{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":43,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":43,"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":"ee4a48787628","filters":{"venue":"ACM Transactions on Database Systems"}},"results":[{"id":"W2740924709","doi":"10.1145/3068335","title":"DBSCAN Revisited, Revisited","year":2017,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":2625,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University; University of Alberta","funders":"","keywords":"Computer science; DBSCAN; Data mining; Heuristics; Information retrieval; Algorithm; Artificial intelligence; Cluster analysis","authors":[{"name":"Erich Schubert","is_ca":false},{"name":"Jörg Sander","is_ca":true},{"name":"Martin Ester","is_ca":true},{"name":"Hans Peter Kriegel","is_ca":false},{"name":"Xiaowei Xu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03953632886572851,"gpt":0.2875278022523509,"spread":0.2479914733866224,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02027269,0.001379363,0.002766035,0.007284625,0.002735003,0.01428949,0.006094941,0.005047181,0.005576116],"category_scores_gemma":[0.07993302,0.001349472,0.001376653,0.02133653,0.00722649,0.01425294,0.004312376,0.01246938,0.002477925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007508904,"about_ca_system_score_gemma":0.0070958,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02380819,"about_ca_topic_score_gemma":0.02336793,"domain_scores_codex":[0.9756469,0.008321157,0.001247288,0.00279739,0.01113803,0.0008492405],"domain_scores_gemma":[0.9648389,0.01724369,0.0009839458,0.005956223,0.009832587,0.001144631],"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":[0.0005875294,0.0000773709,0.001860643,0.0003581035,0.0001921986,0.0001801117,0.0004659012,0.03215596,0.0005022414,0.4700712,0.110616,0.3829328],"study_design_scores_gemma":[0.0001149713,0.0001209845,0.0007180002,0.0003255919,0.00007723065,0.0009507195,0.0008897463,0.2602594,0.004560767,0.5548252,0.1770125,0.0001449492],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02425857,0.06059712,0.7293007,0.1236185,0.00963214,0.0001707081,0.001193943,0.004991226,0.04623705],"genre_scores_gemma":[0.3181241,0.02714698,0.6015165,0.0228382,0.005715937,0.0002135717,0.001451698,0.001684653,0.02130829],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02380819,"threshold_uncertainty_score":0.1072135,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2171332293","doi":"10.1145/1366102.1366103","title":"Conditional functional dependencies for capturing data inconsistencies","year":2008,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":458,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Bell (Canada)","funders":"Engineering and Physical Sciences Research Council","keywords":"Computer science; Functional dependency; Data integrity; Consistency (knowledge bases); Relational database; SQL; Data mining; Set (abstract data type); Schema (genetic algorithms); Database; Programming language; Theoretical computer science; Algorithm; Information retrieval; Artificial intelligence","authors":[{"name":"Wenfei Fan","is_ca":true},{"name":"Floris Geerts","is_ca":false},{"name":"Xibei Jia","is_ca":false},{"name":"Anastasios Kementsietsidis","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5917106512846138,"gpt":0.4156652633131301,"spread":0.1760453879714837,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00596049,0.001493781,0.001008972,0.003373366,0.001390214,0.002483564,0.003619769,0.001471629,0.004050264],"category_scores_gemma":[0.03202393,0.001027983,0.001926781,0.003194467,0.002039955,0.007278857,0.003140515,0.003481669,0.0005684242],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002176254,"about_ca_system_score_gemma":0.003973315,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009169787,"about_ca_topic_score_gemma":0.01112289,"domain_scores_codex":[0.9896253,0.002408356,0.0009249324,0.001820087,0.004562165,0.0006592409],"domain_scores_gemma":[0.9679843,0.01933556,0.002947006,0.005303276,0.004065137,0.0003646371],"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":[0.0005351298,0.0003044803,0.02062658,0.001195576,0.0002601479,0.001353596,0.001027788,0.2548094,0.02310361,0.3538506,0.0165431,0.32639],"study_design_scores_gemma":[0.00006227894,0.0001581655,0.002392726,0.0001791995,0.0001457845,0.0009433918,0.0002501449,0.7545909,0.03303646,0.1809362,0.02716524,0.0001395025],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01428972,0.0002522221,0.9796801,0.0004469699,0.0000645448,0.0001877647,0.001089224,0.002497668,0.001491692],"genre_scores_gemma":[0.24742,0.0003318627,0.7472166,0.0005092593,0.00009172968,0.0003356464,0.002207848,0.0005435083,0.001343643],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009169787,"threshold_uncertainty_score":0.03152251,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1983304353","doi":"10.1145/974750.974757","title":"A normal form for XML documents","year":2004,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":285,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Functional dependency; XML validation; Document Structure Description; XML; Information retrieval; XML database; XML Schema Editor; Relational database; Theoretical computer science; Programming language; World Wide Web","authors":[{"name":"Marcelo Arenas","is_ca":true},{"name":"Leonid Libkin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02221699491570806,"gpt":0.2760328890509727,"spread":0.2538158941352646,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00406712,0.0008807774,0.0006523783,0.00217742,0.001047653,0.004094907,0.001124408,0.001328205,0.004795743],"category_scores_gemma":[0.0108695,0.0006668132,0.0009428809,0.002293765,0.003074534,0.008987268,0.002093597,0.003126266,0.002772734],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001778342,"about_ca_system_score_gemma":0.00186675,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001902441,"about_ca_topic_score_gemma":0.001546237,"domain_scores_codex":[0.9944525,0.00111833,0.001203705,0.001186199,0.001799777,0.0002393653],"domain_scores_gemma":[0.9947821,0.001698977,0.0004421278,0.001332414,0.001588486,0.0001559346],"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":[0.00004789137,0.00002545,0.0003221927,0.0001415049,0.00001036983,0.0001340775,0.0005377028,0.002320949,0.002792965,0.9056709,0.008598844,0.07939696],"study_design_scores_gemma":[0.0000302481,0.0000650839,0.000157004,0.0001218279,0.00001877286,0.0007831081,0.0001897188,0.02325571,0.005744325,0.6832767,0.2863103,0.00004730373],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.003781419,0.0003703671,0.9838071,0.0009529779,0.000372097,0.0002024131,0.0008956241,0.001656982,0.007961028],"genre_scores_gemma":[0.07161827,0.0009324224,0.9108495,0.0005149269,0.0003442329,0.000592833,0.002533094,0.0007806217,0.01183406],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004795743,"threshold_uncertainty_score":0.02150923,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2057545667","doi":"10.1145/1508857.1508858","title":"The design of a query monitoring system","year":2009,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":169,"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; Query optimization; Online aggregation; Joins; Query expansion; Sargable; Query plan; View; Data mining; Database; Query language; Cardinality (data modeling); Web query classification; Information retrieval; Web search query; Search engine; Database design","authors":[{"name":"Chaitanya Mishra","is_ca":true},{"name":"Nick Koudas","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03427797394678277,"gpt":0.266917922125267,"spread":0.2326399481784843,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005352603,0.0007736016,0.001387581,0.001132105,0.001589407,0.004318254,0.003957888,0.002091341,0.003382208],"category_scores_gemma":[0.009265824,0.001312127,0.0006248969,0.001075928,0.001210797,0.004988434,0.002686638,0.002340015,0.002229899],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009907882,"about_ca_system_score_gemma":0.002512266,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001967019,"about_ca_topic_score_gemma":0.0009435122,"domain_scores_codex":[0.9951591,0.0007941028,0.0006581429,0.00137164,0.001706362,0.0003105866],"domain_scores_gemma":[0.9947228,0.001522095,0.0004387884,0.001066183,0.001734793,0.0005153705],"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":[0.003433106,0.001505417,0.0204144,0.001449778,0.0004314356,0.001791768,0.002955761,0.07104931,0.2389501,0.1010524,0.03948348,0.5174831],"study_design_scores_gemma":[0.0002840129,0.0008010718,0.002684423,0.0001041443,0.0002567926,0.001186645,0.0002963297,0.8149929,0.09537934,0.01474604,0.06909358,0.0001748126],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01072415,0.0002215657,0.9742353,0.0004908641,0.00009500238,0.0007574908,0.0001228673,0.01199624,0.001356462],"genre_scores_gemma":[0.2183189,0.0003086808,0.7740512,0.0006478538,0.00023564,0.001181473,0.0005507702,0.0007732977,0.003932154],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005352603,"threshold_uncertainty_score":0.02830762,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2090154982","doi":"10.1145/357775.357778","title":"Emancipating instances from the tyranny of classes in information modeling","year":2000,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":165,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia; Memorial University of Newfoundland","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Schema (genetic algorithms); Interoperability; Relational database; Class (philosophy); Semi-structured model; Theoretical computer science; Information schema; Data mining; Information retrieval; Artificial intelligence; Database model; World Wide Web","authors":[{"name":"Jeffrey Parsons","is_ca":true},{"name":"Yair Wand","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0253603979568016,"gpt":0.2532512764296321,"spread":0.2278908784728305,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01802305,0.000822044,0.001082368,0.002514358,0.003631796,0.01395128,0.002658458,0.00561579,0.001604655],"category_scores_gemma":[0.02406001,0.0008786718,0.001448057,0.003151048,0.03999387,0.02947354,0.01135887,0.01372405,0.0007311671],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005297569,"about_ca_system_score_gemma":0.003912939,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006898737,"about_ca_topic_score_gemma":0.003849934,"domain_scores_codex":[0.9847799,0.008661253,0.0008816159,0.001923471,0.003014071,0.0007396985],"domain_scores_gemma":[0.9756977,0.01471299,0.001155318,0.006333902,0.001386561,0.0007134534],"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":[0.000006006442,0.000003904168,0.0001420781,0.0000157375,0.000003750736,0.00002006982,0.001323893,0.0003540248,0.0000392839,0.9928981,0.000811284,0.004381841],"study_design_scores_gemma":[0.000005915294,0.000006610186,0.00006721102,0.00006644332,0.000006217553,0.00004872068,0.0003342567,0.003096838,0.0001800038,0.9585594,0.03761668,0.00001179719],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01777291,0.008733952,0.7821018,0.1061516,0.001358026,0.00006673618,0.0001188128,0.0003500306,0.0833462],"genre_scores_gemma":[0.5990382,0.01086457,0.3552932,0.01535972,0.002862491,0.0005916795,0.0002372441,0.0006325676,0.0151204],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01802305,"threshold_uncertainty_score":0.09531611,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1973115038","doi":"10.1145/937598.937601","title":"Are quorums an alternative for data replication?","year":2003,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":144,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Replication (statistics); Scalability; Overhead (engineering); Replicate; Distributed computing; Consistency (knowledge bases); Eventual consistency; Selection (genetic algorithm); Data access; Data consistency; Consistency model; Database; Programming language; Artificial intelligence","authors":[{"name":"Marta Patiño-Martı́nez","is_ca":false},{"name":"Gustavo Alonso","is_ca":false},{"name":"Bettina