{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":59,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":59,"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":"c4fafddebffc","filters":{"venue":"The VLDB Journal"}},"results":[{"id":"W2061240327","doi":"10.1007/s007780050006","title":"Distance-based outliers: algorithms and applications","year":2000,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":1196,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Outlier; Computer science; Anomaly detection; Simple (philosophy); Algorithm; Data mining; Curse of dimensionality; Identification (biology); Time complexity; Credit card fraud; Task (project management); Artificial intelligence; Machine learning; Pattern recognition (psychology); Credit card","authors":[{"name":"Edwin M. Knorr","is_ca":true},{"name":"Raymond T. Ng","is_ca":true},{"name":"Vladimir Tucakov","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.009606939193727806,"gpt":0.2379098253314463,"spread":0.2283028861377185,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002091071,0.001409823,0.00232931,0.003823178,0.0009703268,0.002238905,0.00326314,0.002537388,0.001534943],"category_scores_gemma":[0.01146754,0.0006284287,0.0007691493,0.007980758,0.001465115,0.003076773,0.002334451,0.002692567,0.00150229],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007716829,"about_ca_system_score_gemma":0.0007714336,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002487089,"about_ca_topic_score_gemma":0.001570219,"domain_scores_codex":[0.9974463,0.0005715375,0.0002004619,0.0004472636,0.001237778,0.0000966532],"domain_scores_gemma":[0.9940037,0.002686643,0.0006608246,0.0006870644,0.001787614,0.0001741915],"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":[0.0003002487,0.0001622768,0.003354048,0.0002591762,0.000102541,0.0001267865,0.0001583102,0.103678,0.005953399,0.02607104,0.00715067,0.8526835],"study_design_scores_gemma":[0.00003844649,0.00007562499,0.0008128274,0.00003183067,0.00004128646,0.0003568474,0.00008732695,0.9452694,0.006367022,0.04053836,0.006332429,0.00004853649],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004006234,0.001180884,0.9932889,0.0001715977,0.00007993594,0.00001994094,0.00006131666,0.0008832067,0.0003079462],"genre_scores_gemma":[0.1396761,0.002535899,0.8541574,0.00009179443,0.0003632563,0.0001242194,0.0005627967,0.0002794511,0.002209023],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003823178,"threshold_uncertainty_score":0.01105875,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2110557355","doi":"10.1007/s007780100049","title":"Approximate query processing using wavelets","year":2001,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":477,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Bell (Canada)","funders":"","keywords":"Computer science; Wavelet; Query optimization; Data mining; Tuple; Histogram; Algorithm; Artificial intelligence; Mathematics; Image (mathematics)","authors":[{"name":"Kaushik Chakrabarti","is_ca":false},{"name":"Minos Garofalakis","is_ca":true},{"name":"Rajeev Rastogi","is_ca":true},{"name":"Kyuseok Shim","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0412518917975479,"gpt":0.2701677799750619,"spread":0.228915888177514,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002698724,0.0007566524,0.00243333,0.001933353,0.0008061604,0.00291935,0.001408989,0.00108481,0.002794555],"category_scores_gemma":[0.01294768,0.0007078713,0.001120851,0.004012499,0.001021991,0.004041892,0.002198686,0.002001568,0.001187813],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008298015,"about_ca_system_score_gemma":0.001185185,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002402531,"about_ca_topic_score_gemma":0.001779978,"domain_scores_codex":[0.9965302,0.00091922,0.0002913128,0.0004194638,0.001517447,0.0003223026],"domain_scores_gemma":[0.9946956,0.002182281,0.0002738486,0.00174522,0.0009329576,0.0001701513],"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":[0.001878235,0.0002778685,0.003008145,0.0002992157,0.0002790189,0.0002583879,0.0003515654,0.2500666,0.03441801,0.1139692,0.0125973,0.5825965],"study_design_scores_gemma":[0.00002869128,0.00006702182,0.0001983092,0.000007317522,0.00002080276,0.00007342384,0.00004600806,0.9685407,0.003322681,0.02622098,0.001462427,0.00001156798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02235579,0.00068946,0.974838,0.0002732595,0.0001211508,0.00002979667,0.00009092833,0.0007114651,0.0008901074],"genre_scores_gemma":[0.4244093,0.001185465,0.56917,0.0002105767,0.0003757897,0.0001452152,0.0008243181,0.0002669368,0.003412316],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00291935,"threshold_uncertainty_score":0.01427239,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2040374677","doi":"10.1007/s00778-002-0081-x","title":"TIMBER: A native XML database","year":2002,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":408,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia; University of British Columbia Hospital","funders":"","keywords":"Computer science; XPath; XML database; Streaming XML; XML validation; Efficient XML Interchange; XML Schema Editor; XML Encryption; Database; Query optimization; XML Schema (W3C); Information retrieval; XML; XML Signature; Document Structure Description; Rewriting; Relational database; Programming language; World Wide Web","authors":[{"name":"H. V. Jagadish","is_ca":false},{"name":"Shurug Al-Khalifa","is_ca":false},{"name":"Adriane Chapman","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Andrew Nierman","is_ca":false},{"name":"S. Paparizos","is_ca":false},{"name":"Jignesh M. Patel","is_ca":false},{"name":"Divesh Srivastava","is_ca":false},{"name":"Nuwee Wiwatwattana","is_ca":false},{"name":"Yuqing Wu","is_ca":false},{"name":"Cong Yu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02968092915324465,"gpt":0.2434911894461427,"spread":0.2138102602928981,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001711137,0.0009152287,0.0008096745,0.00168004,0.0008198504,0.002951851,0.003335686,0.00111551,0.04198299],"category_scores_gemma":[0.004150643,0.0009285269,0.0006353547,0.001757303,0.0006781311,0.005189545,0.002639361,0.001591249,0.02676995],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004786318,"about_ca_system_score_gemma":0.001339918,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002897083,"about_ca_topic_score_gemma":0.003936544,"domain_scores_codex":[0.9989785,0.0001143501,0.0001414172,0.0001862905,0.0004653207,0.0001142003],"domain_scores_gemma":[0.9983242,0.0003177951,0.00007606765,0.0006834998,0.0003966167,0.0002017934],"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":[0.001367333,0.0004146742,0.002785303,0.0007249621,0.0001369118,0.000646078,0.0005836682,0.002747043,0.02806381,0.03597048,0.629412,0.2971478],"study_design_scores_gemma":[0.0007072585,0.0002653515,0.003005572,0.0001876449,0.0001959872,0.001444672,0.0004483227,0.04910322,0.04197178,0.04970949,0.8527276,0.0002331171],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"software","genre_gemma":"methods","genre_scores_codex":[0.0154738,0.001189172,0.3010935,0.0009803518,0.0007352425,0.0005334383,0.02184816,0.6003305,0.05781583],"genre_scores_gemma":[0.1968368,0.002256912,0.391914,0.002900959,0.0005460267,0.001064974,0.1400591,0.1067669,0.1576543],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.04198299,"threshold_uncertainty_score":0.1404471,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2143682328","doi":"10.1007/s00778-004-0128-2","title":"Supporting top-k join queries in relational databases","year":2004,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":387,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Joins; Computer science; Join (topology); Sort-merge join; Query optimization; Ranking (information retrieval); Hash join; Database; Relational database; Heuristic; Query language; Sargable; Information retrieval; Rank (graph theory); Heuristics; Relational algebra; Query expansion; Theoretical computer science; Web search query; Search engine; Programming language; Mathematics; Artificial intelligence","authors":[{"name":"Ihab F. Ilyas","is_ca":true},{"name":"WalidG. Aref","is_ca":false},{"name":"AhmedK. Elmagarmid","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04003804845750331,"gpt":0.2958259358899701,"spread":0.2557878874324668,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005228457,0.001087648,0.002600929,0.001444186,0.00243184,0.008435749,0.005038479,0.002322149,0.005461701],"category_scores_gemma":[0.024543,0.001496007,0.001209621,0.003480641,0.002048505,0.01740723,0.006402438,0.002378959,0.003628027],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009942695,"about_ca_system_score_gemma":0.002183656,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003360765,"about_ca_topic_score_gemma":0.004614528,"domain_scores_codex":[0.9910232,0.001505855,0.001461646,0.001354981,0.003473944,0.001180373],"domain_scores_gemma":[0.9746523,0.01198386,0.00128786,0.008777367,0.002372137,0.0009265894],"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":[0.009171318,0.001244991,0.03337821,0.002790584,0.0007308449,0.00222956,0.004358225,0.08303259,0.07917921,0.1042169,0.05763362,0.622034],"study_design_scores_gemma":[0.0005810113,0.0006774622,0.003055695,0.0001885187,0.0005458107,0.002211432,0.001933504,0.6167038,0.0685297,0.2791406,0.02617003,0.0002623973],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2319163,0.003700186,0.7254238,0.00212552,0.000427445,0.0002326098,0.002511929,0.02190468,0.01175753],"genre_scores_gemma":[0.7734467,0.001022039,0.2171907,0.0003451017,0.0002487903,0.0000756262,0.002636575,0.001055536,0.00397895],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008435749,"threshold_uncertainty_score":0.02765107,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W760598031","doi":"10.1007/s00778-015-0389-y","title":"Profiling relational data: a survey","year":2015,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":273,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Profiling (computer programming); Metadata; Computer science; Data mining; Data type; Information retrieval; Data science; World Wide Web","authors":[{"name":"Ziawasch Abedjan","is_ca":false},{"name":"Lukasz Golab","is_ca":true},{"name":"Felix