertainty_score":0.03574914,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2027440536","doi":"10.1007/s10707-006-9827-8","title":"Mining Co-Location Patterns with Rare Events from Spatial Data Sets","year":2006,"lang":"en","type":"article","venue":"GeoInformatica","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":156,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Property (philosophy); Data mining; Geography; Scale (ratio); Spatial ecology; Feature (linguistics); Scalability; Computer science; Measure (data warehouse); Common spatial pattern; Spatial analysis; Cartography; Pattern recognition (psychology); Artificial intelligence; Mathematics; Statistics; Remote sensing; Ecology; Database; Biology","authors":[{"name":"Yan Huang","is_ca":false},{"name":"Jian Pei","is_ca":true},{"name":"Hui Xiong","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01955937255258091,"gpt":0.2537781371904637,"spread":0.2342187646378827,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002009122,0.001138846,0.001749836,0.008565336,0.0009377273,0.00225678,0.002334873,0.001544142,0.001025702],"category_scores_gemma":[0.01739815,0.0006433903,0.001728356,0.008916751,0.0008499196,0.002969637,0.001719063,0.001624522,0.0008542837],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005129942,"about_ca_system_score_gemma":0.001254956,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002876813,"about_ca_topic_score_gemma":0.005866559,"domain_scores_codex":[0.9967546,0.0004910897,0.0004749267,0.0009640392,0.001096038,0.0002193522],"domain_scores_gemma":[0.9817386,0.01171778,0.002798076,0.002057901,0.00116469,0.0005229184],"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":[0.001965741,0.001029453,0.4223977,0.001574052,0.002012779,0.008147472,0.001385283,0.0807876,0.01554577,0.0128085,0.01217288,0.4401728],"study_design_scores_gemma":[0.0001903638,0.0004480361,0.07581015,0.0002649736,0.001084139,0.01146979,0.002189005,0.8229499,0.01407841,0.05795076,0.01344211,0.0001224227],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5034913,0.003053418,0.4755706,0.00143847,0.0002202765,0.0003604033,0.01161135,0.0019262,0.00232795],"genre_scores_gemma":[0.8098085,0.00110291,0.1668909,0.0001577492,0.0003105595,0.0002696734,0.02028405,0.0000938624,0.001081728],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.008565336,"threshold_uncertainty_score":0.01062536,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2102589938","doi":"10.1109/icdm.2001.989600","title":"Fast parallel association rule mining without candidacy generation","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":154,"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":"Association rule learning; Computer science; Candidacy; Data mining; Process (computing); Parallel algorithm; Tree (set theory); Parallel computing; Mathematics; Operating system","authors":[{"name":"Osmar R. Zai͏̈ane","is_ca":true},{"name":"Mohammed El‐Hajj","is_ca":true},{"name":"P. Lu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03154291050930846,"gpt":0.246651509559753,"spread":0.2151085990504445,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002377546,0.001156852,0.001753475,0.001749079,0.001206379,0.001764253,0.002564358,0.0008078903,0.004559489],"category_scores_gemma":[0.0095999,0.0007774171,0.0009108959,0.002751606,0.0006931872,0.002817019,0.00179129,0.001332037,0.003170575],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005245333,"about_ca_system_score_gemma":0.001875877,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002785875,"about_ca_topic_score_gemma":0.004483517,"domain_scores_codex":[0.9974933,0.0005080865,0.0002169443,0.000564003,0.001036862,0.0001808805],"domain_scores_gemma":[0.9944931,0.002788428,0.0002737584,0.001439094,0.0008511725,0.0001543898],"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":[0.0007280348,0.0002213613,0.003346614,0.0002072522,0.0001321894,0.0002739376,0.0001563752,0.04783426,0.01111913,0.006962092,0.007088258,0.9219304],"study_design_scores_gemma":[0.0004152693,0.0002576605,0.001355015,0.00002855837,0.0001362306,0.001087681,0.0001056105,0.920005,0.02388537,0.03634364,0.01632858,0.0000515525],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02105332,0.0005189105,0.9699533,0.000298455,0.0001260565,0.0002246733,0.0002288889,0.005691119,0.001905347],"genre_scores_gemma":[0.1336124,0.0003081321,0.8597295,0.0002177144,0.0001324634,0.0005022484,0.001199179,0.0004385333,0.003859835],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004559489,"threshold_uncertainty_score":0.01525301,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2063529879","doi":"10.1145/358108.358114","title":"Beyond