Kemme","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1108332601636947,"gpt":0.3436451194508417,"spread":0.232811859287147,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01157011,0.0006131641,0.001569866,0.001317472,0.002190847,0.005164957,0.003463542,0.004455815,0.01172012],"category_scores_gemma":[0.03014002,0.0005899933,0.001255195,0.002286652,0.003670238,0.01901526,0.002822037,0.002575668,0.004207607],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00167084,"about_ca_system_score_gemma":0.001612587,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001064291,"about_ca_topic_score_gemma":0.0007340512,"domain_scores_codex":[0.9887707,0.004638792,0.0006261876,0.001211788,0.004104638,0.0006479465],"domain_scores_gemma":[0.9815678,0.008125682,0.001207764,0.005266093,0.003157219,0.0006754333],"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":[0.0003302291,0.00008145993,0.0008300834,0.0005075659,0.00006909565,0.0001681942,0.000520527,0.005314398,0.002734002,0.8717052,0.01006431,0.1076749],"study_design_scores_gemma":[0.0001748802,0.0003841927,0.0003704764,0.0002695957,0.00006602512,0.0008231697,0.0005621046,0.03209814,0.003582762,0.7969105,0.1646564,0.0001016923],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02182115,0.02333485,0.8538089,0.02731874,0.006330821,0.0003146299,0.0002766356,0.0009239999,0.06587033],"genre_scores_gemma":[0.5934011,0.01274791,0.3241893,0.007126458,0.004341282,0.0009666965,0.0004033111,0.000573126,0.05625075],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01172012,"threshold_uncertainty_score":0.06118935,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2057058417","doi":"10.1145/1132863.1132873","title":"Approximation and streaming algorithms for histogram construction problems","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":127,"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":"Histogram; Computer science; Algorithm; Approximation algorithm; Artificial intelligence; Image (mathematics)","authors":[{"name":"Sudipto Guha","is_ca":false},{"name":"Nick Koudas","is_ca":true},{"name":"Kyuseok Shim","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02572179603381434,"gpt":0.2438775195296402,"spread":0.2181557234958259,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003229405,0.001799141,0.001895851,0.00182935,0.001014513,0.002560741,0.003266516,0.001901751,0.00675954],"category_scores_gemma":[0.02151073,0.000960675,0.001508952,0.004392195,0.001267668,0.006773975,0.002802458,0.004084916,0.001920504],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002176328,"about_ca_system_score_gemma":0.001773015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004042483,"about_ca_topic_score_gemma":0.003596257,"domain_scores_codex":[0.9975446,0.0007286948,0.000180392,0.0005269393,0.0007223467,0.0002971133],"domain_scores_gemma":[0.9910372,0.006141248,0.0005208579,0.001265912,0.0007874742,0.0002471737],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0004619743,0.0002885616,0.002013644,0.0006725951,0.0001195198,0.0001153475,0.0004225296,0.4449729,0.002640802,0.1588888,0.01893566,0.3704676],"study_design_scores_gemma":[0.00005154217,0.00004881254,0.000149802,0.00002596525,0.00001918796,0.00008077947,0.00006233665,0.8827558,0.0008467631,0.1129096,0.003034397,0.00001505862],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004234343,0.0007106293,0.9924163,0.0003179004,0.00007288921,0.00008143284,0.0001568499,0.0006124793,0.001397205],"genre_scores_gemma":[0.1296832,0.001671365,0.8623673,0.0002487046,0.0003357001,0.0003939544,0.001289884,0.0003542888,0.003655666],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00675954,"threshold_uncertainty_score":0.02261299,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2130696419","doi":"10.1145/1189769.1189774","title":"Towards multidimensional subspace skyline analysis","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":117,"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":"Skyline; Linear subspace; Subspace topology; Computer science; Semantics (computer science); Theoretical computer science; Space (punctuation); Object (grammar); Online analytical processing; Algorithm; Data mining; Mathematics; Artificial intelligence; Pure mathematics; Programming language","authors":[{"name":"Jian Pei","is_ca":true},{"name":"Yidong Yuan","is_ca":false},{"name":"Xuemin Lin","is_ca":false},{"name":"Wen Jin","is_ca":true},{"name":"Martin Ester","is_ca":true},{"name":"Qing Liu","is_ca":false},{"name":"Wei Wang","is_ca":false},{"name":"Yufei Tao","is_ca":false},{"name":"Jeffrey Xu Yu","is_ca":false},{"name":"Qing Zhang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01799995614887762,"gpt":0.2517425082751469,"spread":0.2337425521262693,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001480806,0.001151441,0.001439685,0.00287127,0.0008965224,0.002861235,0.001278792,0.001037056,0.003505152],"category_scores_gemma":[0.006202721,0.0005003476,0.001762146,0.004168312,0.0014085,0.004678326,0.002876229,0.002546737,0.0009227031],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001202725,"about_ca_system_score_gemma":0.001234743,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004613324,"about_ca_topic_score_gemma":0.002814696,"domain_scores_codex":[0.9986136,0.0004836952,0.00008812942,0.0002407784,0.0004206132,0.0001531998],"domain_scores_gemma":[0.9973559,0.001046159,0.0003612429,0.0004348426,0.0006633897,0.000138462],"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":[0.0002268368,0.00009420802,0.004138349,0.0003518979,0.0001487396,0.0002701817,0.0007423505,0.4142062,0.005550241,0.4169173,0.009528934,0.1478247],"study_design_scores_gemma":[0.00001043261,0.00002952208,0.0003035031,0.00002553876,0.00001092368,0.00003888863,0.0001377108,0.8286785,0.0009101667,0.1653244,0.004512894,0.00001754672],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006109124,0.0002814343,0.9918046,0.000137274,0.00001735217,0.00003813971,0.0002465628,0.0003751121,0.0009903526],"genre_scores_gemma":[0.2232306,0.0009245921,0.7713571,0.0001822327,0.0001479352,0.0002849384,0.001745021,0.000329444,0.001798255],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004613324,"threshold_uncertainty_score":0.01172584,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1997533292","doi":"10.1145/503099.503102","title":"SchemaSQL","year":2001,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":94,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; SQL; Data definition language; Relational database; Schema (genetic algorithms); Database schema; Programming language; Database; Conceptual schema; Semi-structured model; Database model; Information retrieval; Database design","authors":[{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Fereidoon Sadri","is_ca":false},{"name":"Subbu N. Subramanian","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03070971332297829,"gpt":0.2692929818947709,"spread":0.2385832685717926,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007928503,0.00163732,0.00136773,0.002863258,0.001608906,0.008521542,0.007712069,0.00263522,0.05195415],"category_scores_gemma":[0.0167231,0.002017213,0.003671398,0.003119037,0.001317966,0.01076673,0.007945903,0.004416022,0.03368536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001806025,"about_ca_system_score_gemma":0.005147182,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007528958,"about_ca_topic_score_gemma":0.005942179,"domain_scores_codex":[0.9920931,0.001793813,0.001389997,0.001025869,0.003011018,0.0006861448],"domain_scores_gemma":[0.9928777,0.001856938,0.0004363914,0.002218464,0.002177951,0.0004325286],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0005714045,0.0002210554,0.002278758,0.001959796,0.000319364,0.0005276238,0.0008744594,0.005345454,0.004737686,0.2642291,0.5624763,0.1564591],"study_design_scores_gemma":[0.0001045692,0.00003734294,0.0002464096,0.000170302,0.00005001328,0.0003946023,0.0001601677,0.01058917,0.003969657,0.05993186,0.9242697,0.00007622639],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.002752317,0.002143795,0.6742128,0.00398384,0.001196718,0.001673391,0.06256807,0.1747596,0.07670942],"genre_scores_gemma":[0.04623395,0.004812209,0.5766754,0.008545077,0.000719474,0.002589094,0.2721344,0.03560749,0.05268292],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.05195415,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2006315879","doi":"10.1145/1093382.1093388","title":"Capturing summarizability with integrity constraints in OLAP","year":2005,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":93,"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":"Online analytical processing; Computer science; Heuristics; Dimension (graph theory); Theoretical computer science; Set (abstract data type); Data integrity; Aggregate (composite); Intrinsic dimension; Class (philosophy); Space (punctuation); Data mining; Curse of dimensionality; Data warehouse; Database; Artificial intelligence; Mathematics; Programming language","authors":[{"name":"Carlos Hurtado","is_ca":false},{"name":"Claudio Gutiérrez","is_ca":false},{"name":"Alberto O. Mendelzon","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02571648555127341,"gpt":0.2608755870518358,"spread":0.2351591015005624,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005878457,0.001310838,0.001289405,0.002267622,0.001894525,0.004699617,0.003100541,0.002210974,0.001860254],"category_scores_gemma":[0.05647692,0.001327072,0.001570479,0.004584523,0.002596858,0.0152272,0.004563915,0.004115283,0.0003504216],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001787486,"about_ca_system_score_gemma":0.00208044,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007446779,"about_ca_topic_score_gemma":0.007024912,"domain_scores_codex":[0.989626,0.003639373,0.001036937,0.001414661,0.003623906,0.0006589844],"domain_scores_gemma":[0.9549343,0.03099595,0.004135116,0.005306885,0.004117804,0.0005098787],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0006044139,0.0003475128,0.008910788,0.0008833761,0.0002704561,0.001265886,0.00245003,0.4576727,0.01342961,0.2216066,0.006796409,0.2857622],"study_design_scores_gemma":[0.00004065207,0.00008267761,0.0007452574,0.00007559753,0.00006857922,0.0002629271,0.000542265,0.7197083,0.009812639,0.2635342,0.00506311,0.00006373149],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04059713,0.000405712,0.9543814,0.000977661,0.00003952698,0.00028467,0.0005845854,0.00132832,0.001401],"genre_scores_gemma":[0.3083235,0.0003747271,0.6884649,0.0002595992,0.0001045192,0.000260051,0.001250289,0.0002516986,0.0007107307],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007446779,"threshold_uncertainty_score":0.03108865,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2121433213","doi":"10.1145/1386118.1386119","title":"Probabilistic top- <i>k</i> and ranking-aggregate queries","year":2008,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":87,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Ranking (information retrieval); Probabilistic logic; Aggregate (composite); Uncertain data; Semantics (computer science); Tuple; Probabilistic database; Search engine indexing; Materialized view; Dimension (graph theory); Information retrieval; Online aggregation; Data mining; Theoretical computer science; Web search query; Sargable; Search engine; Relational database; Artificial intelligence; View; Database theory; Programming language","authors":[{"name":"Mohamed A. Soliman","is_ca":true},{"name":"Ihab F. Ilyas","is_ca":true},{"name":"Kevin Chen–Chuan