Naumann","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.8844173631534311,"gpt":0.5434835487839118,"spread":0.3409338143695193,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01869769,0.0006774935,0.001507291,0.01501409,0.0009242385,0.007538349,0.002744373,0.001273704,0.001958959],"category_scores_gemma":[0.07480663,0.0007422493,0.001051805,0.0222602,0.001718179,0.01735284,0.003062729,0.001598523,0.001021985],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001464438,"about_ca_system_score_gemma":0.002510754,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002434847,"about_ca_topic_score_gemma":0.003064253,"domain_scores_codex":[0.98079,0.003948452,0.002594439,0.002924362,0.009087564,0.0006552319],"domain_scores_gemma":[0.8604422,0.09986163,0.007603403,0.01009019,0.01982508,0.00217755],"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":[0.0001678369,0.0001770685,0.0908149,0.00369295,0.0001898058,0.00007622846,0.001507676,0.001936789,0.00225872,0.01415405,0.01142764,0.8735963],"study_design_scores_gemma":[0.000067401,0.0007884558,0.1373759,0.009211984,0.0006429999,0.003835496,0.01551866,0.03142736,0.01908045,0.06174427,0.7199531,0.0003539018],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.204822,0.4718415,0.2501662,0.02193191,0.0006169153,0.0005095127,0.0106903,0.001940897,0.03748073],"genre_scores_gemma":[0.4661339,0.3631345,0.1421929,0.004883834,0.0009596304,0.0004179132,0.01806351,0.0006093284,0.003604455],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.01869769,"threshold_uncertainty_score":0.09888405,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2085472312","doi":"10.1007/s00778-013-0344-8","title":"Correlated network data publication via differential privacy","year":2013,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":214,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University; McGill University","funders":"","keywords":"Differential privacy; Computer science; Adjacency matrix; Data publishing; Data mining; Privacy software; Information privacy; Enhanced Data Rates for GSM Evolution; Node (physics); Theoretical computer science; Data science; Publishing; Computer security; Artificial intelligence; Graph","authors":[{"name":"Rui Chen","is_ca":false},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Philip S. Yu","is_ca":false},{"name":"Bipin C. Desai","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04447157467703367,"gpt":0.2691169814537689,"spread":0.2246454067767352,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01296624,0.0009974707,0.002950707,0.002667341,0.002177546,0.007111678,0.003374378,0.002181064,0.004190474],"category_scores_gemma":[0.06631979,0.001388941,0.001714032,0.007127909,0.003798183,0.01182233,0.009424089,0.004169993,0.001736036],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002623521,"about_ca_system_score_gemma":0.004045967,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005916122,"about_ca_topic_score_gemma":0.0007321251,"domain_scores_codex":[0.9777625,0.009048078,0.001472712,0.002955856,0.007153261,0.001607482],"domain_scores_gemma":[0.9024111,0.0472557,0.004624547,0.04043598,0.004074038,0.001198631],"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":[0.00149665,0.0002935768,0.00511448,0.000432611,0.000295366,0.0009070004,0.0006753049,0.111295,0.006404969,0.6849306,0.01203865,0.1761159],"study_design_scores_gemma":[0.0001182737,0.0001228873,0.0004719103,0.00007208764,0.0001263477,0.0008036238,0.00008881495,0.4748773,0.01041831,0.5046828,0.008164916,0.00005266606],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02691211,0.0009011136,0.9624556,0.002364346,0.0002278567,0.0001161816,0.0005190169,0.0006778636,0.005825839],"genre_scores_gemma":[0.8601065,0.001632173,0.125281,0.0008728536,0.0007545765,0.0002846365,0.001077702,0.0002554921,0.00973513],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01296624,"threshold_uncertainty_score":0.06857288,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2121041488","doi":"10.1007/s00778-013-0337-7","title":"gStore: a graph-based SPARQL query engine","year":2013,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":202,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"SPARQL; Computer science; Named graph; RDF; Scalability; RDF query language; Information retrieval; Web search query; Query optimization; Graph; Query expansion; RDF Schema; Web query classification; Theoretical computer science; Semantic Web; Database; Search engine","authors":[{"name":"Lei Zou","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true},{"name":"Lei Chen","is_ca":false},{"name":"Xuchuan Shen","is_ca":false},{"name":"Ruizhe Huang","is_ca":false},{"name":"Dongyan Zhao","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01207445201333954,"gpt":0.2032755897971928,"spread":0.1912011377838533,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001611193,0.00136838,0.00151239,0.002483015,0.0006595911,0.002494524,0.003488198,0.001189901,0.01370658],"category_scores_gemma":[0.004040848,0.001053008,0.001654714,0.002876087,0.0007681045,0.004840537,0.002809994,0.001580773,0.006480323],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009718442,"about_ca_system_score_gemma":0.001443759,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008267613,"about_ca_topic_score_gemma":0.0126606,"domain_scores_codex":[0.9982718,0.0002316264,0.0002052728,0.0003370269,0.0008280475,0.0001262023],"domain_scores_gemma":[0.9983897,0.0005805668,0.00009212132,0.0005095821,0.0003362913,0.00009170968],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002156789,0.0006718138,0.005143229,0.003089244,0.001066075,0.0007214469,0.0007467208,0.03544534,0.0642688,0.07019129,0.3737282,0.442771],"study_design_scores_gemma":[0.0007456649,0.0003542758,0.002708531,0.0002479525,0.0005182087,0.00094188,0.0004834453,0.5010182,0.07639446,0.1106202,0.3055898,0.0003772382],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01069539,0.0007946964,0.6307598,0.0005589408,0.0002132722,0.0004487263,0.01935323,0.3291728,0.008003093],"genre_scores_gemma":[0.1596801,0.001690638,0.6629913,0.002151045,0.0001481111,0.0006794217,0.1133544,0.03946584,0.01983907],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01370658,"threshold_uncertainty_score":0.04585302,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2126431245","doi":"10.1007/s00778-006-0029-7","title":"Algorithms and analyses for maximal vector computation","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":177,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Skyline; Curse of dimensionality; Computation; Field (mathematics); Computer science; Running time; Divide and conquer algorithms; Algorithm; Mathematics; Discrete mathematics; Theoretical computer science; Combinatorics; Artificial intelligence; Data mining; Pure mathematics","authors":[{"name":"Parke Godfrey","is_ca":true},{"name":"J. Ryan Shipley","is_ca":false},{"name":"Jarek Gryz","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04801050073090831,"gpt":0.3160797680561167,"spread":0.2680692673252084,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005817893,0.002498648,0.002153084,0.005575067,0.002533806,0.00698099,0.005808397,0.002048922,0.01387799],"category_scores_gemma":[0.03881434,0.001335485,0.002844607,0.007361695,0.005171902,0.01723073,0.005553711,0.006791322,0.004388889],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003480046,"about_ca_system_score_gemma":0.003017663,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002916614,"about_ca_topic_score_gemma":0.002988583,"domain_scores_codex":[0.9932895,0.002234266,0.0004683542,0.00100475,0.002230083,0.0007731044],"domain_scores_gemma":[0.9749116,0.01697567,0.001067221,0.004094638,0.002336722,0.0006141624],"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":[0.0003137983,0.0001582691,0.0009700437,0.0003252389,0.00008653073,0.00004961489,0.000225085,0.02553238,0.0008970145,0.8529319,0.01188647,0.1066236],"study_design_scores_gemma":[0.00001939128,0.00001893421,0.0001651515,0.00004202341,0.00003735662,0.00003040892,0.00003512541,0.09281605,0.000847549,0.9024708,0.003497781,0.00001929792],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006338958,0.001954228,0.9792014,0.000897912,0.0002151101,0.00007990056,0.0003003806,0.001523159,0.009488952],"genre_scores_gemma":[0.3147154,0.003627452,0.6537578,0.001149893,0.002264948,0.00067938,0.002153363,0.001980165,0.01967175],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01387799,"threshold_uncertainty_score":0.04642648,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2064580935","doi":"10.1007/s00778-004-0123-7","title":"Querying XML streams","year":2004,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":126,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; XQuery; XML; Streaming XML; Tuple; XML database; Path (computing); Query plan; Programming language; Database; Efficient XML Interchange; XML Encryption; Concurrency; Information retrieval; Operating system; Search engine; Web search query; Sargable","authors":[{"name":"Vanja Josifovski","is_ca":false},{"name":"Marcus Fontoura","is_ca":false},{"name":"Attila Barta","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01035243184366573,"gpt":0.2293669318428719,"spread":0.2190144999992061,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003633282,0.00120612,0.001591583,0.002797914,0.001137165,0.006531637,0.002240458,0.001973054,0.01020728],"category_scores_gemma":[0.01314679,0.0009185956,0.00128776,0.003516791,0.00111768,0.009381784,0.00329694,0.00189847,0.003597287],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001204164,"about_ca_system_score_gemma":0.00108284,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002557571,"about_ca_topic_score_gemma":0.002188037,"domain_scores_codex":[0.9962938,0.0006468805,0.0004999244,0.0006088804,0.001751194,0.0001992657],"domain_scores_gemma":[0.9954517,0.002002759,0.0001878339,0.001387336,0.0007694774,0.0002008518],"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":[0.002492207,0.0005753894,0.01292798,0.00107042,0.0003826481,0.001322816,0.001637983,0.03220011,0.02944381,0.2324628,0.09486795,0.5906159],"study_design_scores_gemma":[0.0003122075,0.0002352582,0.001594074,0.0002420469,0.0002849004,0.001459385,0.00098274,0.5070152,0.0790376,0.2412768,0.1674296,0.0001301757],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05630952,0.001907133,0.867203,0.00254368,0.0007876902,0.0004269332,0.007390821,0.04423206,0.0191991],"genre_scores_gemma":[0.4196128,0.002820706,0.5065785,0.001408717,0.0007369439,0.0004413061,0.03123961,0.005327012,0.03183442],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01020728,"threshold_uncertainty_score":0.03414673,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2068157134","doi":"10.1007/s00778-002-0076-7","title":"Tree pattern query minimization","year":2002,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":109,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Minification; Computer science; Tree (set theory); XML; Reduction (mathematics); Pattern matching; Algorithm; Query optimization; Data