intratransaction association analysis","year":2000,"lang":"en","type":"article","venue":"ACM Transactions on Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":153,"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":"Badan Riset dan Inovasi Nasional; Helsingin Yliopisto","keywords":"Association rule learning; Apriori algorithm; Database transaction; Computer science; Association (psychology); Transaction data; Affinity analysis; Data mining; Data science; Psychology; Database","authors":[{"name":"Hongjun Lü","is_ca":false},{"name":"Ling Feng","is_ca":false},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.007761500766622136,"gpt":0.2255636195039939,"spread":0.2178021187373717,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01113132,0.001875937,0.002302752,0.008572572,0.001714209,0.006259198,0.004534108,0.001930607,0.003162001],"category_scores_gemma":[0.053749,0.0007605322,0.00260257,0.01121704,0.002134863,0.009649999,0.004894831,0.003945418,0.002204955],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008298,"about_ca_system_score_gemma":0.001951182,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002621876,"about_ca_topic_score_gemma":0.00181592,"domain_scores_codex":[0.9759803,0.005716295,0.002295176,0.00640448,0.008728353,0.0008753921],"domain_scores_gemma":[0.9306611,0.03607225,0.009458841,0.01314494,0.009642364,0.001020508],"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":[0.000577142,0.0004845112,0.1366947,0.001062413,0.001793276,0.002047279,0.001717415,0.07279047,0.003716124,0.1138321,0.00889225,0.6563923],"study_design_scores_gemma":[0.00005483089,0.0002575676,0.01846303,0.0003592691,0.0005655205,0.00257846,0.001339859,0.5614007,0.007468463,0.3578354,0.04948146,0.0001953873],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04174352,0.002331823,0.9470627,0.0008588423,0.0002019643,0.0001958377,0.001189609,0.001032099,0.005383532],"genre_scores_gemma":[0.5820517,0.001686912,0.4050038,0.0009091166,0.0006102166,0.0004093891,0.004974041,0.0004089946,0.003945802],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01113132,"threshold_uncertainty_score":0.05886871,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2500377098","doi":"10.1137/1.9781611972740","title":"Proceedings of the 2004 SIAM International Conference on Data Mining","year":2004,"lang":"en","type":"paratext","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":147,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Queen's University","funders":"","keywords":"Computer science; Library science; Data science","authors":[{"name":"Allan Borodin","is_ca":true},{"name":"Richard M. Karp","is_ca":true},{"name":"Umeshwar Dayal","is_ca":true},{"name":"Chandrika Kamath","is_ca":true},{"name":"David B. Skillicorn","is_ca":true},{"name":"Daniel Barbará","is_ca":true},{"name":"Carlotta Domeniconi","is_ca":true},{"name":"Ning Kang","is_ca":true},{"name":"Hyunsoo Kim","is_ca":true},{"name":"Haesun Park","is_ca":true},{"name":"Sathyakama Sandilya","is_ca":true},{"name":"Rahul Rao","is_ca":true},{"name":"Dimitris Papadopoulos","is_ca":true},{"name":"Dimitrios Gunopulos","is_ca":true},{"name":"Sheng Ma","is_ca":true},{"name":"Jaideep Vaidya","is_ca":true},{"name":"Chris Clifton","is_ca":true},{"name":"Wei‐Guang Teng","is_ca":true},{"name":"Ming-Syan Chen⋆","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07677233096175941,"gpt":0.3190282780559112,"spread":0.2422559470941518,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004984835,0.00144674,0.003487212,0.002357706,0.001295144,0.007638634,0.002146699,0.001527126,0.04636888],"category_scores_gemma":[0.0101239,0.0009064193,0.001131829,0.002660262,0.0009511848,0.003926537,0.002302519,0.003561358,0.03835737],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008182275,"about_ca_system_score_gemma":0.003363977,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004077125,"about_ca_topic_score_gemma":0.007922126,"domain_scores_codex":[0.9980683,0.0004666309,0.0001723405,0.0002732187,0.0009172743,0.0001023333],"domain_scores_gemma":[0.9935331,0.001706189,0.0002127936,0.001537737,0.002288984,0.0007211172],"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":[0.0002611124,0.0006052517,0.001699659,0.000315571,0.0001552702,0.000110887,0.000108398,0.001898218,0.002494809,0.005989239,0.5692426,0.4171189],"study_design_scores_gemma":[0.0001136321,0.000194416,0.003231253,0.0002878381,0.0002890845,0.0007554923,0.000220405,0.07783562,0.008252486,0.03483553,0.8739174,0.00006682827],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.02731902,0.02911725,0.7381778,0.04454402,0.03678926,0.0009024465,0.01597961,0.01989989,0.08727072],"genre_scores_gemma":[0.06367489,0.03792469,0.3639679,0.005441553,0.006054314,0.0009254266,0.05967449,0.003216068,0.4591206],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.04636888,"threshold_uncertainty_score":0.1551193,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2032469362","doi":"10.1007/s10115-006-0032-8","title":"CanTree: a canonical-order tree for incremental frequent-pattern mining","year":2006,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":140,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Computer science; Data mining; Tree (set theory); GSP Algorithm; Database transaction; Tree structure; A priori and a posteriori; Association rule learning; Apriori algorithm; Database; Mathematics; Algorithm; Binary tree","authors":[{"name":"Carson K. Leung","is_ca":true},{"name":"Quamrul Khan","is_ca":true},{"name":"Zhan Li","is_ca":true},{"name":"Tariqul Hoque","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01329467542214002,"gpt":0.2450617536862699,"spread":0.2317670782641299,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001515917,0.001060664,0.001507573,0.004260251,0.001009946,0.002600781,0.002450834,0.001064657,0.008621993],"category_scores_gemma":[0.01106732,0.0008973642,0.001435983,0.005473205,0.0005227218,0.003007544,0.001535109,0.001647342,0.003742522],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006007674,"about_ca_system_score_gemma":0.002887062,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006871248,"about_ca_topic_score_gemma":0.01683791,"domain_scores_codex":[0.9986927,0.0002501215,0.0001691348,0.0001909783,0.0005937807,0.0001032536],"domain_scores_gemma":[0.9958216,0.001742498,0.0002051335,0.0009794168,0.001007629,0.000243681],"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":[0.001130326,0.0003617201,0.003866584,0.001379936,0.000273181,0.0006747421,0.0004561777,0.0217341,0.008582065,0.02837995,0.1013327,0.8318286],"study_design_scores_gemma":[0.0006083248,0.0004746486,0.002462396,0.0003597173,0.0003877381,0.00158242,0.0003367146,0.7047969,0.01561137,0.1211034,0.1520473,0.0002290226],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01109918,0.001114215,0.9371065,0.0003608908,0.0002484014,0.0005986546,0.0114221,0.03531371,0.002736395],"genre_scores_gemma":[0.04933676,0.0006688891,0.9275955,0.0001566743,0.00007818708,0.0003773109,0.01749674,0.001608038,0.00268185],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008621993,"threshold_uncertainty_score":0.0288434,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1482810056","doi":"10.1007/3-540-47887-6_34","title":"Top Down FP-Growth for Association Rule Mining","year":2002,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":136,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Association rule learning; Pruning; Computer science; Tree (set theory); Data mining; Space (punctuation); Depth-first search; Order (exchange); Algorithm; Search algorithm; Mathematics; Combinatorics","authors":[{"name":"Ke Wang","is_ca":true},{"name":"Tang Liu","is_ca":true},{"name":"Jiawei Han","is_ca":true},{"name":"Junqiang Liu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01755477054816567,"gpt":0.2447510518977258,"spread":0.2271962813495601,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004160114,0.001649313,0.002419774,0.003250882,0.001069734,0.002106731,0.002650356,0.001650486,0.004702453],"category_scores_gemma":[0.02013781,0.0007508035,0.001394123,0.004270939,0.0007782606,0.002804583,0.001665313,0.003566591,0.004979011],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000573609,"about_ca_system_score_gemma":0.001242423,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002589772,"about_ca_topic_score_gemma":0.002772877,"domain_scores_codex":[0.9958016,0.001306027,0.0003535417,0.0005799768,0.001699818,0.0002590359],"domain_scores_gemma":[0.9865588,0.008206884,0.0002927264,0.002532829,0.002208454,0.000200354],"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":[0.0005163811,0.0002110511,0.001783234,0.0003932697,0.0001091428,0.0003528225,0.00007121645,0.04496499,0.004809629,0.01273384,0.03179986,0.9022546],"study_design_scores_gemma":[0.00004306068,0.0001500619,0.0005488888,0.00006109945,0.00007695652,0.0005158975,0.0000333767,0.9175825,0.007190314,0.06461658,0.009148632,0.00003268123],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009144033,0.002634874,0.9775622,0.0006316657,0.0004176502,0.0001493978,0.00100019,0.005625545,0.002834463],"genre_scores_gemma":[0.1165473,0.00141813,0.8705649,0.0004070001,0.0005327894,0.0003195851,0.003562004,0.0006663679,0.00598196],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004702453,"threshold_uncertainty_score":0.02200109,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2078064242","doi":"10.1023/b:dami.0000023674.74932.4c","title":"Pushing