Chang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0273376981472049,"gpt":0.2331527256784324,"spread":0.2058150275312275,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006858397,0.001092663,0.002133998,0.00230493,0.001760104,0.006600272,0.003480271,0.002181053,0.004233499],"category_scores_gemma":[0.02158022,0.0009607404,0.002289487,0.005715603,0.002265817,0.01432174,0.003658371,0.002414907,0.001385086],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001583974,"about_ca_system_score_gemma":0.002713371,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002956729,"about_ca_topic_score_gemma":0.004636779,"domain_scores_codex":[0.9911655,0.002234562,0.001252159,0.001461679,0.003307934,0.0005783064],"domain_scores_gemma":[0.9881957,0.004778282,0.001643421,0.003834311,0.001228828,0.0003194635],"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":[0.000524334,0.0003519879,0.004942892,0.001032034,0.0002179672,0.000429306,0.001022333,0.1580803,0.006377633,0.5143227,0.02019545,0.2925031],"study_design_scores_gemma":[0.00005626035,0.0001441691,0.0008622931,0.00006395539,0.00008845164,0.0008789226,0.0004075881,0.6060081,0.005140896,0.3707573,0.01547527,0.0001168516],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008490557,0.0004647756,0.9854857,0.0006847719,0.00005451708,0.0002391189,0.000633522,0.001055037,0.002891975],"genre_scores_gemma":[0.2654943,0.0008102531,0.7283151,0.0004174823,0.0002339408,0.0003359421,0.001291997,0.0002454158,0.002855682],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006858397,"threshold_uncertainty_score":0.0362711,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2150433972","doi":"10.1145/958942.958944","title":"Efficient dynamic mining of constrained frequent sets","year":2003,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":85,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba; University of British Columbia","funders":"","keywords":"Computer science; Context (archaeology); Computation; Exploit; Process (computing); Set (abstract data type); Data mining; Function (biology); Component (thermodynamics); Association rule learning; Algorithm","authors":[{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Carson K. Leung","is_ca":true},{"name":"Raymond T. Ng","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02206422575661429,"gpt":0.269716471572934,"spread":0.2476522458163198,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002063992,0.0007435151,0.001500471,0.003555853,0.0008334368,0.001438092,0.002015759,0.0008768702,0.001258693],"category_scores_gemma":[0.02068395,0.0006825438,0.0008662645,0.004189549,0.000557373,0.003065337,0.001572322,0.0009170439,0.0004948013],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005963174,"about_ca_system_score_gemma":0.001389302,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002008993,"about_ca_topic_score_gemma":0.002441973,"domain_scores_codex":[0.9976305,0.0005482539,0.0002089138,0.000454629,0.0009604961,0.0001972262],"domain_scores_gemma":[0.988818,0.007849645,0.0007416127,0.001198848,0.001197315,0.0001946965],"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":[0.0005615812,0.0002276889,0.008264381,0.0004167715,0.0001845195,0.0006806522,0.0004820435,0.1973359,0.0120729,0.02294302,0.005280123,0.7515504],"study_design_scores_gemma":[0.00003201035,0.00008248608,0.000979432,0.00003015301,0.00003114674,0.0004605775,0.000157434,0.9518961,0.005587383,0.03748089,0.003240937,0.00002144172],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04971014,0.0003800102,0.9472575,0.0002296591,0.00002338786,0.0001569842,0.0004836491,0.0009351385,0.0008235155],"genre_scores_gemma":[0.2993985,0.0003227469,0.6969597,0.00008380792,0.00005631451,0.0002790654,0.00212184,0.0001176622,0.0006603795],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003555853,"threshold_uncertainty_score":0.01091558,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4256119157","doi":"10.1145/1189769.1189778","title":"Peer data exchange","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":69,"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; Data exchange; Tuple; Peer-to-peer; Schema (genetic algorithms); Data sharing; Generalization; Data source; Theoretical computer science; Information retrieval; Distributed computing; World Wide Web","authors":[{"name":"Ariel Fuxman","is_ca":true},{"name":"Phokion G. Kolaitis","is_ca":false},{"name":"Renée J. Miller","is_ca":true},{"name":"Wang-Chiew Tan","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0697215860345619,"gpt":0.2892360718961638,"spread":0.2195144858616019,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008101888,0.001024668,0.001478076,0.001703267,0.002705371,0.00742876,0.006299261,0.002965062,0.03799364],"category_scores_gemma":[0.02550192,0.0007195282,0.00173151,0.003862312,0.001912917,0.01862973,0.01274984,0.003270698,0.01164776],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002143204,"about_ca_system_score_gemma":0.003359398,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002607535,"about_ca_topic_score_gemma":0.001736887,"domain_scores_codex":[0.9871514,0.003276474,0.001403456,0.002588112,0.004486065,0.001094542],"domain_scores_gemma":[0.9833254,0.004333498,0.0009418947,0.008292156,0.00242569,0.0006814611],"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":[0.0002005533,0.0001068285,0.001310169,0.0004978994,0.00009301199,0.0004435896,0.0008353336,0.01114484,0.001094419,0.7859198,0.05530261,0.143051],"study_design_scores_gemma":[0.00009244546,0.00008378852,0.0002993617,0.0001549779,0.0000464619,0.0007383129,0.000483797,0.02700027,0.002485359,0.3553015,0.6132597,0.00005396669],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01175337,0.003030333,0.8350553,0.007096178,0.001691359,0.001400343,0.003066248,0.00343111,0.1334759],"genre_scores_gemma":[0.2979111,0.005308218,0.5440571,0.004950394,0.001605137,0.001927889,0.01416356,0.001873459,0.1282031],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03799364,"threshold_uncertainty_score":0.1271014,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2553567329","doi":"10.1145/3004295","title":"Smart Meter Data Analytics","year":2016,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":67,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Data analysis; Analytics; Smart meter; Big data; Data science; Data mining; Smart grid; Electrical engineering","authors":[{"name":"Xiufeng Liu","is_ca":false},{"name":"Lukasz Golab","is_ca":true},{"name":"Wojciech Golab","is_ca":true},{"name":"Ihab F. Ilyas","is_ca":true},{"name":"Shichao Jin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09289352135828077,"gpt":0.2750155793791962,"spread":0.1821220580209155,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001504524,0.001547865,0.001088516,0.002840151,0.0005625002,0.002966426,0.001777546,0.0007381661,0.01081272],"category_scores_gemma":[0.008826906,0.0004278427,0.0007255081,0.005655748,0.0003680058,0.004133721,0.001936985,0.001264438,0.009049088],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009387495,"about_ca_system_score_gemma":0.001276905,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003514862,"about_ca_topic_score_gemma":0.002756365,"domain_scores_codex":[0.9975321,0.0003220468,0.0002283426,0.0005749078,0.001177527,0.000165071],"domain_scores_gemma":[0.9950044,0.0009157693,0.0003808374,0.001627319,0.001904489,0.000167196],"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":[0.0005341498,0.0004017917,0.02799456,0.0009732327,0.0002824021,0.0003839718,0.0003366106,0.07179943,0.01314963,0.03920014,0.257101,0.5878431],"study_design_scores_gemma":[0.00008415136,0.0001854596,0.01528653,0.000246736,0.00009614063,0.0005062041,0.0005770042,0.5123786,0.04217698,0.07801474,0.350323,0.0001244772],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06512266,0.003470585,0.6509171,0.005853023,0.001510152,0.001337661,0.102508,0.0765393,0.09274152],"genre_scores_gemma":[0.5547987,0.003953628,0.2715014,0.00142777,0.0006869314,0.0005603342,0.1433324,0.002841459,0.02089739],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01081272,"threshold_uncertainty_score":0.03617221,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2054655669","doi":"10.1145/383891.383892","title":"Querying ATSQL databases with temporal logic","year":2001,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":66,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; SQL; Temporal logic; Expressive power; Temporal database; Extension (predicate logic); Database; Translation (biology); Programming language; Representation (politics); Theoretical computer science","authors":[{"name":"Jan Chomicki","is_ca":false},{"name":"David Toman","is_ca":true},{"name":"Michael H. Böhlen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05051783123629092,"gpt":0.2832165114231688,"spread":0.2326986801868779,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002744949,0.0005990789,0.000585182,0.0008457893,0.0006552491,0.00311686,0.001570559,0.0006220808,0.00312904],"category_scores_gemma":[0.007968815,0.0004623743,0.001091362,0.00135565,0.0008998648,0.006252846,0.002212898,0.001512107,0.0006683274],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007246883,"about_ca_system_score_gemma":0.001071635,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002070762,"about_ca_topic_score_gemma":0.001725825,"domain_scores_codex":[0.9956501,0.001140775,0.0006971725,0.0004340682,0.001846306,0.000231466],"domain_scores_gemma":[0.9949386,0.002177153,0.0005139877,0.001131729,0.001100921,0.0001375598],"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":[0.0008290342,0.0002455966,0.003112835,0.0006475863,0.0001504424,0.001251149,0.00127732,0.05012334,0.04116383,0.7193992,0.0179529,0.1638468],"study_design_scores_gemma":[0.0001269976,0.0002144222,0.0005331871,0.0001445234,0.0001487889,0.001196719,0.0004936118,0.5247963,0.05065151,0.3367366,0.08487537,0.00008191195],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02181625,0.0001884276,0.9672636,0.0006163117,0.00007785459,0.00009233407,0.000753834,0.003816975,0.005374375],"genre_scores_gemma":[0.4775787,0.0007945401,0.5091717,0.001011105,0.000310918,0.0003644187,0.003954601,0.0008913259,0.005922742],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00312904,"threshold_uncertainty_score":0.01451683,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2112527958","doi":"10.1145/1538909.1538913","title":"Snapshot isolation and integrity constraints in replicated databases","year":2009,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Seventh Framework Programme; Federación Española de Enfermedades Raras","keywords":"Computer science; Serializability; Correctness; Replica; Data integrity; Snapshot (computer storage); Database; Isolation (microbiology); Distributed computing; Rollback; Concurrency control; Two-phase locking; Distributed database; Fault tolerance; Database transaction; Distributed transaction; Transaction processing; Programming language","authors":[{"name":"Yi Lin","is_ca":true},{"name":"Bettina Kemme","is_ca":true},{"name":"Marta Patiño-Martı́nez","is_ca":false},{"name":"José Enrique Armendáriz-Íñigo","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04464981699830389,"gpt":0.2969985777953138,"spread":0.2523487607970099,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004761039,0.0006814023,0.0007569343,0.001284008,0.001154748,0.003939955,0.002604337,0.001881292,0.001917357],"category_scores_gemma":[0.01512338,0.0007747728,0.0009741287,0.002360591,0.004676317,0.00833667,0.003135661,0.003178799,0.0004947254],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00131503,"about_ca_system_score_gemma":0.001985819,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002052251,"about_ca_topic_score_gemma":0.001319135,"domain_scores_codex":[0.992069,0.002391767,0.00103155,0.001083004,0.002892823,0.0005319468],"domain_scores_gemma":[0.9878407,0.006686011,0.001386865,0.002240202,0.00149998,0.0003462012],"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":[0.00004369821,0.00003834313,0.0003872179,0.0001786209,0.00002275079,0.0005031926,0.0005667345,0.01277136,0.002530899,0.965076,0.001342726,0.01653843],"study_design_scores_gemma":[0.00007298148,0.00009162417,0.0004016838,0.0001195807,0.00005377669,0.0007562116,0.0002860719,0.07632276,0.008460076,0.8764246,0.03695431,0.00005641061],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02412757,0.002673093,0.9545694,0.001631581,0.0002090422,0.0001835933,0.0003711981,0.0004134653,0.01582098],"genre_scores_gemma":[0.6112283,0.00335811,0.3743851,0.0008834837,0.0007010823,0.0006294785,0.0007677842,0.0002196213,0.007826959],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004761039,"threshold_uncertainty_score":0.02517909,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1967155922","doi":"10.1145/1670243.1670250","title":"Automatic