mining; Theoretical computer science; Mathematics; Artificial intelligence","authors":[{"name":"Sihem Amer-Yahia","is_ca":false},{"name":"Sung-Eun Cho","is_ca":false},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Divesh Srivastava","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02434631267444179,"gpt":0.2217052823758772,"spread":0.1973589697014354,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001356983,0.0009073167,0.001592495,0.001408145,0.0008805991,0.002378471,0.002333503,0.001197262,0.0172053],"category_scores_gemma":[0.005846284,0.0005049155,0.001429913,0.002674246,0.000725089,0.003728926,0.002431853,0.002081223,0.004724224],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00098216,"about_ca_system_score_gemma":0.001925922,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001544195,"about_ca_topic_score_gemma":0.002778013,"domain_scores_codex":[0.9969784,0.0005933564,0.0001962874,0.0006492062,0.001195886,0.0003868044],"domain_scores_gemma":[0.9970939,0.001051161,0.0001341071,0.001067592,0.0005310765,0.000122149],"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":[0.0009952836,0.0006250551,0.002670848,0.0005871067,0.0002170205,0.0002241971,0.0002308863,0.03899916,0.01851721,0.1441998,0.0741691,0.7185643],"study_design_scores_gemma":[0.000211138,0.0003842603,0.001466717,0.00005030237,0.0002626363,0.001119764,0.0002900104,0.6463464,0.03733354,0.2834758,0.02901756,0.00004182628],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04524384,0.0008133336,0.9194626,0.002359509,0.0001743238,0.0004324718,0.00209992,0.004549313,0.02486483],"genre_scores_gemma":[0.3250385,0.0006863857,0.6233127,0.001096584,0.0003159645,0.000467711,0.006900965,0.001764973,0.04041621],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0172053,"threshold_uncertainty_score":0.05755746,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1988668397","doi":"10.1007/s00778-007-0058-x","title":"Temporal XML: modeling, indexing, and query processing","year":2007,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; XPath; Information retrieval; Search engine indexing; Query language; XML database; XML validation; Streaming XML; XML; Temporal database; Document Structure Description; Data mining; World Wide Web","authors":[{"name":"Flavio Rizzolo","is_ca":true},{"name":"Alejandro Vaisman","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01980670494016319,"gpt":0.2688788118537077,"spread":0.2490721069135445,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006209496,0.0008504153,0.001146559,0.003331033,0.0008235184,0.007952701,0.00352269,0.00188221,0.003000307],"category_scores_gemma":[0.01482789,0.001070613,0.001386147,0.007056882,0.001781276,0.01204973,0.001391137,0.001603553,0.001307655],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002044306,"about_ca_system_score_gemma":0.002257144,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01180383,"about_ca_topic_score_gemma":0.007591634,"domain_scores_codex":[0.995899,0.001521052,0.0006113211,0.0005381175,0.001239528,0.0001909253],"domain_scores_gemma":[0.9939814,0.003388081,0.0005122092,0.001247461,0.0006882609,0.0001824809],"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":[0.0003808387,0.0002804229,0.005675923,0.0008557687,0.0001243745,0.000439146,0.00108017,0.1178453,0.00411377,0.5937322,0.02043135,0.2550408],"study_design_scores_gemma":[0.00004540715,0.00009833643,0.0009664568,0.0001606011,0.0001243215,0.0006240569,0.0003332647,0.6637654,0.004334326,0.2849685,0.04450912,0.00007027278],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01700853,0.007450273,0.9620278,0.002255805,0.0002162901,0.0001484673,0.001798737,0.003751532,0.005342662],"genre_scores_gemma":[0.3203615,0.01232639,0.6507199,0.0005374959,0.0007260865,0.0004096286,0.006692308,0.0008562956,0.00737043],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01180383,"threshold_uncertainty_score":0.03283936,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2755658662","doi":"10.1007/s00778-019-00548-x","title":"The ubiquity of large graphs and surprising challenges of graph processing: extended survey","year":2019,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":84,"is_retracted":false,"has_abstract":false,"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":"Suite; Graph; Scalability; Graph drawing; Visualization; Call graph; Data visualization","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02115508178165599,"gpt":0.263779211117652,"spread":0.242624129335996,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003521405,0.0008484069,0.001525213,0.003811397,0.001094366,0.004307706,0.002686842,0.002150926,0.003870468],"category_scores_gemma":[0.0199426,0.001033507,0.0007591307,0.007085608,0.003670955,0.01496999,0.002497525,0.002925726,0.00127783],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001587449,"about_ca_system_score_gemma":0.001459882,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002360604,"about_ca_topic_score_gemma":0.002826448,"domain_scores_codex":[0.9978035,0.0005830643,0.0001008938,0.0007293304,0.0006286118,0.0001545708],"domain_scores_gemma":[0.9681472,0.02594262,0.001152995,0.002463004,0.001640864,0.0006534325],"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":[0.000164054,0.0001794589,0.01260582,0.004672566,0.0001931829,0.000267284,0.0004383849,0.01874274,0.001927419,0.2863099,0.04613047,0.6283687],"study_design_scores_gemma":[0.00002156128,0.0000994178,0.006019396,0.0008891069,0.00009043835,0.001478394,0.0007428388,0.04954101,0.001610467,0.7115039,0.2279352,0.00006834722],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.05113453,0.5899597,0.2586038,0.05807485,0.001236712,0.0001152062,0.001843052,0.00136651,0.03766572],"genre_scores_gemma":[0.3021656,0.5546473,0.1201978,0.008320748,0.006998458,0.0001373027,0.00286681,0.0005662181,0.004099707],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.004307706,"threshold_uncertainty_score":0.01862317,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2012848182","doi":"10.1007/s00778-009-0176-8","title":"Supporting ranking queries on uncertain and incomplete data","year":2010,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":78,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Ranking (information retrieval); Computer science; Data mining; Uncertain data; Set (abstract data type); Rank (graph theory); Probabilistic logic; Ranking SVM; Information retrieval; Semantics (computer science); Learning to rank; Artificial intelligence; Mathematics","authors":[{"name":"Mohamed A. Soliman","is_ca":true},{"name":"Ihab F. Ilyas","is_ca":true},{"name":"Shalev Ben-David","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04986115690612704,"gpt":0.3238459435933017,"spread":0.2739847866871746,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008457639,0.001692377,0.004651676,0.003070237,0.002000741,0.008426222,0.004591455,0.00370967,0.004689343],"category_scores_gemma":[0.05408572,0.001375184,0.001737949,0.007149325,0.001494348,0.01735496,0.004027179,0.00329189,0.001610734],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001216387,"about_ca_system_score_gemma":0.002501567,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004796967,"about_ca_topic_score_gemma":0.007261544,"domain_scores_codex":[0.9873517,0.003325354,0.001359166,0.00170742,0.005217439,0.001038915],"domain_scores_gemma":[0.9317869,0.05248009,0.002535132,0.0075056,0.004570733,0.001121587],"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":[0.003391317,0.001092153,0.01310177,0.002725439,0.0006572658,0.002605977,0.001616614,0.3940114,0.02069813,0.07969754,0.05031104,0.4300914],"study_design_scores_gemma":[0.00009664708,0.0001911753,0.00050888,0.00005246739,0.0001025157,0.0004871429,0.0004358298,0.8932379,0.005257257,0.09556155,0.004028921,0.00003970831],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1685836,0.003097147,0.8067501,0.004632933,0.0004892536,0.0003647204,0.005317983,0.005529271,0.005234917],"genre_scores_gemma":[0.6873715,0.001315837,0.2971462,0.0006839785,0.0009049015,0.0001459412,0.008768324,0.0005362898,0.00312701],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008457639,"threshold_uncertainty_score":0.04472882,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1977320861","doi":"10.1007/s00778-004-0136-2","title":"Preserving mapping consistency under schema changes","year":2004,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Schema (genetic algorithms); Rewriting; Conceptual schema; Theoretical computer science; Semantics (computer science); Information retrieval; Programming language","authors":[{"name":"Yannis Velegrakis","is_ca":true},{"name":"Renée J. Miller","is_ca":true},{"name":"Lucian Popa","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05595234486683632,"gpt":0.2594263878383082,"spread":0.2034740429714719,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01300161,0.001129487,0.002590392,0.002770867,0.00226968,0.008142119,0.004986068,0.003247301,0.003316373],"category_scores_gemma":[0.06665141,0.002409852,0.002274935,0.006554347,0.00401183,0.02130204,0.008825484,0.006619938,0.001315991],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001195653,"about_ca_system_score_gemma":0.003784974,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002574996,"about_ca_topic_score_gemma":0.001984447,"domain_scores_codex":[0.9795458,0.007216014,0.002198812,0.003905329,0.005715683,0.001418334],"domain_scores_gemma":[0.896283,0.03124994,0.004375931,0.05926326,0.007802707,0.001025065],"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":[0.002464077,0.0006779814,0.01408276,0.0009997725,0.001072475,0.002562204,0.003967999,0.08508351,0.0377629,0.4106396,0.01397659,0.4267102],"study_design_scores_gemma":[0.0002716072,0.0002107889,0.00206447,0.000137482,0.0009361424,0.001425374,0.00123699,0.1641964,0.06518991,0.7464266,0.01776903,0.0001352009],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.133088,0.0006537027,0.8490754,0.002090947,0.0004448183,0.0002323964,0.001040967,0.005222672,0.008151084],"genre_scores_gemma":[0.74642,0.0007839896,0.2396941,0.001160537,0.0004563306,0.0002161824,0.002587638,0.002230331,0.006451026],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01300161,"threshold_uncertainty_score":0.06875986,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2141724398","doi":"10.1007/s00778-005-0172-6","title":"Query optimization in XML structured-document databases","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":68,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; XPath; Query optimization; XML validation; Streaming XML; XML database; Information retrieval; Efficient XML Interchange; Query language; XML Schema Editor; XML; Document