Convertible Constraints in Frequent Itemset Mining","year":2004,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":134,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Constraint (computer-aided design); Computer science; Data mining; Context (archaeology); Association rule learning; Convertible; A priori and a posteriori; Class (philosophy); Apriori algorithm; Monotonic function; Sequential Pattern Mining; Artificial intelligence; Mathematics; Engineering; Geography","authors":[{"name":"Jian Pei","is_ca":false},{"name":"Jiawei Han","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04812423122534112,"gpt":0.2999252304469889,"spread":0.2518009992216478,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007770571,0.001120368,0.002495039,0.005193333,0.001746164,0.005434884,0.004381738,0.002693982,0.005785996],"category_scores_gemma":[0.07125326,0.001964246,0.002228376,0.007130831,0.002327159,0.01350291,0.005022739,0.005739216,0.001191658],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000814135,"about_ca_system_score_gemma":0.001601174,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003388511,"about_ca_topic_score_gemma":0.003720558,"domain_scores_codex":[0.9913293,0.002363333,0.001001851,0.001404176,0.003482599,0.0004185825],"domain_scores_gemma":[0.9512032,0.03725308,0.002019739,0.005807577,0.003012704,0.0007036111],"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":[0.00125527,0.0006220003,0.009343117,0.001632052,0.0004388645,0.003067011,0.001541019,0.1079403,0.004943031,0.2536542,0.01321659,0.6023466],"study_design_scores_gemma":[0.000134146,0.0002046629,0.001323905,0.0003445988,0.0001979504,0.001268606,0.0004854435,0.428769,0.004620282,0.5465281,0.01603468,0.00008863499],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06120999,0.003186459,0.9239631,0.002204013,0.0004418405,0.0003971095,0.001540459,0.00144744,0.005609586],"genre_scores_gemma":[0.3256478,0.002244064,0.6614846,0.000805379,0.0004445596,0.0003639216,0.003477396,0.0003387751,0.005193418],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007770571,"threshold_uncertainty_score":0.0410952,"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":"W2066881707","doi":"10.1016/j.fss.2004.09.014","title":"Genetic algorithm based framework for mining fuzzy association rules","year":2004,"lang":"en","type":"article","venue":"Fuzzy Sets and Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":133,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Association rule learning; Data mining; Cluster analysis; Apriori algorithm; Fuzzy logic; Fuzzy clustering; Centroid; Computer science; Fuzzy set; Membership function; Mathematics; Fuzzy classification; Defuzzification; Fuzzy set operations; Midpoint; Algorithm; Artificial intelligence; Fuzzy number","authors":[{"name":"Mehmet Kaya","is_ca":false},{"name":"R. Alhajj","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02101407464046413,"gpt":0.2648853785248908,"spread":0.2438713038844267,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001727691,0.0007557167,0.001378483,0.00289299,0.0007857105,0.001368818,0.002376877,0.001576846,0.001569158],"category_scores_gemma":[0.004083386,0.0004172467,0.001156978,0.002999929,0.0008352832,0.001068518,0.0008154025,0.001354367,0.0005697476],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007753518,"about_ca_system_score_gemma":0.001834439,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007117502,"about_ca_topic_score_gemma":0.00574416,"domain_scores_codex":[0.9989292,0.0002934731,0.00008074621,0.0002234753,0.000409503,0.00006355849],"domain_scores_gemma":[0.9988735,0.0006116869,0.00008052623,0.00009733413,0.0003021228,0.00003487795],"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":[0.0001533869,0.0003693632,0.004506623,0.0002134843,0.0005011143,0.0005300371,0.0002782934,0.4493774,0.006882809,0.04849429,0.003326705,0.4853666],"study_design_scores_gemma":[0.00003475226,0.00009088978,0.0006351084,0.000034105,0.00009392959,0.0002062662,0.0000369749,0.9671977,0.001632069,0.02810367,0.001914027,0.00002057927],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00956544,0.0002908913,0.9884279,0.0001368199,0.00003667692,0.00009036373,0.00007759775,0.0004307918,0.000943387],"genre_scores_gemma":[0.1415534,0.0003654182,0.8551487,0.0001848772,0.00005418007,0.0002924609,0.0004129169,0.00004215513,0.001945883],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007117502,"threshold_uncertainty_score":0.01415217,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1514516794","doi":"","title":"Mining