virtual machine configuration for database workloads","year":2008,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":61,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"IBM (Canada); University of Waterloo","funders":"","keywords":"Computer science; Online transaction processing; Workload; Database; Software deployment; Virtual machine; Virtualization; Temporal isolation among virtual machines; Distributed computing; Hypervisor; Operating system; Cloud computing; Transaction processing; Database transaction","authors":[{"name":"Ahmed A. Soror","is_ca":true},{"name":"Umar Farooq Minhas","is_ca":true},{"name":"Ashraf Aboulnaga","is_ca":true},{"name":"Kenneth Salem","is_ca":true},{"name":"Peter Kokosielis","is_ca":true},{"name":"Sunil Kamath","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03302318292325333,"gpt":0.2566651177257133,"spread":0.22364193480246,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001503598,0.001367509,0.001301223,0.001696091,0.000886708,0.001494872,0.001869814,0.0006121615,0.001119522],"category_scores_gemma":[0.01018509,0.000691788,0.0004014382,0.001338522,0.0003609108,0.00192015,0.0006926797,0.0008463988,0.0006737361],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009219307,"about_ca_system_score_gemma":0.001417575,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003735709,"about_ca_topic_score_gemma":0.005414796,"domain_scores_codex":[0.9977378,0.0007526432,0.0001931339,0.0005834186,0.0005329253,0.0002000262],"domain_scores_gemma":[0.9959487,0.001630639,0.0004702038,0.001091512,0.0007133852,0.000145582],"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":[0.001306497,0.0003669829,0.02935312,0.0002296421,0.0001215187,0.0002216253,0.0004467648,0.1545494,0.06592966,0.002219539,0.01067612,0.7345793],"study_design_scores_gemma":[0.00005307561,0.00008211908,0.004208793,0.00001533911,0.00002575396,0.0001137237,0.00009857824,0.9709901,0.02098877,0.001684544,0.001703653,0.000035611],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3182343,0.001137228,0.6262749,0.0003227658,0.0001130304,0.0002929646,0.0007050153,0.04982266,0.003097045],"genre_scores_gemma":[0.783838,0.0001578528,0.2132886,0.00006677936,0.00004166602,0.0001021681,0.001131371,0.0007871939,0.0005863814],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003735709,"threshold_uncertainty_score":0.007951915,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2118265701","doi":"10.1145/1189769.1189772","title":"Adaptive rank-aware query optimization in relational databases","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Query optimization; Query plan; Ranking (information retrieval); Rank (graph theory); Cardinality (data modeling); Key (lock); Query language; Relational database; Join (topology); Operator (biology); Sargable; Theoretical computer science; Database; Data mining; Information retrieval; Web search query; Search engine","authors":[{"name":"Ihab F. Ilyas","is_ca":true},{"name":"Walid G. Aref","is_ca":false},{"name":"Ahmed K. Elmagarmid","is_ca":false},{"name":"Hicham G. Elmongui","is_ca":false},{"name":"Rahul Shah","is_ca":false},{"name":"Jeffrey Scott Vitter","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03803229702306821,"gpt":0.2493029119004517,"spread":0.2112706148773835,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002333811,0.001004599,0.001002168,0.0007864573,0.0008179749,0.001642685,0.002125149,0.0007602485,0.0007830902],"category_scores_gemma":[0.004435072,0.0004756686,0.000742527,0.001447423,0.0009257116,0.002723709,0.001579061,0.001227966,0.0003223787],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001117957,"about_ca_system_score_gemma":0.001790185,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009449322,"about_ca_topic_score_gemma":0.01002792,"domain_scores_codex":[0.9965239,0.0008742684,0.0002433738,0.0006374824,0.001373598,0.0003472264],"domain_scores_gemma":[0.9973817,0.001226128,0.0002494766,0.0006344712,0.0003996836,0.0001084938],"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":[0.0007820624,0.0004634802,0.004880896,0.0003034742,0.0001340101,0.0002367611,0.0004714799,0.6879132,0.05093918,0.02222398,0.005960777,0.2256907],"study_design_scores_gemma":[0.00002664925,0.00009804669,0.0004630563,0.000005059003,0.00002669086,0.00006972445,0.00006931829,0.981397,0.009056041,0.007243744,0.001522963,0.00002161896],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09871926,0.001551674,0.8900695,0.0003835795,0.00003305569,0.0001632943,0.0002899571,0.006524642,0.002264976],"genre_scores_gemma":[0.5875003,0.0004896038,0.4096378,0.0001639084,0.00004255056,0.00008522291,0.0004544175,0.0004109014,0.001215206],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009449322,"threshold_uncertainty_score":0.01878864,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1993831977","doi":"10.1145/1242524.1242529","title":"Estimating the selectivity of approximate string queries","year":2007,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":41,"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":"Substring; String (physics); Estimator; Approximate string matching; Computer science; Inverse; String searching algorithm; String metric; Algorithm; String kernel; Data structure; Mathematics; Statistics; Pattern matching; Artificial intelligence","authors":[{"name":"Artūras Mažeika","is_ca":false},{"name":"Michael H. Böhlen","is_ca":false},{"name":"Nick Koudas","is_ca":true},{"name":"Divesh Srivastava","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0257870433612683,"gpt":0.2745853299599472,"spread":0.2487982865986789,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005085166,0.0005060796,0.001518626,0.00309252,0.0006027263,0.002121457,0.001213995,0.00124379,0.0009385201],"category_scores_gemma":[0.04692527,0.0004425911,0.0004132787,0.003505429,0.0008848613,0.004911946,0.002109946,0.0009016392,0.0005464421],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001069317,"about_ca_system_score_gemma":0.001333863,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002231394,"about_ca_topic_score_gemma":0.001714583,"domain_scores_codex":[0.9932328,0.001651903,0.0006822789,0.0007519881,0.003124716,0.0005562415],"domain_scores_gemma":[0.9667745,0.02367123,0.002313762,0.003843604,0.003056146,0.0003408174],"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":[0.003018658,0.0003022316,0.111644,0.0003467071,0.0001796053,0.0006601934,0.0007440451,0.2710517,0.03521158,0.0305765,0.006439193,0.5398257],"study_design_scores_gemma":[0.00003424431,0.0001083725,0.005065416,0.00001775943,0.00001864137,0.0004362765,0.0002497644,0.9638097,0.01507376,0.01399223,0.001168167,0.0000256569],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.421999,0.000864077,0.5711316,0.0005590445,0.00004554847,0.0001413995,0.001100776,0.001957788,0.00220072],"genre_scores_gemma":[0.8812813,0.0004835319,0.1151395,0.0001409624,0.0001121225,0.0001214611,0.001835047,0.0001308115,0.0007553216],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005085166,"threshold_uncertainty_score":0.02689326,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3166524313","doi":"10.1145/3446980","title":"Optimizing One-time and Continuous Subgraph Queries using Worst-case Optimal Joins","year":2021,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Joins; Query plan; Query optimization; Spatial query; Vertex (graph theory); Computation; Theoretical computer science; Graph; Sargable; Algorithm; Database; Web search query; Search engine; Information retrieval","authors":[{"name":"Amine Mhedhbi","is_ca":true},{"name":"Chathura Kankanamge","is_ca":true},{"name":"Semih Salihoğlu","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03081097010833498,"gpt":0.24837275789465,"spread":0.2175617877863151,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002596465,0.00156181,0.001544197,0.0008175771,0.0007957364,0.002458572,0.00194608,0.001020904,0.002735033],"category_scores_gemma":[0.00552527,0.0006355086,0.00160293,0.001595981,0.001564346,0.003597496,0.002161975,0.00187706,0.0004317099],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002204553,"about_ca_system_score_gemma":0.002776509,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008217824,"about_ca_topic_score_gemma":0.009994023,"domain_scores_codex":[0.9962053,0.000866317,0.0001891881,0.0007288205,0.001415471,0.000594979],"domain_scores_gemma":[0.9970862,0.001703308,0.0002422659,0.0004741844,0.0003131272,0.0001810089],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0003593125,0.0002692761,0.00172318,0.0001781436,0.00009899172,0.0001554274,0.0002167011,0.8705576,0.006669228,0.03302196,0.003574761,0.08317532],"study_design_scores_gemma":[0.00003634483,0.0001096924,0.0001947685,0.000008450147,0.00002459657,0.00004463968,0.00008472594,0.973066,0.002795742,0.02227557,0.001346654,0.00001275783],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07978179,0.0006375266,0.9092821,0.0005463997,0.00006422085,0.0002457045,0.0004664778,0.002230835,0.006744858],"genre_scores_gemma":[0.4450021,0.0002994135,0.5498627,0.0001925811,0.0000599535,0.0002407262,0.0009073002,0.0007447496,0.00269038],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008217824,"threshold_uncertainty_score":0.01633996,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2042315498","doi":"10.1145/581751.581752","title":"Searching for dependencies at multiple abstraction levels","year":2002,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Functional dependency; Dependency (UML); Tuple; Online analytical processing; Dependency theory (database theory); Generalization; Abstraction; Scope (computer science); Data mining; Extension (predicate logic); Theoretical computer science; Database; Data warehouse; Relational database; Artificial intelligence; Programming language","authors":[{"name":"Toon Calders","is_ca":false},{"name":"Raymond T. Ng","is_ca":true},{"name":"Jef Wijsen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09694704259971397,"gpt":0.2879969122087368,"spread":0.1910498696090228,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001463022,0.0006634902,0.0007590489,0.003131927,0.0012602,0.001468439,0.001168328,0.0008536466,0.002637543],"category_scores_gemma":[0.009311095,0.000784022,0.001251972,0.001721649,0.0006713869,0.003895084,0.002227258,0.001449586,0.0005226152],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000696687,"about_ca_system_score_gemma":0.001413849,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004696145,"about_ca_topic_score_gemma":0.00770706,"domain_scores_codex":[0.9985217,0.0001884411,0.0001176249,0.0003682005,0.0005274123,0.0002765589],"domain_scores_gemma":[0.9939516,0.003236403,0.000628978,0.001057046,0.0008634249,0.0002626669],"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":[0.0008020498,0.000442295,0.1137378,0.0008394884,0.0003172687,0.003234311,0.002324157,0.09193343,0.03930933,0.07596149,0.01289513,0.6582033],"study_design_scores_gemma":[0.0000449653,0.0003082459,0.02914173,0.000170636,0.0003204052,0.001496567,0.00105286,0.7130979,0.03586806,0.2005826,0.01778738,0.0001286964],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2966595,0.0005798722,0.6925,0.0005916096,0.00003378923,0.0001887519,0.001838196,0.003811893,0.003796342],"genre_scores_gemma":[0.7284258,0.0002298157,0.266701,0.000113445,0.00002280193,0.00008746044,0.002549436,0.0001902721,0.00168001],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004696145,"threshold_uncertainty_score":0.009337604,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1984237017","doi":"10.1145/1735886.1735887","title":"Transparent anonymization","year":2010,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":34,"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":"Nanyang Technological University; Glaucoma Research Foundation","keywords":"Computer science; Data publishing; Adversary; Exploit; Guard (computer science); Generalization; Data anonymization; Privacy protection; Computer security; Adversary model; Information privacy; Data mining; Publishing","authors":[{"name":"Xiaokui Xiao","is_ca":false},{"name":"Yufei