Structure Description; Query expansion; XML Schema (W3C); XML Encryption; Database; World Wide Web","authors":[{"name":"Dunren Che","is_ca":false},{"name":"Karl Aberer","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01090803058863072,"gpt":0.2453907915984678,"spread":0.2344827610098371,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00575842,0.00077326,0.002290678,0.001409252,0.0008271624,0.003744515,0.002465655,0.001175058,0.00256719],"category_scores_gemma":[0.01187872,0.0009052192,0.001091062,0.003707587,0.00117574,0.004623153,0.001860078,0.00145026,0.0005994705],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001502751,"about_ca_system_score_gemma":0.001543937,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003985494,"about_ca_topic_score_gemma":0.003770109,"domain_scores_codex":[0.9920763,0.002825018,0.0008538063,0.0008178975,0.002856656,0.0005704078],"domain_scores_gemma":[0.9919504,0.005921576,0.0003999489,0.0007658566,0.0007672529,0.0001950852],"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":[0.002366523,0.0007564998,0.004721431,0.001227665,0.0003110024,0.0006014063,0.0006835173,0.4787256,0.01656339,0.09702289,0.01998481,0.3770352],"study_design_scores_gemma":[0.0001190887,0.0001382344,0.0004603981,0.00002353141,0.00007088928,0.0001895924,0.0001889401,0.9415951,0.007931672,0.04611,0.003148252,0.00002428794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07426152,0.003200453,0.9136544,0.00139615,0.00008989628,0.0001939711,0.0005454906,0.002866185,0.003791858],"genre_scores_gemma":[0.5193705,0.001610844,0.4709459,0.0004394113,0.000209969,0.0002340623,0.00173081,0.0006280487,0.004830433],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00575842,"threshold_uncertainty_score":0.0304538,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2100257623","doi":"10.1007/s00778-010-0214-6","title":"Anonymity meets game theory: secure data integration with malicious participants","year":2010,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":66,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Anonymity; Scalability; Private information retrieval; Adversary; Data sharing; k-anonymity; Computer security; Data integration; Information privacy; Data anonymization; Partition (number theory); Table (database); Set (abstract data type); Data mining; Database","authors":[{"name":"Noman Mohammed","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05734070335716644,"gpt":0.3061374292867121,"spread":0.2487967259295457,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01491403,0.001296086,0.002297611,0.001271691,0.003275107,0.009136052,0.004351279,0.004044156,0.002940186],"category_scores_gemma":[0.03760947,0.001476554,0.002372873,0.001579639,0.006647523,0.01527542,0.01026618,0.006531689,0.0009796727],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002574654,"about_ca_system_score_gemma":0.004648136,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001841172,"about_ca_topic_score_gemma":0.001336729,"domain_scores_codex":[0.9784581,0.01237374,0.0008469552,0.002129656,0.004565814,0.001625755],"domain_scores_gemma":[0.9592054,0.02778701,0.001784456,0.007209888,0.002639523,0.001373638],"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":[0.000441616,0.0001347645,0.000632867,0.0001030228,0.0001203872,0.000292289,0.0006694301,0.05942507,0.002445117,0.9135897,0.002109691,0.02003607],"study_design_scores_gemma":[0.00006408378,0.00005083964,0.00004836168,0.00001519662,0.00004180093,0.00009576361,0.00009708409,0.3713385,0.001685601,0.6244267,0.002107316,0.00002869491],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01799138,0.00009473358,0.9756848,0.001178106,0.00007087906,0.0001518152,0.00005061882,0.0002269584,0.004550739],"genre_scores_gemma":[0.7691895,0.0002368759,0.2226928,0.0004722784,0.0001540521,0.0004389903,0.000125373,0.0001624691,0.006527773],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01491403,"threshold_uncertainty_score":0.07887381,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2092324613","doi":"10.1007/s007780200062","title":"Query processing techniques for arrays","year":2002,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Variety (cybernetics); Class (philosophy); Rewriting; Query plan; Function (biology); Query optimization; Data structure; Space (punctuation); Theoretical computer science; Programming language; Database; Information retrieval; Sargable; Artificial intelligence; Search engine; Web search query","authors":[{"name":"A. Marathe","is_ca":true},{"name":"Kenneth Salem","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03430227086175861,"gpt":0.2687840089182429,"spread":0.2344817380564843,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002466884,0.00136827,0.001808624,0.002930044,0.00203488,0.005329168,0.00290099,0.001194893,0.01153482],"category_scores_gemma":[0.009459683,0.001204903,0.00199368,0.0069329,0.001754737,0.009747375,0.003306026,0.003454196,0.006127321],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001095645,"about_ca_system_score_gemma":0.001158955,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002498285,"about_ca_topic_score_gemma":0.002115507,"domain_scores_codex":[0.9954388,0.0008560827,0.0005256697,0.0007204802,0.002023531,0.0004354947],"domain_scores_gemma":[0.9947647,0.002037353,0.0002332329,0.001919211,0.0009383738,0.000107095],"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":[0.0004838608,0.0001273707,0.0008632146,0.0009100059,0.0001927181,0.0002308424,0.00100221,0.01067104,0.01552306,0.3672738,0.04947043,0.5532514],"study_design_scores_gemma":[0.0001416794,0.000199843,0.0005265097,0.0001937979,0.0003040991,0.0008597743,0.0004622342,0.1335589,0.02918915,0.6769528,0.1575005,0.0001106998],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.003336313,0.003060627,0.9804765,0.0006342405,0.0002307952,0.00009245538,0.0003377556,0.004758921,0.007072314],"genre_scores_gemma":[0.1046228,0.004929546,0.8683001,0.0007808017,0.0009335958,0.0003516893,0.00190713,0.001864398,0.01630994],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01153482,"threshold_uncertainty_score":0.03858787,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1995541812","doi":"10.1007/s00778-009-0161-2","title":"Creating probabilistic databases from duplicated data","year":2009,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":61,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Probabilistic logic; Data mining; Scalability; Data deduplication; Task (project management); Cluster analysis; Database; String (physics); Probabilistic database; Information retrieval; Machine learning; Artificial intelligence; Database theory; Relational database","authors":[{"name":"Oktie Hassanzadeh","is_ca":true},{"name":"Renée J. Miller","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.4798807441543463,"gpt":0.4770797275385156,"spread":0.002801016615830709,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011814,0.001015899,0.002131192,0.005695032,0.001478626,0.006886034,0.006783619,0.002236261,0.003579563],"category_scores_gemma":[0.07379757,0.002510847,0.002979721,0.00944797,0.001422532,0.0117748,0.007296313,0.002279842,0.001551984],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001218541,"about_ca_system_score_gemma":0.003040248,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002422068,"about_ca_topic_score_gemma":0.003470686,"domain_scores_codex":[0.9881392,0.002553105,0.001274037,0.002108176,0.005480946,0.0004445137],"domain_scores_gemma":[0.937184,0.0342419,0.002999665,0.01905036,0.005741648,0.0007822549],"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":[0.001034216,0.0004380929,0.01609243,0.001047934,0.0007365667,0.002295628,0.001905103,0.1630905,0.01249567,0.07947756,0.02216232,0.6992239],"study_design_scores_gemma":[0.0001079395,0.0001601663,0.001609265,0.0001122813,0.0003219376,0.001748828,0.0005635697,0.8275191,0.02251518,0.1282526,0.01696126,0.0001278225],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0336339,0.0003936307,0.9544981,0.0005678157,0.00009633327,0.0002014845,0.001916215,0.007510467,0.001182081],"genre_scores_gemma":[0.1832422,0.000406788,0.8093007,0.000148443,0.00009252705,0.0002299795,0.004675475,0.0009120769,0.0009918391],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.011814,"threshold_uncertainty_score":0.06247914,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2156892372","doi":"10.1007/s00778-009-0162-1","title":"Threshold-based probabilistic top-k dominating queries","year":2009,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Randomized algorithm; Scalability; Probabilistic logic; Computer science; Computation; Uncertain data; Set (abstract data type); Efficient algorithm; Dominating set; Algorithm; Theoretical computer science; Data mining; Artificial intelligence; Database","authors":[{"name":"Wenjie Zhang","is_ca":false},{"name":"Xuemin Lin","is_ca":false},{"name":"Ying Zhang","is_ca":false},{"name":"Jian Pei","is_ca":true},{"name":"Wei Wang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01581869942841299,"gpt":0.2478756448548271,"spread":0.2320569454264141,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005255802,0.001267975,0.004297161,0.001927595,0.001973754,0.004476781,0.006446841,0.002418821,0.004531319],"category_scores_gemma":[0.02844143,0.001154295,0.001415844,0.004948397,0.001708401,0.008687461,0.005646341,0.00221494,0.001221483],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002524327,"about_ca_system_score_gemma":0.002218434,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00154627,"about_ca_topic_score_gemma":0.002668314,"domain_scores_codex":[0.9925354,0.00210052,0.000662991,0.001342293,0.002537078,0.0008217219],"domain_scores_gemma":[0.9810138,0.01192972,0.0008573743,0.003523069,0.001817405,0.0008586394],"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":[0.004472099,0.0005500252,0.004568313,0.001216027,0.0003095719,0.0003294176,0.000650089,0.6389386,0.01341984,0.09765514,0.02020927,0.2176817],"study_design_scores_gemma":[0.0001331216,0.000175215,0.000318502,0.00002336534,0.00008163668,0.0002834029,0.0001162709,0.9083197,0.002371334,0.0863077,0.001836449,0.0000332671],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07652448,0.002181008,0.9089671,0.001817242,0.0003082876,0.0004050494,0.001332521,0.001878556,0.006585797],"genre_scores_gemma":[0.8017426,0.000989435,0.1896081,0.0005458335,0.0003200917,0.000290124,0.001499074,0.000345593,0.004659254],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006446841,"threshold_uncertainty_score":0.02779567,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2172197636","doi":"10.1007/s00778-011-0260-8","title":"Scalability of write-ahead logging on