Frequent Itemsets Using Support Constraints","year":2000,"lang":"en","type":"article","venue":"National University of Singapore","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":132,"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","authors":[{"name":"Ke Wang","is_ca":false},{"name":"Yu He","is_ca":false},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03351622836162168,"gpt":0.2488833181669882,"spread":0.2153670898053665,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006301444,0.001749058,0.002658737,0.01072505,0.001239713,0.004111799,0.003236637,0.002089959,0.003047208],"category_scores_gemma":[0.04358103,0.001461345,0.002073273,0.01172142,0.0008334661,0.007431414,0.002495256,0.001981865,0.002138536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005601174,"about_ca_system_score_gemma":0.00190253,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001319084,"about_ca_topic_score_gemma":0.001152641,"domain_scores_codex":[0.988754,0.00259995,0.002391665,0.001825782,0.003954821,0.0004737522],"domain_scores_gemma":[0.9636247,0.02537197,0.00387286,0.001998907,0.004421963,0.0007096903],"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":[0.001234299,0.0007934439,0.04355316,0.003426083,0.001465357,0.006758423,0.001166873,0.09633197,0.0142933,0.04808441,0.02139006,0.7615026],"study_design_scores_gemma":[0.0003764578,0.0007691521,0.008468237,0.001170469,0.00054232,0.005973066,0.001833636,0.6878414,0.02336549,0.2285323,0.04084178,0.0002856512],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06611908,0.003679477,0.911036,0.001659017,0.0002122865,0.0008472798,0.01076457,0.001887099,0.003795211],"genre_scores_gemma":[0.2130783,0.002781996,0.7650843,0.0003232636,0.0003049218,0.0009904788,0.01603237,0.0001569177,0.001247393],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01072505,"threshold_uncertainty_score":0.03332561,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2116064560","doi":"10.1007/s10618-005-0248-3","title":"Mining Web Log Sequential Patterns with Position Coded Pre-Order Linked WAP-Tree","year":2005,"lang":"en","type":"article","venue":"Data Mining and Knowledge Discovery","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":131,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Windsor","funders":"","keywords":"Computer science; Web log analysis software; Tree (set theory); Trie; Suffix tree; Prefix; Data mining; Subsequence; Web mining; Node (physics); Web server; Fractal tree index; Tree structure; Interval tree; Web page; Data structure; Algorithm; The Internet; World Wide Web; Mathematics; Binary tree; Static web page","authors":[{"name":"C. I. Ezeife","is_ca":true},{"name":"Yi Lu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03043223887843423,"gpt":0.2874098398765441,"spread":0.2569776009981099,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005223575,0.000408365,0.0005828464,0.004766338,0.0008250256,0.001168149,0.0009660025,0.0007547274,0.001963287],"category_scores_gemma":[0.005336094,0.0003361931,0.0007076811,0.004870439,0.0003288419,0.001636212,0.0006796703,0.0007584738,0.001058037],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004136906,"about_ca_system_score_gemma":0.001311161,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003828702,"about_ca_topic_score_gemma":0.008566475,"domain_scores_codex":[0.9993704,0.00007417148,0.00007431507,0.0001242522,0.0002737573,0.00008307207],"domain_scores_gemma":[0.997134,0.001107086,0.0003126317,0.0005005373,0.0007855956,0.0001601738],"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":[0.001794458,0.001291771,0.1252399,0.0007896124,0.0002964315,0.002702913,0.0005911925,0.07023752,0.02773121,0.01470895,0.01710113,0.737515],"study_design_scores_gemma":[0.00007652842,0.0003251839,0.01447759,0.00008038871,0.0001469751,0.001153518,0.000408414,0.93617,0.01384274,0.02474358,0.008537252,0.00003790231],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5177175,0.0005613393,0.4576658,0.0004768656,0.0001407863,0.0006583247,0.01054572,0.007308394,0.004925318],"genre_scores_gemma":[0.7005172,0.0001977091,0.2846425,0.00008240846,0.00004467002,0.0002708926,0.01110097,0.0001843702,0.002959259],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004766338,"threshold_uncertainty_score":0.007612824,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2131225292","doi":"10.1109/ride.1997.583715","title":"Generalization