Tao","is_ca":false},{"name":"Nick Koudas","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04600808504767626,"gpt":0.2870118829402188,"spread":0.2410037978925426,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01167546,0.001101779,0.00174642,0.002466794,0.003131662,0.004603988,0.003558084,0.002314262,0.002845151],"category_scores_gemma":[0.03434421,0.000900611,0.002577141,0.004107696,0.00390726,0.01045885,0.01019356,0.003882447,0.001279559],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001992857,"about_ca_system_score_gemma":0.0035242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008325664,"about_ca_topic_score_gemma":0.0006849193,"domain_scores_codex":[0.9805659,0.007691221,0.001373148,0.003694986,0.005673845,0.001000922],"domain_scores_gemma":[0.9428051,0.01151058,0.005117999,0.03730102,0.002704519,0.0005609458],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003609377,0.0002130696,0.003624159,0.0004217153,0.0002732486,0.0003518387,0.00141303,0.07484934,0.01020522,0.6811881,0.01029994,0.2167994],"study_design_scores_gemma":[0.0001100919,0.000204642,0.001235907,0.0001814416,0.0001438535,0.002023523,0.0004250088,0.3152133,0.02584674,0.5886887,0.06575921,0.0001675691],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01060707,0.0006134803,0.9803712,0.0008514724,0.0001055941,0.0003259316,0.0002684853,0.0008055572,0.006051157],"genre_scores_gemma":[0.5811741,0.001795194,0.4068354,0.0009319015,0.0004923658,0.0007903954,0.0008257914,0.0002844612,0.006870356],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01167546,"threshold_uncertainty_score":0.06174648,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1971604907","doi":"10.1145/1189769.1189770","title":"Expressive power of an algebra for data mining","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Tuple; Relational algebra; Data mining; Data model (GIS); Relational database; Information retrieval; Theoretical computer science; Algebra over a field; Artificial intelligence; Mathematics; Discrete mathematics","authors":[{"name":"Toon Calders","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Raymond T. Ng","is_ca":true},{"name":"Jan Paredaens","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04314760138777388,"gpt":0.2949769826758806,"spread":0.2518293812881067,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009194762,0.000659631,0.0008335595,0.001738825,0.00167205,0.00597976,0.002484025,0.001395005,0.003602246],"category_scores_gemma":[0.01277294,0.00070165,0.002479599,0.002666438,0.005451893,0.01258819,0.00388147,0.00387823,0.001104486],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00168777,"about_ca_system_score_gemma":0.001738739,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001167556,"about_ca_topic_score_gemma":0.0008031138,"domain_scores_codex":[0.9921499,0.002726253,0.00115508,0.001284668,0.002368671,0.0003154164],"domain_scores_gemma":[0.9900038,0.005505347,0.0006847448,0.002347066,0.001043127,0.0004159093],"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":[0.00002137606,0.0000214975,0.0001827526,0.00007300366,0.00001652428,0.00005239875,0.0002680618,0.001991876,0.0008136031,0.9873696,0.0007066041,0.008482634],"study_design_scores_gemma":[0.00001031906,0.00001757359,0.00005582797,0.00002005421,0.00001040636,0.00008412787,0.0000491363,0.01206711,0.0005956349,0.9794174,0.00765925,0.00001306764],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01312358,0.001206443,0.9675691,0.001865507,0.0001124469,0.00009486209,0.0005946875,0.000532236,0.01490112],"genre_scores_gemma":[0.340354,0.001831931,0.6468545,0.001277994,0.0006556034,0.0005581253,0.00162329,0.0002194492,0.006625156],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009194762,"threshold_uncertainty_score":0.04862714,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3212747350","doi":"10.1145/3483940","title":"On Directed Densest Subgraph Discovery","year":2021,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"University of Hong Kong; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Computer science; Scalability; Induced subgraph isomorphism problem; Enhanced Data Rates for GSM Evolution; Core (optical fiber); Graph; Directed graph; Subgraph isomorphism problem; Efficient algorithm; Theoretical computer science; Algorithm; Artificial intelligence; Database; Line graph","authors":[{"name":"Chenhao Ma","is_ca":false},{"name":"Yixiang Fang","is_ca":false},{"name":"Reynold Cheng","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Wenjie Zhang","is_ca":false},{"name":"Xuemin Lin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01990446779099111,"gpt":0.2521827545106705,"spread":0.2322782867196793,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002531888,0.001817312,0.002175238,0.005323214,0.001790802,0.002454726,0.003381835,0.002139315,0.003285509],"category_scores_gemma":[0.01699894,0.001310837,0.001806253,0.008397704,0.001910776,0.006331075,0.003670482,0.002197367,0.001513414],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002419546,"about_ca_system_score_gemma":0.003373355,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01427237,"about_ca_topic_score_gemma":0.02054987,"domain_scores_codex":[0.9964361,0.0008633066,0.000181354,0.001227121,0.0009943385,0.0002977465],"domain_scores_gemma":[0.9907951,0.004780336,0.0008681868,0.002201963,0.001013846,0.0003405492],"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":[0.000437577,0.0004349783,0.009328983,0.00117219,0.0003538116,0.0006077841,0.0007827538,0.3172434,0.005059159,0.0930625,0.05656396,0.5149528],"study_design_scores_gemma":[0.00007890674,0.00007170536,0.0012896,0.00007865641,0.00007589885,0.0005020746,0.0002982804,0.8021528,0.00298575,0.1767881,0.0156404,0.0000378197],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03344009,0.00251983,0.9513903,0.002017594,0.0001599188,0.0004129443,0.002569981,0.003544391,0.003945029],"genre_scores_gemma":[0.1973065,0.001713229,0.7843827,0.0008372274,0.0002764254,0.0003793303,0.009217323,0.0005950376,0.005292333],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01427237,"threshold_uncertainty_score":0.02837855,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2139000699","doi":"10.1145/2338626.2338633","title":"Reordering rows for better compression","year":2012,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of New Brunswick","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Speedup; Lexicographical order; Row; Sorting; Compression (physics); Algorithm; Data compression; Heuristic; Heuristics; Huffman coding; Compression ratio; Encoding (memory); Parallel computing; Mathematics; Combinatorics; Artificial intelligence; Database","authors":[{"name":"Daniel Lemire","is_ca":false},{"name":"Owen Kaser","is_ca":true},{"name":"Eduardo Gutarra","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04066127576892296,"gpt":0.289049769394661,"spread":0.248388493625738,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008232598,0.001279708,0.001041405,0.001478333,0.0006685902,0.002378405,0.001317414,0.0009073094,0.01438599],"category_scores_gemma":[0.007893777,0.0005491061,0.0006762217,0.002980309,0.0007031323,0.004133627,0.001246539,0.001414162,0.007007372],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005860187,"about_ca_system_score_gemma":0.001498546,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001083248,"about_ca_topic_score_gemma":0.002479769,"domain_scores_codex":[0.9986203,0.0002223532,0.0002174289,0.0002303942,0.0005053591,0.0002042996],"domain_scores_gemma":[0.9934715,0.001740957,0.0004224986,0.003034873,0.001198745,0.0001314277],"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":[0.00100222,0.0003503983,0.003524898,0.0007283795,0.0001109017,0.0005643019,0.0004085753,0.02535396,0.1934589,0.02515553,0.04149568,0.7078462],"study_design_scores_gemma":[0.0003211125,0.0009703787,0.003596814,0.0003803832,0.0002395801,0.002404997,0.0008068283,0.2302873,0.5603638,0.05274783,0.1476525,0.0002285774],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1726577,0.004081518,0.774421,0.003313716,0.001819395,0.0005603269,0.002799648,0.01901615,0.02133063],"genre_scores_gemma":[0.2510335,0.001268035,0.7286454,0.001453515,0.0003893854,0.0002006292,0.003628042,0.002571691,0.01080971],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01438599,"threshold_uncertainty_score":0.04812598,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2010916971","doi":"10.1145/1132863.1132868","title":"Integrating XML data sources using approximate joins","year":2006,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":28,"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; Joins; XML validation; XML; Efficient XML Interchange; Document Structure Description; XML Schema (W3C); XML database; Information retrieval; Streaming XML; Data mining; Set (abstract data type); Tree (set theory); XML Encryption; Database; Programming language; World Wide Web","authors":[{"name":"Sudipto Guha","is_ca":false},{"name":"H. V. Jagadish","is_ca":false},{"name":"Nick Koudas","is_ca":true},{"name":"Divesh Srivastava","is_ca":false},{"name":"Ting Yu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06748389564404962,"gpt":0.2828255905583581,"spread":0.2153416949143085,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01359359,0.001268061,0.00234265,0.004707168,0.001813167,0.007316389,0.003677216,0.001997215,0.001168524],"category_scores_gemma":[0.04275217,0.001227675,0.002008241,0.01020504,0.001651267,0.01106493,0.00909385,0.002113866,0.0008928873],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001571089,"about_ca_system_score_gemma":0.002551043,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003814023,"about_ca_topic_score_gemma":0.003764135,"domain_scores_codex":[0.9802822,0.006270072,0.001510047,0.003074972,0.00837563,0.0004871135],"domain_scores_gemma":[0.977114,0.01119548,0.002524092,0.006072334,0.002619655,0.000474393],"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":[0.000540636,0.0002671052,0.01320487,0.0004129099,0.0004222191,0.0007013024,0.002189641,0.3480946,0.01541524,0.1503491,0.004384737,0.4640176],"study_design_scores_gemma":[0.0000320051,0.0001460049,0.0008764026,0.0000440965,0.00006431135,0.0003683513,0.0004017719,0.903037,0.008182772,0.08020989,0.006586715,0.00005068056],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01470884,0.0003687578,0.9829029,0.0001556621,0.00002253231,0.00009485552,0.0001553124,0.0009198168,0.0006712509],"genre_scores_gemma":[0.1569842,0.0004293297,0.8401152,0.0001206393,0.00006784721,0.0001603939,0.001103348,0.0002200469,0.0007988826],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01359359,"threshold_uncertainty_score":0.07189059,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1971512992","doi":"10.1145/1005566.1005570","title":"A compressed accessibility map for XML","year":2004,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Access Control and Trust","field":"Social Sciences","cited_by":28,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Efficient XML Interchange; XML validation; XML Schema (W3C); XML database; XML; Streaming XML; XML framework; XML Encryption; Simple API for XML; Database; Document Structure Description; XML Schema Editor; XML Signature; Information retrieval; Data mining; World Wide Web","authors":[{"name":"Ting Yu","is_ca":false},{"name":"Divesh