multicore and multisocket hardware","year":2011,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Scalability; Multi-core processor; Serialization; Operating system; Parallel computing; Database","authors":[{"name":"Ryan Johnson","is_ca":true},{"name":"Ippokratis Pandis","is_ca":false},{"name":"Radu Stoica","is_ca":false},{"name":"Manos Athanassoulis","is_ca":false},{"name":"Anastasia Ailamaki","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04511224232357768,"gpt":0.2497341470458297,"spread":0.2046219047222521,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002332588,0.0008444802,0.0008002491,0.001236411,0.0009047691,0.001653621,0.001595659,0.0006628449,0.004637957],"category_scores_gemma":[0.01095811,0.0004839129,0.0002580027,0.001330553,0.001033298,0.0042172,0.001343323,0.001159931,0.0003651628],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001252011,"about_ca_system_score_gemma":0.002016804,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004870123,"about_ca_topic_score_gemma":0.006298512,"domain_scores_codex":[0.9971671,0.0004607956,0.0001768786,0.0004645162,0.001012444,0.0007183834],"domain_scores_gemma":[0.9876426,0.006438448,0.0007625024,0.002434784,0.001905354,0.0008163844],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.01293821,0.002268733,0.04639383,0.001214454,0.0003599546,0.0008254221,0.001099338,0.4467735,0.1558176,0.01237715,0.02336645,0.2965653],"study_design_scores_gemma":[0.0003442533,0.0009536664,0.01078879,0.0000590118,0.0001674101,0.0002547102,0.000628337,0.9268247,0.05278238,0.005514705,0.001599621,0.00008241854],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9782657,0.002033339,0.0106836,0.0005334045,0.000195889,0.00004194151,0.0003331036,0.003159921,0.00475299],"genre_scores_gemma":[0.9958111,0.0001713778,0.002629645,0.00004939362,0.00003510424,0.00001551118,0.0001873854,0.00006505865,0.001035552],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004870123,"threshold_uncertainty_score":0.01551551,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2129088288","doi":"10.1007/s00778-012-0297-3","title":"Modeling the execution semantics of stream processing engines with SECRET","year":2012,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":54,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Semantics (computer science); Tuple; Key (lock); Range (aeronautics); Variation (astronomy); Window (computing); Programming language; Theoretical computer science; World Wide Web; Operating system; Engineering","authors":[{"name":"Nihal Dindar","is_ca":false},{"name":"Nesime Tatbul","is_ca":false},{"name":"Renée J. Miller","is_ca":true},{"name":"Laura M. Haas","is_ca":false},{"name":"Irina Botan","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01984968428036772,"gpt":0.2453065712035755,"spread":0.2254568869232078,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002627453,0.0007554023,0.0006266485,0.001026538,0.0006783857,0.004750716,0.002361227,0.001305334,0.002184543],"category_scores_gemma":[0.008479436,0.0008807061,0.001216701,0.001080638,0.00190651,0.006652239,0.001487507,0.001632015,0.0005171864],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001788277,"about_ca_system_score_gemma":0.002496806,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009280956,"about_ca_topic_score_gemma":0.006805511,"domain_scores_codex":[0.9982061,0.0004352336,0.0001869517,0.0002830054,0.0006142219,0.0002744115],"domain_scores_gemma":[0.9961868,0.001759415,0.0004265931,0.0008833627,0.0006065163,0.0001374416],"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":[0.0003207278,0.0001097946,0.002696827,0.00008796752,0.00005130572,0.0003138934,0.0003753256,0.6986148,0.003254558,0.279989,0.001222956,0.01296274],"study_design_scores_gemma":[0.00001740435,0.00001108843,0.00005024698,0.000007523701,0.00002096608,0.00002283145,0.00002301794,0.9400804,0.002249144,0.05631048,0.001199256,0.000007768394],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08936688,0.000215911,0.9022364,0.000544658,0.00007740187,0.0001023999,0.0004816846,0.002188887,0.004785767],"genre_scores_gemma":[0.8408599,0.0004197824,0.1528312,0.0001452582,0.00007984701,0.0001274195,0.0006593601,0.0005459359,0.004331274],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009280956,"threshold_uncertainty_score":0.01845384,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2065714066","doi":"10.1007/s00778-005-0177-1","title":"Optimization and evaluation of shortest path queries","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Constrained Shortest Path First; Shortest path problem; K shortest path routing; Computer science; Yen's algorithm; Dijkstra's algorithm; Shortest Path Faster Algorithm; Euclidean shortest path; Scalability; Pruning; Path (computing); Query optimization; Algorithm; Theoretical computer science; Data mining; Database; Graph; Computer network","authors":[{"name":"Edward P. F. Chan","is_ca":true},{"name":"Heechul Lim","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01871367238197218,"gpt":0.2452819636881325,"spread":0.2265682913061603,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006121316,0.001390724,0.002382547,0.001817322,0.0009320736,0.002972526,0.002683273,0.001625605,0.004655022],"category_scores_gemma":[0.02688518,0.0006165013,0.0007899499,0.003154302,0.001317867,0.004186526,0.001633395,0.00117886,0.0006676768],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003460742,"about_ca_system_score_gemma":0.004201186,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01259749,"about_ca_topic_score_gemma":0.01194494,"domain_scores_codex":[0.9900511,0.004032047,0.000537075,0.001052948,0.003371754,0.00095502],"domain_scores_gemma":[0.9805335,0.01482385,0.0006311806,0.00145736,0.002030664,0.0005233619],"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":[0.00374782,0.0007408289,0.005166725,0.0006029709,0.0001584102,0.0001695537,0.0001887199,0.7288913,0.007308987,0.02021152,0.01265479,0.2201584],"study_design_scores_gemma":[0.00007874976,0.0001188591,0.0004863992,0.000006346985,0.0000211749,0.00004067344,0.0000548426,0.9899634,0.001811924,0.006854699,0.0005549653,0.000008072207],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.5108377,0.006259776,0.4530629,0.003076758,0.0003948516,0.0004570504,0.00218195,0.006253441,0.01747551],"genre_scores_gemma":[0.8432462,0.0006774544,0.1507577,0.0001252694,0.0001171874,0.0001003842,0.001994996,0.0005094907,0.002471229],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.01259749,"threshold_uncertainty_score":0.03237301,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2136058416","doi":"10.1007/s00778-006-0026-x","title":"Extending DBMSs with satellite databases","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Scalability; Middleware (distributed applications); Database; Architecture; Satellite; Distributed computing","authors":[{"name":"Christian Plattner","is_ca":false},{"name":"Gustavo Alonso","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01229410304909796,"gpt":0.2241104797065471,"spread":0.2118163766574492,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001847423,0.0004404899,0.0007182898,0.001106367,0.0007095409,0.002579496,0.001705114,0.0008796058,0.003569941],"category_scores_gemma":[0.005719007,0.0008101552,0.0008806658,0.002411155,0.0008637839,0.006261216,0.004119097,0.001421304,0.001201683],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003935452,"about_ca_system_score_gemma":0.0005474991,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002332545,"about_ca_topic_score_gemma":0.002979173,"domain_scores_codex":[0.9987649,0.0002770021,0.0001916245,0.0001738444,0.0004543703,0.0001384277],"domain_scores_gemma":[0.9959899,0.000834033,0.0001833121,0.002347108,0.0004827012,0.0001630118],"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":[0.0008458516,0.0003879266,0.0171979,0.001019211,0.0003706437,0.001445432,0.002744967,0.05821241,0.05354891,0.185716,0.02147584,0.6570349],"study_design_scores_gemma":[0.0002537331,0.0004380896,0.002946903,0.0002995011,0.0005892937,0.001906354,0.0009524247,0.2978427,0.06632508,0.2485948,0.3796961,0.0001550483],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1331743,0.002428932,0.8242638,0.001274051,0.0003755764,0.0001502258,0.0005476364,0.0168593,0.02092636],"genre_scores_gemma":[0.5479262,0.00229539,0.4364319,0.0005325383,0.0003139777,0.0001036697,0.001438129,0.0014134,0.00954474],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003569941,"threshold_uncertainty_score":0.01194263,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2021255304","doi":"10.1007/s00778-013-0316-z","title":"Sampling from repairs of conditional functional dependency violations","year":2013,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":46,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Dependency (UML); Functional dependency; Data mining; Sampling (signal processing); Class (philosophy); Space (punctuation); Variety (cybernetics); Semantics (computer science); Artificial intelligence; Relational database; Programming language","authors":[{"name":"George Beskales","is_ca":false},{"name":"Ihab F. Ilyas","is_ca":false},{"name":"Lukasz Golab","is_ca":true},{"name":"Artur Galiullin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.3025801455678561,"gpt":0.4077661213846789,"spread":0.1051859758168228,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01053574,0.0008267699,0.001814197,0.002020031,0.0006876005,0.001163122,0.003652363,0.001932743,0.003161123],"category_scores_gemma":[0.08984218,0.000849051,0.001434632,0.00165677,0.001430734,0.002351098,0.00213183,0.00260057,0.00057097],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007438241,"about_ca_system_score_gemma":0.002185455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00518581,"about_ca_topic_score_gemma":0.007995241,"domain_scores_codex":[0.9915009,0.00385371,0.0005659398,0.001716659,0.00157081,0.0007920372],"domain_scores_gemma":[0.8830783,0.08502354,0.004930286,0.01952668,0.005538608,0.001902634],"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":[0.005039938,0.001281608,0.2275777,0.0007922354,0.0007807433,0.00185057,0.00145903,0.3723983,0.00941426,0.03346537,0.01706895,0.3288713],"study_design_scores_gemma":[0.0001377366,0.0002794929,0.008612437,0.00005206377,0.0001354173,0.0003793274,0.0003094088,0.9500205,0.003451037,0.03525864,0.001332081,0.00003184853],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6686205,0.0004916348,0.3253549,0.0005609143,0.0001415093,0.0002412539,0.001292522,0.001825752,0.001471055],"genre_scores_gemma":[0.9538024,0.00008114153,0.04148501,0.0001472229,0.00005922643,0.0001255861,0.002948642,0.0001932471,0.00115744],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01053574,"threshold_uncertainty_score":0.05571902,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2123080243","doi":"10.1007/s00778-009-0159-9","title":"Schema