and decision tree induction: efficient classification in data mining","year":2002,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":118,"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; Scalability; Decision tree; Data mining; Abstraction; Generalization; Relevance (law); Machine learning; Artificial intelligence; Database; Mathematics","authors":[{"name":"Micheline Kamber","is_ca":true},{"name":"Lara Winstone","is_ca":true},{"name":"Wan Gong","is_ca":true},{"name":"Shan Cheng","is_ca":true},{"name":"Jiawei Han","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1137594219951989,"gpt":0.2990887928556513,"spread":0.1853293708604524,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00579642,0.0008039897,0.001786639,0.001856209,0.0006811387,0.001255989,0.002137853,0.001207077,0.001561359],"category_scores_gemma":[0.01090347,0.0004915234,0.001123613,0.004159565,0.001174659,0.003141054,0.001940169,0.002192922,0.001232911],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00067584,"about_ca_system_score_gemma":0.001286525,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001404513,"about_ca_topic_score_gemma":0.001577087,"domain_scores_codex":[0.9959316,0.001843526,0.0002290964,0.0006286651,0.001178977,0.000188206],"domain_scores_gemma":[0.9952112,0.002928444,0.0002648733,0.0008848792,0.0006007531,0.0001098063],"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":[0.0001470477,0.0001452978,0.002335729,0.0003876797,0.0001197314,0.0001750821,0.0002368747,0.06207373,0.003613119,0.07778572,0.01155258,0.8414273],"study_design_scores_gemma":[0.000057927,0.00009443378,0.00101911,0.00008525736,0.00006233215,0.0002632428,0.00005132469,0.7903185,0.004234712,0.1914354,0.01233948,0.00003831813],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004889735,0.001206194,0.9913154,0.0004129019,0.00009244621,0.00009331659,0.0001047855,0.000728648,0.001156616],"genre_scores_gemma":[0.1188905,0.001725698,0.8753523,0.0003788934,0.0004099535,0.0003516329,0.0005797871,0.0001561097,0.002155087],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00579642,"threshold_uncertainty_score":0.03065473,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2970828623","doi":"10.14778/3342263.3342645","title":"Efficient algorithms for densest subgraph discovery","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":115,"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":"Induced subgraph isomorphism problem; Intuition; Computer science; Subgraph isomorphism problem; Graph; Algorithm; Color-coding; Efficient algorithm; Theoretical computer science; Artificial intelligence; Line graph","authors":[{"name":"Yixiang Fang","is_ca":false},{"name":"Kaiqiang Yu","is_ca":false},{"name":"Reynold Cheng","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Xuemin Lin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01307049411882908,"gpt":0.2375503892464292,"spread":0.2244798951276001,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002762877,0.002450887,0.002683559,0.005454037,0.001825824,0.002922295,0.004275587,0.002704507,0.006323712],"category_scores_gemma":[0.015302,0.001469997,0.002781273,0.007930484,0.001277073,0.006116601,0.004554974,0.002406863,0.003039423],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002329322,"about_ca_system_score_gemma":0.004490763,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006597212,"about_ca_topic_score_gemma":0.01349676,"domain_scores_codex":[0.9961489,0.0007727728,0.0002957236,0.001166436,0.001172072,0.0004441442],"domain_scores_gemma":[0.990545,0.004801808,0.0007362175,0.002406641,0.001161817,0.000348474],"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":[0.0003835076,0.0006338007,0.004093499,0.0009383431,0.0002894562,0.0002669595,0.0005675493,0.2048263,0.006386471,0.04434923,0.03735498,0.6999099],"study_design_scores_gemma":[0.0001494718,0.00008014928,0.0006908114,0.00004501609,0.00007703641,0.0003581885,0.0001911556,0.870224,0.002918431,0.1166183,0.008617361,0.00003011276],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01313038,0.0009566081,0.9772779,0.0006265506,0.00007726126,0.0003665573,0.001011948,0.003949999,0.002602762],"genre_scores_gemma":[0.07665639,0.0005479236,0.914748,0.0002441538,0.00008888763,0.000437499,0.004659056,0.0004448317,0.002173252],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006597212,"threshold_uncertainty_score":0.02115494,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2115866740","doi":"10.1145/1008694.1008705","title":"An