Srivastava","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"H. V. Jagadish","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05599528690366123,"gpt":0.3495931178292262,"spread":0.293597830925565,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004691216,0.0004323388,0.0005204419,0.001730321,0.0008155746,0.001342063,0.0009376166,0.0006156758,0.007865149],"category_scores_gemma":[0.008138038,0.0002691023,0.0005298576,0.002285307,0.0007888674,0.003469142,0.002504644,0.0009023851,0.001487688],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007047498,"about_ca_system_score_gemma":0.001005508,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00382801,"about_ca_topic_score_gemma":0.002952745,"domain_scores_codex":[0.9991061,0.0001617612,0.00006736426,0.0001595935,0.000428358,0.00007668817],"domain_scores_gemma":[0.9975442,0.001030188,0.0001239887,0.0007003669,0.0005135685,0.00008751998],"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":[0.0005008763,0.0001481887,0.001681469,0.0004021597,0.00003189439,0.0007170132,0.0008736888,0.09208555,0.01452308,0.2699888,0.04108736,0.5779599],"study_design_scores_gemma":[0.00006280319,0.0001643724,0.001058097,0.00006391917,0.00002967756,0.001061631,0.0004321215,0.6672449,0.01726962,0.2246378,0.08790958,0.00006537665],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02066894,0.0003276867,0.9622872,0.0004524059,0.0001116407,0.0002629413,0.002553617,0.005315608,0.00801992],"genre_scores_gemma":[0.3415627,0.000551429,0.6417908,0.000198183,0.0001570898,0.000691215,0.006158098,0.0006011294,0.008289333],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007865149,"threshold_uncertainty_score":0.02631158,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2053495332","doi":"10.1145/2487259.2487260","title":"Analysis and optimization for boolean expression indexing","year":2013,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":28,"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; Regular expression; Theoretical computer science; Search engine indexing; Boolean expression; Tree (set theory); Matching (statistics); Search tree; Data structure; Tree structure; String searching algorithm; Data mining; Algorithm; Pattern matching; Boolean function; Binary tree; Search algorithm; Artificial intelligence; Mathematics","authors":[{"name":"Mohammad Sadoghi","is_ca":true},{"name":"Hans‐Arno Jacobsen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02055275887691247,"gpt":0.2477562282413695,"spread":0.227203469364457,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001793019,0.0007183676,0.001065706,0.001482319,0.0007510118,0.003008807,0.001965148,0.0006793574,0.004935978],"category_scores_gemma":[0.01121253,0.000394244,0.000975094,0.004083353,0.001163289,0.005621689,0.001666186,0.001464263,0.001139146],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002402986,"about_ca_system_score_gemma":0.00326325,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003546303,"about_ca_topic_score_gemma":0.005173323,"domain_scores_codex":[0.9961573,0.0007042459,0.0002750265,0.0003671852,0.002035857,0.0004603751],"domain_scores_gemma":[0.9951615,0.002689472,0.0003229961,0.0009371674,0.0007833068,0.0001054825],"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":[0.0005240041,0.0002933208,0.003186145,0.000523604,0.00007034847,0.0001231824,0.0002288591,0.2545762,0.01379962,0.2498584,0.01672461,0.4600917],"study_design_scores_gemma":[0.00003023218,0.00008715544,0.0003147385,0.0000230143,0.0000242873,0.00007724085,0.00006156626,0.9092592,0.006120636,0.07774493,0.006239515,0.00001752145],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03071239,0.001323686,0.9537104,0.0005846177,0.00009096297,0.0001640744,0.0006802697,0.002330491,0.01040305],"genre_scores_gemma":[0.3858693,0.001150598,0.6025813,0.0003091113,0.0001502223,0.000394422,0.002134524,0.0007629616,0.006647552],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004935978,"threshold_uncertainty_score":0.01743501,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2104120015","doi":"10.1145/1538909.1538910","title":"Anonymization-based attacks in privacy-preserving data publishing","year":2009,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":27,"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":"Research Grants Council, University Grants Committee","keywords":"Computer science; Adversary; Data publishing; Overhead (engineering); Confidentiality; Computer security; Information loss; k-anonymity; Information sensitivity; Publishing; Information privacy; Internet privacy; Artificial intelligence","authors":[{"name":"Raymond Chi-Wing Wong","is_ca":false},{"name":"Ada Wai-Chee Fu","is_ca":false},{"name":"Ke Wang","is_ca":true},{"name":"Jian Pei","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07597995798891426,"gpt":0.3100199144698071,"spread":0.2340399564808928,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01705674,0.0008382078,0.002123404,0.00199752,0.003471215,0.005810082,0.003694669,0.004506863,0.0009468051],"category_scores_gemma":[0.04646838,0.001258143,0.002080936,0.003278606,0.006053536,0.01571141,0.009024018,0.004878683,0.000410896],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002207114,"about_ca_system_score_gemma":0.002576445,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004633697,"about_ca_topic_score_gemma":0.0002781994,"domain_scores_codex":[0.9641204,0.02060343,0.002420731,0.00338389,0.007822008,0.001649555],"domain_scores_gemma":[0.9185688,0.0446494,0.006225948,0.02780049,0.001889821,0.0008656333],"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":[0.001246032,0.0003227284,0.006576279,0.0004738539,0.000380189,0.0008553498,0.002316958,0.1349982,0.01066379,0.6992227,0.00390619,0.1390378],"study_design_scores_gemma":[0.0001119433,0.0002781649,0.0005268621,0.0001143601,0.0001233913,0.001791732,0.000385651,0.4335323,0.02613119,0.5223861,0.01453148,0.00008681895],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05414714,0.001052606,0.9359488,0.003165745,0.0001046674,0.0002553584,0.0001259009,0.00064878,0.004551047],"genre_scores_gemma":[0.8135076,0.0006547622,0.1828237,0.0006502374,0.0002000567,0.0002462239,0.0001615146,0.00008718859,0.001668705],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01705674,"threshold_uncertainty_score":0.09020573,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1998597078","doi":"10.1145/1670243.1670248","title":"An information-theoretic analysis of worst-case redundancy in database design","year":2008,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"Seventh Framework Programme; Engineering and Physical Sciences Research Council","keywords":"Computer science; Redundancy (engineering); Schema (genetic algorithms); Functional dependency; Database schema; Theoretical computer science; Data integrity; Relational database; Data mining; Database design; Information retrieval; Database","authors":[{"name":"Solmaz Kolahi","is_ca":true},{"name":"Leonid Libkin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03518479320693964,"gpt":0.275027395432391,"spread":0.2398426022254513,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01224363,0.001280755,0.001113995,0.003294484,0.0009495663,0.004377198,0.002695252,0.001421061,0.002386144],"category_scores_gemma":[0.05150817,0.001081934,0.001241537,0.00375176,0.005014334,0.01085169,0.001890431,0.002488062,0.0004928748],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003522623,"about_ca_system_score_gemma":0.001565015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007024063,"about_ca_topic_score_gemma":0.000611911,"domain_scores_codex":[0.9830073,0.006897724,0.001115581,0.001404444,0.006760076,0.0008148199],"domain_scores_gemma":[0.9477338,0.03699489,0.003410354,0.007671615,0.003742899,0.0004464647],"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":[0.0001770875,0.00008537819,0.001147099,0.0003209697,0.00009803643,0.0001927236,0.0004042596,0.3119412,0.004790825,0.6201838,0.001471719,0.05918695],"study_design_scores_gemma":[0.00002261202,0.0001572303,0.0003657597,0.00005468235,0.00006207993,0.000305021,0.00007572305,0.4110947,0.004040639,0.5806116,0.003169416,0.0000405118],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03455323,0.001074143,0.9562877,0.0009785042,0.00004430057,0.00007382144,0.0001312122,0.0002076006,0.006649425],"genre_scores_gemma":[0.6734915,0.001543706,0.3209257,0.0003375761,0.0003230463,0.000383941,0.0003414386,0.0002010411,0.002451919],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01224363,"threshold_uncertainty_score":0.06475127,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2051445745","doi":"10.1145/507234.507237","title":"A logical foundation for deductive object-oriented databases","year":2002,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Logic, Reasoning, and Knowledge","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University","funders":"","keywords":"Computer science; Programming language; Deductive database; Syntax; Semantics (computer science); Class (philosophy); Object (grammar); Well-founded semantics; Logic programming; Inheritance (genetic algorithm); Operational semantics; Database; Natural language processing; Artificial intelligence; Denotational semantics","authors":[{"name":"Mengchi Liu","is_ca":true},{"name":"Gillian Dobbie","is_ca":false},{"name":"Tok Wang Ling","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07233550198391613,"gpt":0.2909125887188853,"spread":0.2185770867349692,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008905284,0.0007730386,0.001022499,0.003470689,0.003239306,0.009901508,0.003121301,0.002969469,0.003997218],"category_scores_gemma":[0.01280269,0.001459718,0.002258172,0.003078677,0.01063456,0.019102,0.004774418,0.006357907,0.002040011],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002952732,"about_ca_system_score_gemma":0.003697742,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002254774,"about_ca_topic_score_gemma":0.001323051,"domain_scores_codex":[0.9940128,0.001461782,0.0007951354,0.0008406765,0.002505831,0.0003838234],"domain_scores_gemma":[0.9916986,0.00417873,0.0006791211,0.001040406,0.001947707,0.0004554542],"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":[0.000005690585,0.00002012545,0.00007391792,0.00005714865,0.000006104362,0.00008243231,0.0002636209,0.0005476339,0.0002057233,0.9920258,0.001125829,0.005585916],"study_design_scores_gemma":[0.00002023863,0.00001596669,0.00004898186,0.0000614052,0.00001024291,0.0001075838,0.00008008367,0.003917661,0.0003844614,0.9625782,0.03276023,0.00001496075],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004130984,0.003413402,0.9695057,0.006341223,0.0004672771,0.0001921729,0.0003018084,0.0006547534,0.01499286],"genre_scores_gemma":[0.1374639,0.005820825,0.8409525,0.004437851,0.001561882,0.0006327687,0.0009320486,0.0003185922,0.007879667],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009901508,"threshold_uncertainty_score":0.04709619,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4245823248","doi":"10.1145/1114244.1114250","title":"Optimization of query streams using semantic prefetching","year":2005,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Latency (audio); Context (archaeology); Server; Query optimization; STREAMS; Data mining; Information retrieval; Database; Computer network","authors":[{"name":"Ivan T. Bowman","is_ca":true},{"name":"Kenneth Salem","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03100413104897517,"gpt":0.2743439780064526,"spread":0.2433398469574774,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009190089,0.0008454323,0.0008359904,0.0006126515,0.0003955376,0.0009152305,0.001200068,0.0004980967,0.001479559],"category_scores_gemma":[0.005832576,0.0003694281,0.0003353211,0.001127286,0.0004402454,0.001912281,0.0008571456,0.000782013,0.0003032121],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008016896,"about_ca_system_score_gemma":0.001417621,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005737296,"about_ca_topic_score_gemma":0.008995467,"domain_scores_codex":[0.9990371,0.0001721844,0.00007690735,0.0002204392,0.0003580508,0.0001353268],"domain_scores_gemma":[0.9981303,0.0008649971,0.0001484019,0.0004084588,0.0003687101,0.00007916435],"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":[0.001562968,0.0005829624,0.01173961,0.0001673898,0.0000954908,0.0003082168,0.0002176159,0.4761916,0.05679208,0.004506723,0.008721353,0.439114],"study_design_scores_gemma":[0.00003736059,0.0000623432,0.0006391784,0.00000196193,0.00001092353,0.00003112136,0.00002787778,0.9846859,0.01229747,0.001624621,0.000573143,0.000008187427],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5548804,0.0009343303,0.4227339,0.0006581889,0.00008899773,0.0002899914,0.0007340335,0.01497391,0.004706382],"genre_scores_gemma":[0.8958384,0.0001781861,0.1010137,0.0001139459,0.00004800389,0.00008303603,0.0007642427,0.0002627667,0.00169763],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005737296,"threshold_uncertainty_score":0.01140779,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2002932370","doi":"10.1145/1061318.1061324","title":"Concise