mapping and query translation in heterogeneous P2P XML databases","year":2009,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":42,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia Hospital","funders":"","keywords":"Computer science; XQuery; Information retrieval; Scalability; Schema (genetic algorithms); Document Structure Description; Metadata; XML; Query language; XML Schema Editor; Database; World Wide Web","authors":[{"name":"Angela Bonifati","is_ca":false},{"name":"Elaine Chang","is_ca":true},{"name":"Terence Ho","is_ca":true},{"name":"Laks V. S. Lakshmanan","is_ca":true},{"name":"Rachel Pottinger","is_ca":true},{"name":"Yongik Chung","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04612929082107779,"gpt":0.2739027125305562,"spread":0.2277734217094784,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008666717,0.0006861103,0.001179836,0.001981994,0.001529199,0.006195565,0.002818827,0.001994882,0.003615539],"category_scores_gemma":[0.01546216,0.001147948,0.001438255,0.003282381,0.001855516,0.008977483,0.004362226,0.001955658,0.001419409],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001043667,"about_ca_system_score_gemma":0.001806628,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004554855,"about_ca_topic_score_gemma":0.002707478,"domain_scores_codex":[0.992761,0.00296797,0.0009762949,0.0008142027,0.002021222,0.00045933],"domain_scores_gemma":[0.9916053,0.004748578,0.0002988676,0.002218127,0.0009218721,0.0002072663],"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":[0.002195796,0.000952195,0.008024548,0.001313444,0.0003490633,0.004070913,0.005851466,0.07184397,0.03974527,0.272345,0.03441301,0.5588953],"study_design_scores_gemma":[0.0003420759,0.0002913464,0.002272367,0.0002167929,0.0003449768,0.001659737,0.003495942,0.6088299,0.1104695,0.2223496,0.0495418,0.0001861172],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06355721,0.0008864255,0.9141378,0.001622883,0.0002222154,0.0004591151,0.0009818865,0.01072583,0.007406521],"genre_scores_gemma":[0.51872,0.001249623,0.4652964,0.0008243945,0.0001464525,0.0004916566,0.004052925,0.001739913,0.00747857],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008666717,"threshold_uncertainty_score":0.04583454,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2099525917","doi":"10.1007/s00778-009-0157-y","title":"A framework for testing DBMS features","year":2009,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":41,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Database; Generator (circuit theory); Set (abstract data type); Feature (linguistics); Test case; Test suite; Test (biology); Data mining; Programming language; Power (physics); Machine learning","authors":[{"name":"Eric Lo","is_ca":false},{"name":"Carsten Binnig","is_ca":false},{"name":"Donald Kossmann","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true},{"name":"Wing-Kai Hon","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0426218733597992,"gpt":0.3124691026543456,"spread":0.2698472292945464,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01756854,0.002010628,0.00124445,0.00456622,0.001707732,0.006557925,0.007544273,0.004018087,0.005546407],"category_scores_gemma":[0.05526622,0.001865669,0.003094841,0.001909553,0.003932649,0.01529851,0.00580922,0.004642787,0.001399016],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002176106,"about_ca_system_score_gemma":0.00397787,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009674661,"about_ca_topic_score_gemma":0.007729436,"domain_scores_codex":[0.9720072,0.009471626,0.003317426,0.003379909,0.009620407,0.002203445],"domain_scores_gemma":[0.9446307,0.02869226,0.002712462,0.01539436,0.007135031,0.001435193],"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":[0.002055209,0.005468007,0.05006683,0.002061671,0.000857314,0.001645508,0.003530375,0.0960044,0.07654272,0.3425821,0.02628129,0.3929045],"study_design_scores_gemma":[0.0003871399,0.001703622,0.005842577,0.0004011229,0.0004282267,0.0009649803,0.001424494,0.7316642,0.07261649,0.153759,0.03048437,0.0003238978],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05392641,0.0001967339,0.9164899,0.0006941945,0.0001028681,0.001084625,0.001396775,0.02126331,0.004845141],"genre_scores_gemma":[0.2970859,0.0001145352,0.6957729,0.0003760609,0.00004877413,0.0008969759,0.002069521,0.001531177,0.002104207],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01756854,"threshold_uncertainty_score":0.09291244,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2086395677","doi":"10.1007/s00778-011-0217-y","title":"Interaction-aware scheduling of report-generation workloads","year":2011,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Online aggregation; Query optimization; View; Sargable; Scheduling (production processes); Web query classification; Leverage (statistics); Query language; IBM; Database; Workload; Query plan; Query expansion; Web search query; Data mining; Database design; Information retrieval; Search engine; Artificial intelligence; Operating system","authors":[{"name":"Mumtaz Ahmad","is_ca":true},{"name":"Ashraf Aboulnaga","is_ca":true},{"name":"Shivnath Babu","is_ca":false},{"name":"Kamesh Munagala","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08655948616922705,"gpt":0.2882161129723615,"spread":0.2016566268031345,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00199921,0.001037624,0.000960589,0.001026904,0.001039119,0.001844844,0.002867721,0.0005770706,0.002662172],"category_scores_gemma":[0.008098383,0.0006900636,0.0004130514,0.001422113,0.0003607253,0.001476037,0.001362784,0.0009406987,0.0008115894],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000745464,"about_ca_system_score_gemma":0.002279126,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003007676,"about_ca_topic_score_gemma":0.006438581,"domain_scores_codex":[0.9974306,0.0007958046,0.0001865725,0.0003235949,0.0007817353,0.0004816462],"domain_scores_gemma":[0.993395,0.002378829,0.0005509715,0.00170616,0.001162017,0.0008069121],"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":[0.01142434,0.002359899,0.02305819,0.0005415319,0.0003844998,0.00100653,0.00170399,0.2633903,0.2231225,0.01065392,0.04473553,0.4176187],"study_design_scores_gemma":[0.0001426031,0.0004877853,0.004331651,0.00001143715,0.00009833279,0.0001800278,0.0002198249,0.9587099,0.02729709,0.002687273,0.005780621,0.00005346302],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7204347,0.001752849,0.2436886,0.0009427477,0.0006094374,0.0002993814,0.0006683132,0.01897158,0.01263237],"genre_scores_gemma":[0.958801,0.0001219565,0.03833957,0.00009190742,0.0001373195,0.00005366672,0.0005850564,0.0004735469,0.001395902],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003007676,"threshold_uncertainty_score":0.01057297,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4364362023","doi":"10.1007/s00778-023-00794-0","title":"ByShard: sharding in a Byzantine environment","year":2023,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Computer science; Scalability; Distributed computing; Atomicity; Commit; Flexibility (engineering); Latency (audio); Isolation (microbiology); Byzantine fault tolerance; Protocol (science); Software deployment; Throughput; Database transaction; Computer network; Fault tolerance; Database; Operating system; Wireless","authors":[{"name":"Jelle Hellings","is_ca":true},{"name":"Mohammad Sadoghi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0163908489472987,"gpt":0.2284934902706444,"spread":0.2121026413233457,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002073474,0.0009285569,0.00101971,0.0008857762,0.00146124,0.002501484,0.00239647,0.00110602,0.009207987],"category_scores_gemma":[0.003457119,0.000866059,0.0006249112,0.001312058,0.001545625,0.005331272,0.00407541,0.00175881,0.002073103],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005625294,"about_ca_system_score_gemma":0.001228787,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001590279,"about_ca_topic_score_gemma":0.001728409,"domain_scores_codex":[0.9988574,0.0002887611,0.00009258645,0.0001783301,0.0003965568,0.0001863648],"domain_scores_gemma":[0.997583,0.0004824325,0.0001302778,0.001379246,0.0002145748,0.0002105488],"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":[0.006576627,0.0006631613,0.005906114,0.001266168,0.0003392592,0.001253871,0.002426211,0.04701097,0.04692199,0.154786,0.1566947,0.576155],"study_design_scores_gemma":[0.001415147,0.001239532,0.002030935,0.000298473,0.0002909037,0.001060495,0.0007023344,0.4488291,0.1085325,0.1552643,0.2800564,0.0002799403],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0816086,0.001353782,0.7442208,0.0009373897,0.001024641,0.0002720196,0.001281691,0.1522357,0.0170653],"genre_scores_gemma":[0.5976306,0.001222143,0.3479454,0.0006624365,0.0003177685,0.0003116488,0.004291348,0.01428552,0.03333308],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009207987,"threshold_uncertainty_score":0.0308038,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2014871820","doi":"10.1007/s00778-011-0238-6","title":"Answering pattern match queries in large graph databases via graph embedding","year":2011,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Reachability; Graph database; Graph; Factor-critical graph; Theoretical computer science; Web search query; Data mining; Information retrieval; Line graph; Search engine; Voltage graph","authors":[{"name":"Lei Zou","is_ca":false},{"name":"Lei Chen","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true},{"name":"Dongyan Zhao","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03072226293727882,"gpt":0.2576678277286159,"spread":0.226945564791337,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001842251,0.001312567,0.003172606,0.004395175,0.001218763,0.00486026,0.003630594,0.003088491,0.002675725],"category_scores_gemma":[0.01662506,0.001258675,0.001673419,0.007504971,0.001211982,0.01623135,0.003637803,0.00240974,0.001090825],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001204907,"about_ca_system_score_gemma":0.001401393,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004410752,"about_ca_topic_score_gemma":0.007966775,"domain_scores_codex":[0.995567,0.001019758,0.0005335532,0.001329583,0.001263476,0.0002865019],"domain_scores_gemma":[0.9845557,0.00989353,0.001134136,0.003077355,0.000910357,0.0004290743],"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":[0.003032586,0.002043365,0.02887049,0.002256018,0.001006259,0.001218014,0.001456556,0.2765237,0.02825109,0.0500709,0.05078782,0.5544832],"study_design_scores_gemma":[0.0001532705,0.0001712054,0.001330664,0.00003580336,0.000177047,0.0004438621,0.0007721808,0.8356786,0.005777818,0.1521964,0.00322344,0.00003972288],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3495128,0.003650072,0.6187513,0.005936002,0.000264139,0.0004587381,0.007378902,0.01054663,0.003501387],"genre_scores_gemma":[0.6912882,0.001262445,0.2905577,0.0008018891,0.0002860343,0.0001603419,0.01232036,0.0005903595,0.002732602],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00486026,"threshold_uncertainty_score":0.009742856,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2145957360","doi":"10.1007/s00778-011-0230-1","title":"Maximizing