associative classifier based on positive and negative rules","year":2004,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":106,"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":"Associative property; Association rule learning; Classifier (UML); Computer science; Negation; Artificial intelligence; Correlation; Data mining; Machine learning; Pattern recognition (psychology); Mathematics","authors":[{"name":"Maria-Luiza Antonie","is_ca":true},{"name":"Osmar R. Zai͏̈ane","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01325472196489189,"gpt":0.2646293309659183,"spread":0.2513746090010264,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006198284,0.001085805,0.00201234,0.004335983,0.001648105,0.002783202,0.002593696,0.002225543,0.002590849],"category_scores_gemma":[0.01958128,0.000442028,0.001071766,0.002814491,0.001293365,0.004753964,0.001403352,0.002140178,0.002080984],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000802714,"about_ca_system_score_gemma":0.002282409,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001898457,"about_ca_topic_score_gemma":0.002982385,"domain_scores_codex":[0.994822,0.0008429664,0.0005311761,0.0009184253,0.002594553,0.00029084],"domain_scores_gemma":[0.9867606,0.006343564,0.0008480236,0.001401494,0.00421264,0.0004337037],"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":[0.0004839592,0.0006835004,0.02307659,0.0003738663,0.0002571528,0.000675532,0.0002625295,0.03252986,0.009630866,0.0390249,0.02042492,0.8725764],"study_design_scores_gemma":[0.00009479286,0.0003671423,0.003303732,0.0001331447,0.0002832403,0.001653378,0.0001402375,0.8888863,0.01748005,0.06825415,0.01929204,0.0001118944],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05965774,0.001077503,0.9243758,0.001142531,0.0006157666,0.0004774563,0.0006010299,0.001962784,0.01008948],"genre_scores_gemma":[0.400057,0.0007264052,0.5848842,0.000995758,0.0007829133,0.0006193856,0.001569243,0.0001626056,0.01020248],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006198284,"threshold_uncertainty_score":0.03278005,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W170258093","doi":"","title":"A Unified Framework for Utility Based Measures for Mining Itemsets","year":2006,"lang":"en","type":"article","venue":"","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":103,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Regina","funders":"","keywords":"Data mining; Computer science; Process (computing); Function (biology)","authors":[{"name":"Hong Yao","is_ca":true},{"name":"Howard J. Hamilton","is_ca":true},{"name":"Liqiang Geng","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05371318287178579,"gpt":0.3005510801555023,"spread":0.2468378972837165,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02114251,0.002836423,0.003668793,0.01029049,0.001479352,0.008254201,0.005925688,0.002665036,0.00281415],"category_scores_gemma":[0.04895392,0.001237284,0.004618944,0.01374268,0.00317818,0.01129208,0.004534556,0.005422066,0.001822037],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002853303,"about_ca_system_score_gemma":0.002556284,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00186827,"about_ca_topic_score_gemma":0.0015743,"domain_scores_codex":[0.9797664,0.009034262,0.002739043,0.002225376,0.005726306,0.0005086381],"domain_scores_gemma":[0.9751012,0.01512012,0.001701516,0.003485051,0.004059757,0.0005324051],"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":[0.00007531593,0.0001828854,0.001481741,0.0005445519,0.0003076881,0.0002186421,0.0004216422,0.04359,0.001118881,0.8074175,0.005741848,0.1388994],"study_design_scores_gemma":[0.00004104071,0.0002161459,0.0007460488,0.000315625,0.000145077,0.0004993311,0.0001638758,0.3137647,0.001479177,0.6522124,0.03030483,0.0001116968],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.000611325,0.0007217227,0.997236,0.0002441836,0.00005685146,0.0001073963,0.0001395732,0.0001217305,0.0007613677],"genre_scores_gemma":[0.03799504,0.001853238,0.9562419,0.0002956078,0.0004379739,0.001067479,0.0006674488,0.0001087226,0.0013326],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02114251,"threshold_uncertainty_score":0.1118136,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}