descriptions of subsets of structured sets","year":2005,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Automatic summarization; Theoretical computer science; Set (abstract data type); Minimum description length; Online analytical processing; Hierarchy; Cover (algebra); Context (archaeology); Representation (politics); Data mining; Algorithm; Artificial intelligence; Data warehouse","authors":[{"name":"Ken Q. Pu","is_ca":true},{"name":"Alberto O. Mendelzon","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0319806076430219,"gpt":0.2639264952448165,"spread":0.2319458876017946,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00221181,0.0006411542,0.001026165,0.002618103,0.0007449456,0.003211886,0.001784012,0.0009176666,0.00416925],"category_scores_gemma":[0.01179663,0.0007064867,0.001002792,0.004164882,0.001209973,0.00964782,0.001979602,0.001231166,0.0007707657],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001378297,"about_ca_system_score_gemma":0.0008772755,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001106882,"about_ca_topic_score_gemma":0.001655707,"domain_scores_codex":[0.9971274,0.0009212605,0.0003468556,0.0004465827,0.0009473841,0.0002105905],"domain_scores_gemma":[0.9936215,0.003480811,0.0008641322,0.001112746,0.0007096498,0.0002111538],"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":[0.0002181742,0.00008357257,0.001282646,0.0006732363,0.00008319858,0.000565473,0.001424717,0.07523867,0.004728945,0.7949336,0.009184256,0.1115835],"study_design_scores_gemma":[0.00004785293,0.000148856,0.0005138396,0.0001562373,0.00007466788,0.0005333452,0.0007957797,0.2015044,0.005666014,0.7435764,0.04693099,0.00005169573],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07561506,0.002277653,0.9070789,0.001566524,0.0001013536,0.0002128193,0.003597157,0.000788936,0.00876161],"genre_scores_gemma":[0.4650638,0.002031705,0.515063,0.0003682663,0.000219762,0.0005280953,0.007841733,0.000353566,0.008530146],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00416925,"threshold_uncertainty_score":0.01394755,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3172623769","doi":"10.1145/3451159","title":"Graph Indexing for Efficient Evaluation of Label-constrained Reachability Queries","year":2021,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Winnipeg","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Reachability; Computer science; Vertex (graph theory); Search engine indexing; Transitive closure; Combinatorics; Graph; Theoretical computer science; Discrete mathematics; Mathematics; Information retrieval","authors":[{"name":"Yangjun Chen","is_ca":true},{"name":"Gagandeep Singh","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06869570056074788,"gpt":0.3123199311397553,"spread":0.2436242305790074,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001428895,0.001276264,0.001663495,0.003192141,0.001104943,0.002818245,0.002965133,0.001131501,0.005496247],"category_scores_gemma":[0.01060308,0.0006474162,0.001119495,0.005570073,0.000885601,0.007714794,0.002332157,0.001497706,0.00165791],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002178643,"about_ca_system_score_gemma":0.003218999,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007362012,"about_ca_topic_score_gemma":0.01088459,"domain_scores_codex":[0.9969888,0.0004459215,0.0003402056,0.0005408675,0.001339283,0.0003449882],"domain_scores_gemma":[0.9934728,0.002859826,0.0004901552,0.002087177,0.0008483431,0.0002417273],"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":[0.002815051,0.0007358065,0.006244778,0.001391868,0.0002516806,0.0005981717,0.0008166725,0.08818924,0.08228969,0.06010265,0.05407065,0.7024937],"study_design_scores_gemma":[0.0002340279,0.000283384,0.001444494,0.00006239052,0.0000863442,0.0005046757,0.0003706815,0.864736,0.04248076,0.07646699,0.01324978,0.00008046638],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08774786,0.001757005,0.8453371,0.0007236975,0.0001937649,0.0006517923,0.007715393,0.04781965,0.008053695],"genre_scores_gemma":[0.4025085,0.0005033346,0.5793366,0.0002921001,0.0001183913,0.0004517451,0.01292519,0.001662584,0.002201657],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007362012,"threshold_uncertainty_score":0.01838672,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2080555007","doi":"10.1145/1806907.1806909","title":"Continuous online index tuning in moving object databases","year":2010,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":15,"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":"Media Development Authority - Singapore","keywords":"Computer science; Granularity; B-tree; Overhead (engineering); Search engine indexing; Tree (set theory); Grid; Data mining; Workload; Set (abstract data type); Object (grammar); Database; Binary tree; Information retrieval; Algorithm; Artificial intelligence; Mathematics","authors":[{"name":"Chen Su","is_ca":false},{"name":"Mário A. Nascimento","is_ca":true},{"name":"Beng Chin Ooi","is_ca":false},{"name":"Kian‐Lee Tan","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02813362435550357,"gpt":0.2745241931610565,"spread":0.2463905688055529,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002764337,0.0008555252,0.001457661,0.001739928,0.001170492,0.0029372,0.003988688,0.0009649663,0.001234371],"category_scores_gemma":[0.01169054,0.0008134115,0.0004406682,0.003571848,0.0008167914,0.005784246,0.003230914,0.001010173,0.0007534404],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001079001,"about_ca_system_score_gemma":0.001329181,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004184304,"about_ca_topic_score_gemma":0.003741119,"domain_scores_codex":[0.9971294,0.0004205107,0.0003096644,0.0006414478,0.001208231,0.000290712],"domain_scores_gemma":[0.995137,0.001529926,0.0004266973,0.001780939,0.0007201069,0.0004053336],"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":[0.001861351,0.0006953197,0.01551937,0.0005334949,0.0001937813,0.0006266456,0.0008018925,0.1586429,0.05557637,0.0150023,0.02525389,0.7252927],"study_design_scores_gemma":[0.000125762,0.0002449724,0.00293917,0.00003152365,0.00004676721,0.0003772349,0.0002107677,0.9523668,0.01697745,0.01369774,0.01291677,0.00006496484],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2483349,0.006258551,0.7087886,0.0005011486,0.0003841905,0.0004225628,0.00100842,0.02613952,0.008162039],"genre_scores_gemma":[0.7060131,0.0009675279,0.2880599,0.0002683095,0.0001781021,0.000216054,0.001605979,0.0008411545,0.001849918],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004184304,"threshold_uncertainty_score":0.01461935,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2766916843","doi":"10.1145/3110214","title":"Blazes","year":2017,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Air Force Office of Scientific Research; Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Computer science; Debugging; Consistency (knowledge bases); Scalability; Distributed computing; Programming language; Database; Artificial intelligence","authors":[{"name":"Peter Alvaro","is_ca":false},{"name":"Neil Conway","is_ca":false},{"name":"Joseph M. Hellerstein","is_ca":false},{"name":"David Maier","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03784811562260383,"gpt":0.288350591027029,"spread":0.2505024754044252,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001252693,0.001097084,0.0005866899,0.001251886,0.00121197,0.003239119,0.002169505,0.001178945,0.126576],"category_scores_gemma":[0.003894606,0.0007226714,0.0007934598,0.0008883757,0.0008143197,0.004817856,0.003395005,0.002078132,0.07684606],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001193685,"about_ca_system_score_gemma":0.001879151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004654299,"about_ca_topic_score_gemma":0.006614449,"domain_scores_codex":[0.998725,0.0001577023,0.00008425568,0.0002863727,0.000585271,0.0001613791],"domain_scores_gemma":[0.9982772,0.0002846111,0.0001071403,0.0006670548,0.000534786,0.0001293042],"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":[0.0008645735,0.0001999682,0.003359259,0.0008792671,0.00008109592,0.0003465731,0.0008728915,0.004466765,0.01814054,0.1676185,0.4445945,0.358576],"study_design_scores_gemma":[0.00005149277,0.00005396195,0.0007398321,0.00009507358,0.00002417401,0.0002393984,0.0001165642,0.007549614,0.009382621,0.02150631,0.9602034,0.00003758546],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.0181041,0.001986402,0.3983406,0.004140202,0.001720025,0.0006793982,0.01843169,0.2199671,0.3366305],"genre_scores_gemma":[0.1388855,0.002314582,0.2682436,0.003666,0.0005123036,0.0007615981,0.04406777,0.0471832,0.4943657],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.126576,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2036787492","doi":"10.1145/2560796","title":"Sharing across Multiple MapReduce Jobs","year":2014,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Cloud computing; Merge (version control); Key (lock); Batch processing; Database; Distributed computing; Context (archaeology); Parallel computing; Operating system","authors":[{"name":"Tomasz Nykiel","is_ca":true},{"name":"Michalis Potamias","is_ca":false},{"name":"Chaitanya Mishra","is_ca":false},{"name":"George Kollios","is_ca":false},{"name":"Nick Koudas","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02870516225659378,"gpt":0.265099848590538,"spread":0.2363946863339443,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002924526,0.001391553,0.001699148,0.0007783233,0.002184439,0.002834382,0.005009312,0.0008587926,0.003421268],"category_scores_gemma":[0.00484811,0.0009013995,0.001837534,0.001324777,0.0009719788,0.003521536,0.004795258,0.001346896,0.001492345],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001321375,"about_ca_system_score_gemma":0.003646484,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006102161,"about_ca_topic_score_gemma":0.005765615,"domain_scores_codex":[0.9951001,0.0007445594,0.0003291403,0.001235631,0.001753264,0.0008372616],"domain_scores_gemma":[0.9959973,0.0004879944,0.0001613636,0.002076427,0.0008044888,0.0004723935],"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":[0.002340558,0.001680397,0.01352137,0.0007048374,0.0009671646,0.001496017,0.001948967,0.371493,0.08391492,0.03782113,0.0308384,0.4532731],"study_design_scores_gemma":[0.0001556436,0.000465717,0.004621177,0.00003093931,0.0001749249,0.0005285574,0.000927632,0.8758264,0.04122069,0.04751777,0.02837293,0.0001577155],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.186904,0.0006699495,0.7822013,0.001093278,0.0003519816,0.0009878745,0.0007079911,0.007930519,0.01915316],"genre_scores_gemma":[0.800671,0.0002136727,0.1880477,0.0002821024,0.0001599708,0.000378338,0.0012259,0.000698611,0.008322841],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006102161,"threshold_uncertainty_score":0.01546657,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4309617049","doi":"10.1145/3571281","title":"Efficiently Cleaning Structured Event Logs: A Graph Repair Approach","year":2022,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":5,"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":"Beijing National Research Center For Information Science And Technology; National Natural Science Foundation of China","keywords":"Computer science; Workflow; Event (particle physics); Pruning; Data mining; Graph; Profiling (computer programming); Information retrieval; Theoretical computer science; Database; Programming language","authors":[{"name":"Ruihong Huang","is_ca":false},{"name":"Jianmin Wang","is_ca":false},{"name":"Shaoxu Song","is_ca":false},{"name":"Xuemin Lin","is_ca":false},{"name":"Xiaochen Zhu","is_ca":false},{"name":"Jian