bichromatic reverse nearest neighbor for L p -norm in two- and three-dimensional spaces","year":2011,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Curse of dimensionality; Computer science; k-nearest neighbors algorithm; Norm (philosophy); Fixed-radius near neighbors; Best bin first; Space (punctuation); Algorithm; Nearest neighbor search; Theoretical computer science; Combinatorics; Mathematics; Data mining; Artificial intelligence; Cluster analysis","authors":[{"name":"Raymond Chi-Wing Wong","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true},{"name":"Ada Wai-Chee Fu","is_ca":false},{"name":"Philip S. Yu","is_ca":false},{"name":"Lian Liu","is_ca":false},{"name":"Yubao Liu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04697737075967021,"gpt":0.2579352204925931,"spread":0.2109578497329229,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00359088,0.0009867304,0.001782469,0.001254801,0.0008232119,0.00167725,0.002509172,0.001667448,0.001827541],"category_scores_gemma":[0.01531178,0.0005442859,0.0007234583,0.00150866,0.001267911,0.002738819,0.002670983,0.001176265,0.0006806678],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009535821,"about_ca_system_score_gemma":0.001184222,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003937567,"about_ca_topic_score_gemma":0.004985914,"domain_scores_codex":[0.9973219,0.001175621,0.0001427492,0.0004019975,0.0008214145,0.0001362851],"domain_scores_gemma":[0.995,0.002784934,0.0003327577,0.0006539825,0.001033251,0.0001951981],"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":[0.0008845334,0.0002836921,0.002740845,0.0005641581,0.0001389672,0.0002095421,0.0003538934,0.5709987,0.01230466,0.07155029,0.00767636,0.3322944],"study_design_scores_gemma":[0.00001257538,0.0000493976,0.0002065479,0.00001027196,0.000008167142,0.0000649469,0.00003339589,0.9796363,0.001535231,0.01794044,0.0004899972,0.00001278758],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02451885,0.0003408119,0.9735702,0.0001678781,0.00003250904,0.00004076932,0.00009572914,0.0002123433,0.00102088],"genre_scores_gemma":[0.2807637,0.0003894029,0.7153053,0.0001533829,0.00008752726,0.0001750451,0.0005488069,0.0002152592,0.002361519],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003937567,"threshold_uncertainty_score":0.01899064,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2386452338","doi":"10.1007/s00778-017-0463-8","title":"Efficiently making (almost) any concurrency control mechanism serializable","year":2017,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Serialization; Concurrency control; Serializability; Commit; Transactional memory; Distributed transaction; Online transaction processing; Database transaction","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0182709251801704,"gpt":0.2767483095853744,"spread":0.258477384405204,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004282639,0.001183663,0.001144165,0.0008822659,0.001304288,0.003839021,0.00282738,0.001006845,0.006701923],"category_scores_gemma":[0.01325451,0.001168532,0.0008838439,0.001595222,0.002224283,0.005379848,0.004380147,0.002408984,0.002012749],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007705898,"about_ca_system_score_gemma":0.003085833,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002168544,"about_ca_topic_score_gemma":0.003015362,"domain_scores_codex":[0.9947259,0.001000041,0.0004270723,0.0008942952,0.002168217,0.0007843787],"domain_scores_gemma":[0.984302,0.003307709,0.0007012943,0.01003773,0.001250756,0.0004004167],"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":[0.00386812,0.000764676,0.005819903,0.0006029899,0.0002458489,0.0005565429,0.001135739,0.09781514,0.1404056,0.168762,0.02194476,0.5580787],"study_design_scores_gemma":[0.0009028547,0.0006556595,0.0013628,0.0001169829,0.0003959349,0.0004161755,0.0004137798,0.4831798,0.1474164,0.3281739,0.03681803,0.0001477126],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1055866,0.0004201545,0.8624067,0.0007853316,0.000409085,0.0002991787,0.0002206264,0.01722453,0.01264782],"genre_scores_gemma":[0.7097064,0.0003176954,0.2713014,0.0004653296,0.0002210348,0.0002560393,0.0006009695,0.001826227,0.0153049],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006701923,"threshold_uncertainty_score":0.02264905,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1974932383","doi":"10.1007/s00778-011-0228-8","title":"Elastic SI-Cache: consistent and scalable caching in multi-tier architectures","year":2011,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Cache; Scalability; Cloud computing; Smart Cache; Distributed computing; Cache invalidation; Cache coloring; Cache algorithms; Benchmark (surveying); Software deployment; Operating system; CPU cache","authors":[{"name":"Francisco Perez-Sorrosal","is_ca":false},{"name":"Marta Patiño-Martı́nez","is_ca":false},{"name":"Bettina Kemme","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03874119627521677,"gpt":0.2340408466965079,"spread":0.1952996504212912,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002030291,0.0007935461,0.001007172,0.0009861566,0.00139446,0.002969901,0.004685883,0.0009739334,0.00301596],"category_scores_gemma":[0.004670437,0.0006873924,0.0004915883,0.002683265,0.0009247199,0.004068816,0.002288194,0.001566935,0.0007922933],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001334314,"about_ca_system_score_gemma":0.003221431,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009194011,"about_ca_topic_score_gemma":0.02195845,"domain_scores_codex":[0.9985096,0.0002729014,0.0001344825,0.0001928756,0.0005939289,0.0002961927],"domain_scores_gemma":[0.9958915,0.0006532358,0.0002121328,0.002095151,0.0008660267,0.0002819407],"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":[0.00543092,0.0009791035,0.01672876,0.0005206286,0.000508772,0.0007612021,0.0005012627,0.3125937,0.05275954,0.07778084,0.0915332,0.439902],"study_design_scores_gemma":[0.0002051441,0.000215294,0.001010121,0.00002023642,0.0001086338,0.0002143574,0.0001305927,0.95847,0.01416955,0.01834727,0.007053705,0.00005515484],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2566962,0.006606243,0.6889049,0.001798563,0.0008846679,0.0003345222,0.001532927,0.0268524,0.01638954],"genre_scores_gemma":[0.8210273,0.0008110158,0.1692946,0.0003735695,0.0001944614,0.0001016519,0.001667021,0.0005545751,0.005975769],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009194011,"threshold_uncertainty_score":0.01828098,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2046474027","doi":"10.1007/s00778-013-0318-x","title":"Consistency anomalies in multi-tier architectures: automatic detection and prevention","year":2013,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Consistency (knowledge bases); Computer science; Anomaly detection; Isolation (microbiology); Database transaction; Set (abstract data type); Data mining; Database; Distributed computing; Artificial intelligence; Programming language","authors":[{"name":"Kamal Zellag","is_ca":true},{"name":"Bettina Kemme","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0173761358373862,"gpt":0.2477361317347878,"spread":0.2303599958974016,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003576198,0.0006783207,0.001007138,0.003006988,0.0008076833,0.002079629,0.0020933,0.001355735,0.0007172665],"category_scores_gemma":[0.02071556,0.0006908653,0.0005507961,0.002186673,0.0008296196,0.002405793,0.001675058,0.001721138,0.0003384245],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009190132,"about_ca_system_score_gemma":0.002357247,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003406652,"about_ca_topic_score_gemma":0.005865923,"domain_scores_codex":[0.9932188,0.00101438,0.0006081504,0.001092431,0.003510432,0.0005558304],"domain_scores_gemma":[0.9679357,0.01020362,0.007712883,0.007344904,0.006023142,0.000779603],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001631017,0.0006726473,0.4160343,0.0005142884,0.0003467167,0.001087252,0.0007645724,0.05918337,0.08937742,0.01140177,0.008194437,0.4107921],"study_design_scores_gemma":[0.00008068988,0.0003392546,0.04792956,0.00006737489,0.0002307933,0.001871686,0.0003587384,0.8731759,0.05829308,0.01448726,0.003081803,0.00008382606],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5713867,0.002327119,0.4081288,0.001014926,0.000328427,0.0002206025,0.0007075186,0.01370274,0.002183099],"genre_scores_gemma":[0.9479831,0.0002240859,0.05054151,0.00009339856,0.00004843601,0.00003049189,0.0004332561,0.0001653124,0.0004802577],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003576198,"threshold_uncertainty_score":0.01891303,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2026808466","doi":"10.1007/s00778-010-0196-4","title":"Ranking queries on uncertain data","year":2010,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Ranking (information retrieval); Uncertain data; Rank (graph theory); Computer science; Set (abstract data type); Probabilistic logic; Data mining; Ranking SVM; Information retrieval; Mathematics; Combinatorics; Artificial intelligence","authors":[{"name":"Hua Ming","is_ca":false},{"name":"Jian Pei","is_ca":true},{"name":"Xuemin Lin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05671574245012417,"gpt":0.2978760619772347,"spread":0.2411603195271105,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00537132,0.001213324,0.003343462,0.004147835,0.001146764,0.005552773,0.002323704,0.002219984,0.003991379],"category_scores_gemma":[0.02581682,0.0007750656,0.00114909,0.007318537,0.001105433,0.008517276,0.002005896,0.002163668,0.001341978],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001408579,"about_ca_system_score_gemma":0.001141643,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002119123,"about_ca_topic_score_gemma":0.002882676,"domain_scores_codex":[0.9909115,0.002909408,0.0009129805,0.001070551,0.003642256,0.0005532038],"domain_scores_gemma":[0.9774842,0.01653299,0.00104873,0.00267237,0.001683091,0.0005786863],"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":[0.002624922,0.0005547741,0.009920379,0.002154478,0.0006211415,0.001445938,0.0006204713,0.3148805,0.01253988,0.1151754,0.06069643,0.4787657],"study_design_scores_gemma":[0.00007718947,0.0002432091,0.0009125895,0.00006372145,0.0001364698,0.000604653,0.0002309345,0.8334959,0.004217181,0.1522383,0.007732552,0.00004737453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1456787,0.009827557,0.8245667,0.005510726,0.0006015592,0.0002959726,0.004708241,0.00289983,0.005910596],"genre_scores_gemma":[0.7564956,0.003676836,0.2227139,0.0006690067,0.001573025,0.0001842504,0.008271473,0.0004117038,0.006004146],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005552773,"threshold_uncertainty_score":0.02840656,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2141833902","doi":"10.1007/s00778-012-0278-6","title":"Scalable