Pei","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1167959570941646,"gpt":0.3589782289255208,"spread":0.2421822718313563,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002421997,0.001803952,0.001477332,0.003763601,0.001701707,0.001418174,0.003188455,0.001676271,0.001925949],"category_scores_gemma":[0.01445816,0.0008608645,0.002058743,0.003386246,0.001402858,0.003526282,0.002380341,0.002210292,0.00107113],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00105662,"about_ca_system_score_gemma":0.002783895,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008777151,"about_ca_topic_score_gemma":0.01302549,"domain_scores_codex":[0.9968985,0.0008022708,0.0002484814,0.0007986059,0.00104486,0.0002073245],"domain_scores_gemma":[0.9857898,0.006698771,0.001362722,0.003988374,0.001871797,0.0002885988],"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":[0.0004402264,0.0005072728,0.007456467,0.001112413,0.0002321999,0.001051079,0.001064945,0.2851672,0.02856133,0.01839515,0.02607188,0.6299399],"study_design_scores_gemma":[0.00006552428,0.0001579421,0.001329349,0.0000515255,0.000122218,0.0005202333,0.0004784882,0.9183147,0.0157549,0.0504079,0.01274408,0.0000532913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01294394,0.0003162917,0.978907,0.0004995849,0.00006408377,0.0002850334,0.0008466721,0.005584656,0.0005527845],"genre_scores_gemma":[0.1093605,0.0002941921,0.8830175,0.000243858,0.0000641066,0.0002193214,0.004326333,0.0006688405,0.001805258],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008777151,"threshold_uncertainty_score":0.01745212,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4410830018","doi":"10.1145/3736756","title":"Efficient Parallel Boolean Expression Matching","year":2025,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"Science and Technology Commission of Shanghai Municipality","keywords":"Computer science; Boolean expression; Expression (computer science); Matching (statistics); Regular expression; Theoretical computer science; Boolean function; Algorithm; Programming language; Mathematics","authors":[{"name":"Shuping Ji","is_ca":false},{"name":"Jianguo Yao","is_ca":false},{"name":"Wei Wang","is_ca":false},{"name":"Jun Wei","is_ca":false},{"name":"Hans‐Arno Jacobsen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01736739543533445,"gpt":0.2672524681699296,"spread":0.2498850727345952,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008107847,0.0008546436,0.001089018,0.001568643,0.0007826638,0.001553189,0.001823725,0.0006562033,0.005355327],"category_scores_gemma":[0.004309177,0.0003124761,0.001038006,0.003734235,0.0005519976,0.003865451,0.001958179,0.0007883497,0.001798014],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001040266,"about_ca_system_score_gemma":0.001673849,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002956373,"about_ca_topic_score_gemma":0.003557867,"domain_scores_codex":[0.9980222,0.0002004943,0.0002242284,0.0004302727,0.000861285,0.0002615867],"domain_scores_gemma":[0.9987102,0.0004213525,0.0001083591,0.000371687,0.0003411009,0.00004741739],"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":[0.0004497915,0.0003025185,0.002836668,0.0003580634,0.00007490815,0.0002971461,0.0002839493,0.03383618,0.05269723,0.03842133,0.02133904,0.8491033],"study_design_scores_gemma":[0.000136705,0.0002291005,0.00123843,0.00003033958,0.00009163177,0.0006357789,0.000250235,0.7879449,0.1001787,0.0777259,0.03148187,0.00005648344],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04689043,0.0006916011,0.9331259,0.000456828,0.0001043106,0.0003023249,0.001091501,0.008818247,0.008518941],"genre_scores_gemma":[0.3570755,0.0005356852,0.6255549,0.0005622661,0.0000849779,0.0003404102,0.005157297,0.0008580303,0.009830938],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005355327,"threshold_uncertainty_score":0.01791537,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W7117687674","doi":"10.1145/3786785","title":"Finding Smallest Witnesses for Conjunctive Queries","year":2025,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Conjunctive query; Rewriting; Property (philosophy); Logarithm; Class (philosophy); Time complexity; Polynomial; Constant (computer programming)","authors":[{"name":"Xiao Hu","is_ca":true},{"name":"Stavros Sintos","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03221601730685047,"gpt":0.2997566851307163,"spread":0.2675406678238658,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004470903,0.001585583,0.002071186,0.001006619,0.001561171,0.005324111,0.003887047,0.002000483,0.007540354],"category_scores_gemma":[0.03845896,0.00136824,0.002752077,0.002158978,0.002227003,0.01892512,0.004916113,0.004149882,0.001253639],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002093408,"about_ca_system_score_gemma":0.002685674,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002434527,"about_ca_topic_score_gemma":0.00315572,"domain_scores_codex":[0.9934914,0.0009590645,0.00067256,0.002473644,0.001664088,0.0007392475],"domain_scores_gemma":[0.9648377,0.02394155,0.001707859,0.007165348,0.00157179,0.0007756352],"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":[0.003180324,0.0008722564,0.01611427,0.003882586,0.0007940118,0.001327288,0.004512968,0.1840881,0.04634937,0.3443542,0.0237843,0.3707403],"study_design_scores_gemma":[0.0003026444,0.0002763353,0.001095411,0.0001688645,0.0003023351,0.0008534955,0.001087873,0.4746567,0.0258436,0.4855161,0.009813081,0.00008360172],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1494918,0.001327371,0.831946,0.003451832,0.000165035,0.0005278337,0.003259824,0.004653929,0.005176408],"genre_scores_gemma":[0.5633989,0.0008429206,0.4247015,0.0008864978,0.000207571,0.0003641118,0.005261706,0.0009452009,0.003391538],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007540354,"threshold_uncertainty_score":0.02522498,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4417496854","doi":"10.1145/3785662","title":"Semi-Oblivious Chase Termination for Linear Existential Rules and Beyond: An Experimental Analysis","year":2025,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; European Commission","keywords":"Existentialism; Chase; Set (abstract data type); Generalization; Linearization; Class (philosophy); Focus (optics)","authors":[{"name":"A.R. Alizad","is_ca":true},{"name":"Marco Calautti","is_ca":false},{"name":"Mostafa Milani","is_ca":true},{"name":"Andréas Pieris","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02356188656044739,"gpt":0.3092196439563378,"spread":0.2856577573958904,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007777838,0.001073943,0.001474575,0.000977125,0.001542414,0.002496689,0.003083466,0.002010335,0.01212659],"category_scores_gemma":[0.05304022,0.0005569426,0.001191926,0.001490299,0.002266707,0.005984626,0.002669798,0.004763674,0.002540305],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002001313,"about_ca_system_score_gemma":0.002697128,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002085644,"about_ca_topic_score_gemma":0.00220894,"domain_scores_codex":[0.9902869,0.002798481,0.0006972447,0.001501399,0.003489636,0.0012263],"domain_scores_gemma":[0.9106302,0.06457847,0.002658955,0.01580043,0.00505486,0.001277022],"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":[0.01077795,0.01289341,0.01791382,0.004457474,0.0007837142,0.001400803,0.002700428,0.274298,0.09989347,0.2083988,0.05543042,0.3110518],"study_design_scores_gemma":[0.0007167508,0.001941975,0.002363006,0.0001944777,0.0001634128,0.0004532638,0.0006705613,0.8380319,0.05399719,0.0893058,0.01205358,0.0001080845],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7630474,0.003720193,0.174807,0.001941542,0.0005833557,0.001167797,0.002874948,0.007634396,0.04422342],"genre_scores_gemma":[0.8824702,0.0008589999,0.1042774,0.0005012107,0.0001377792,0.0007728923,0.003499452,0.001381974,0.006100023],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01212659,"threshold_uncertainty_score":0.04113364,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4392139269","doi":"10.1145/3649133","title":"Sharing Queries with Nonequivalent User-defined Aggregate Functions","year":2024,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Rewriting; Computer science; Aggregate (composite); Class (philosophy); Query language; Function (biology); Materialized view; Extension (predicate logic); Theoretical computer science; Database query; Reuse; Programming language; Information retrieval; Artificial intelligence; View","authors":[{"name":"Chao Zhang","is_ca":true},{"name":"Farouk Toumani","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02668953795672124,"gpt":0.2596981473377908,"spread":0.2330086093810696,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009485825,0.0008451871,0.001596357,0.0008678156,0.0009711473,0.003817434,0.003108014,0.001229627,0.001545143],"category_scores_gemma":[0.02033746,0.0007015517,0.002041604,0.00113093,0.001913386,0.007458214,0.004144976,0.002048774,0.000435146],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001515778,"about_ca_system_score_gemma":0.001994753,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00223362,"about_ca_topic_score_gemma":0.001932036,"domain_scores_codex":[0.9877605,0.003636215,0.001595194,0.001668449,0.004347802,0.0009918321],"domain_scores_gemma":[0.9733683,0.01024735,0.001332092,0.01247506,0.002146706,0.0004304407],"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":[0.001809364,0.0006966187,0.01606016,0.000853774,0.0007167416,0.001896873,0.005172081,0.1124729,0.1115744,0.4276199,0.006922066,0.314205],"study_design_scores_gemma":[0.0001290096,0.0004955193,0.001988901,0.00009090883,0.0004307533,0.001056842,0.000837968,0.5715659,0.1834489,0.1980732,0.04166315,0.0002190436],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09895613,0.0003798429,0.892356,0.0003024645,0.0000692153,0.000157,0.0002306833,0.004731388,0.002817312],"genre_scores_gemma":[0.6947412,0.0002467934,0.2996871,0.0003042493,0.00007830546,0.0001305916,0.0006433126,0.001245978,0.002922463],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009485825,"threshold_uncertainty_score":0.05016643,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4415991463","doi":"10.1145/3774322","title":"Proof-of-Execution: Low-Latency Consensus via Speculative Execution","year":2025,"lang":"en","type":"article","venue":"ACM Transactions on Database Systems","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Protocol (science); Latency (audio); Key (lock); Distributed transaction; Transaction processing; Two-phase commit protocol; Database transaction","authors":[{"name":"Jelle Hellings","is_ca":true},{"name":"Suyash Gupta","is_ca":false},{"name":"Sajjad Rahnama","is_ca":false},{"name":"J. Bradley Chen","is_ca":false},{"name":"Celine Sana","is_ca":true},{"name":"Mohammad Sadoghi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01594775058131951,"gpt":0.2646067206446487,"spread":0.2486589700633292,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004388858,0.0007286519,0.0008236761,0.0007298107,0.001160488,0.001822151,0.002678034,0.001162548,0.003093684],"category_scores_gemma":[0.0156668,0.0006180533,0.0006072868,0.0007511814,0.002707676,0.004404407,0.004804255,0.002627427,0.0008341861],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008743202,"about_ca_system_score_gemma":0.002171788,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00122068,"about_ca_topic_score_gemma":0.001117736,"domain_scores_codex":[0.9963157,0.001371397,0.0002282057,0.0005237276,0.001122212,0.000438767],"domain_scores_gemma":[0.9874315,0.0060184,0.00110172,0.003744762,0.00115367,0.0005498887],"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":[0.0009623872,0.000244095,0.001787155,0.0004599832,0.000120204,0.001112812,0.001189296,0.2771728,0.05533247,0.5345024,0.005671299,0.121445],"study_design_scores_gemma":[0.0001094009,0.0001489599,0.0001160405,0.00003529103,0.00002685189,0.0001413643,0.00008567076,0.818656,0.01922858,0.1551455,0.006266492,0.00003994763],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02381319,0.000277902,0.9700969,0.0004942185,0.0001211278,0.0001395211,0.00003878909,0.001147195,0.003871088],"genre_scores_gemma":[0.7886961,0.0004597977,0.2049353,0.0002285628,0.0001106347,0.000319665,0.0001520032,0.0003053138,0.004792658],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004388858,"threshold_uncertainty_score":0.02321076,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}