and dynamically balanced shared-everything OLTP with physiological partitioning","year":2012,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Scalability; Thread (computing); Heap (data structure); Multi-core processor; Distributed computing; Parallel computing; Database transaction; Operating system; Programming language","authors":[{"name":"Pınar Tözün","is_ca":false},{"name":"Ippokratis Pandis","is_ca":false},{"name":"Ryan Johnson","is_ca":true},{"name":"Anastasia Ailamaki","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01529434217783311,"gpt":0.2264422298769785,"spread":0.2111478876991454,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009503695,0.001023388,0.0009651217,0.0008141769,0.001663369,0.002582104,0.002920201,0.0006474412,0.003217463],"category_scores_gemma":[0.003468811,0.000531869,0.0003363095,0.001309752,0.0007249746,0.003361661,0.003186025,0.0008348643,0.0008338459],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001053018,"about_ca_system_score_gemma":0.00195608,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004078036,"about_ca_topic_score_gemma":0.008466125,"domain_scores_codex":[0.9985865,0.0002368301,0.0001021953,0.0003208041,0.0003889868,0.0003646812],"domain_scores_gemma":[0.9975429,0.0005738781,0.0001459951,0.001010424,0.0004861619,0.000240685],"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":[0.006392353,0.001139561,0.008866444,0.0004291719,0.0002968229,0.0009498124,0.000771019,0.4017914,0.08720358,0.01612647,0.02899563,0.4470378],"study_design_scores_gemma":[0.0003069294,0.000408421,0.00176331,0.00002360683,0.0001273822,0.0003224757,0.0005387131,0.958234,0.01828028,0.01575557,0.004185177,0.00005431319],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4514824,0.001830858,0.5090193,0.001204997,0.0006647793,0.0004434352,0.0009833635,0.01419289,0.02017797],"genre_scores_gemma":[0.9311538,0.000129899,0.06520716,0.0001179593,0.00007389898,0.00008985013,0.0006235191,0.0002269776,0.00237684],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004078036,"threshold_uncertainty_score":0.01076347,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2002220286","doi":"10.1007/s00778-013-0312-3","title":"Eliminating unscalable communication in transaction processing","year":2013,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Online transaction processing; Transaction processing; Transaction processing system; Scalability; Distributed transaction; Distributed computing; Database transaction; Concurrency; Thread (computing); Compensating transaction; Operating system; Database","authors":[{"name":"Ryan Johnson","is_ca":true},{"name":"Ippokratis Pandis","is_ca":false},{"name":"Anastasia Ailamaki","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01070044184859699,"gpt":0.2297111019525497,"spread":0.2190106601039527,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005058601,0.0008664032,0.001283111,0.001314187,0.00306252,0.002855452,0.004634964,0.001968764,0.004516142],"category_scores_gemma":[0.02680689,0.001295404,0.0007330774,0.00254282,0.00325073,0.008164631,0.004316665,0.003966322,0.001371823],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00130574,"about_ca_system_score_gemma":0.004321704,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001937121,"about_ca_topic_score_gemma":0.003419194,"domain_scores_codex":[0.989915,0.003511229,0.0007066514,0.0008002752,0.003824685,0.001242097],"domain_scores_gemma":[0.9593445,0.01621717,0.002244409,0.01713648,0.004272508,0.000784993],"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":[0.002822626,0.001117379,0.008083401,0.0008209459,0.0001507556,0.0005531142,0.00188184,0.1241548,0.02846061,0.1773231,0.01733802,0.6372934],"study_design_scores_gemma":[0.000300485,0.0006723965,0.001192652,0.0001284492,0.0002425731,0.0004269388,0.0006449567,0.6871874,0.05421243,0.2370526,0.01783727,0.000101802],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1827174,0.002779995,0.7932744,0.002480312,0.0007085939,0.0002833699,0.0001493889,0.005859603,0.01174689],"genre_scores_gemma":[0.833163,0.0006302186,0.1568234,0.0007536038,0.0004290492,0.0002257582,0.0002004942,0.0008404313,0.006933896],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005058601,"threshold_uncertainty_score":0.02675277,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1999968379","doi":"10.1007/s00778-010-0180-z","title":"Streaming multiple aggregations using phantoms","year":2010,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Granularity; Data stream mining; Aggregate (composite); Petabyte; Computation; Volume (thermodynamics); Distributed computing; Set (abstract data type); Data mining; The Internet; Service (business); Computer network; Real-time computing; Big data; Algorithm","authors":[{"name":"Rui Zhang","is_ca":false},{"name":"Nick Koudas","is_ca":true},{"name":"Beng Chin Ooi","is_ca":false},{"name":"Divesh Srivastava","is_ca":false},{"name":"Pu Zhou","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02017665443236114,"gpt":0.2689327796126706,"spread":0.2487561251803095,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004304619,0.001252595,0.002254617,0.002089137,0.001619651,0.003337681,0.002995435,0.001473573,0.01048976],"category_scores_gemma":[0.01893803,0.001539362,0.0009351251,0.003397919,0.001238235,0.006386018,0.004475502,0.002232794,0.002208334],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001360616,"about_ca_system_score_gemma":0.00173306,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002426693,"about_ca_topic_score_gemma":0.002368277,"domain_scores_codex":[0.9966425,0.0007818069,0.0003015049,0.0005887366,0.001338811,0.0003465448],"domain_scores_gemma":[0.9837992,0.00557074,0.000496161,0.006867094,0.00231849,0.0009482407],"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":[0.008996285,0.001010461,0.005625395,0.00050151,0.0003904451,0.001451145,0.001286327,0.1032499,0.0805121,0.09944651,0.05910219,0.6384276],"study_design_scores_gemma":[0.0003454161,0.0002825224,0.0005153343,0.00004502426,0.00007915028,0.000390483,0.0001257551,0.9110613,0.02549686,0.04392191,0.01765995,0.00007632569],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05224688,0.0007239908,0.9024526,0.000766878,0.0006949221,0.0002043744,0.0006936922,0.03383715,0.008379376],"genre_scores_gemma":[0.4219685,0.0003451949,0.5657052,0.0003026213,0.0004156815,0.0002221389,0.001879816,0.00186809,0.00729268],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01048976,"threshold_uncertainty_score":0.03509176,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2170688652","doi":"10.1007/s00778-007-0067-9","title":"Applications of corpus-based semantic similarity and word segmentation to database schema matching","year":2007,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":26,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Schema matching; Computer science; Semantic similarity; Artificial intelligence; Schema (genetic algorithms); Star schema; Natural language processing; Information retrieval; Segmentation; Pattern recognition (psychology); Database schema; Data mining; Data integration","authors":[{"name":"Aminul Islam","is_ca":true},{"name":"Diana Inkpen","is_ca":true},{"name":"Iluju Kiringa","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1166235600128447,"gpt":0.4294959857511475,"spread":0.3128724257383028,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004904095,0.000542073,0.001111243,0.01069785,0.001374315,0.004032878,0.001548591,0.001018202,0.00392179],"category_scores_gemma":[0.03056155,0.0004899811,0.000765036,0.01182292,0.001463695,0.005064588,0.002082872,0.0008781508,0.000922015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001422218,"about_ca_system_score_gemma":0.002580957,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006346751,"about_ca_topic_score_gemma":0.006656075,"domain_scores_codex":[0.9953908,0.001990187,0.0006650296,0.0006860617,0.001145322,0.0001225396],"domain_scores_gemma":[0.9822104,0.01159737,0.0007956791,0.002304488,0.002822255,0.0002697332],"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":[0.0006697194,0.0004790099,0.009518186,0.0007994192,0.0003343703,0.0003659847,0.00244703,0.03078659,0.02509066,0.08084552,0.00869566,0.8399679],"study_design_scores_gemma":[0.000188009,0.0002686151,0.006707333,0.0002107029,0.0002709236,0.0009488639,0.002455052,0.7357318,0.03135468,0.1949406,0.02678681,0.0001365485],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08773176,0.001642928,0.8973171,0.00069617,0.000191193,0.000377973,0.001284557,0.003974736,0.006783556],"genre_scores_gemma":[0.3245328,0.0007950961,0.669481,0.0001539444,0.0001072612,0.000257093,0.002998144,0.0005318079,0.001142874],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01069785,"threshold_uncertainty_score":0.02593565,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2147440220","doi":"10.1007/s007780000029","title":"One-dimensional and multi-dimensional substring selectivity estimation","year":2000,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Substring; Computer science; String (physics); Context (archaeology); Matching (statistics); String searching algorithm; Constraint (computer-aided design); Theoretical computer science; Algorithm; Data structure; Mathematics; Pattern matching; Artificial intelligence; Statistics","authors":[{"name":"H. V. Jagadish","is_ca":false},{"name":"Olga Kapitskaia","is_ca":false},{"name":"Raymond T. Ng","is_ca":true},{"name":"Divesh Srivastava","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02152338838573285,"gpt":0.2492014349661,"spread":0.2276780465803671,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001139288,0.0003868374,0.0008862528,0.001482466,0.0003773544,0.0008367156,0.0008822953,0.0008137988,0.001846734],"category_scores_gemma":[0.005531182,0.0002651454,0.0004396738,0.001854872,0.0003719233,0.001523749,0.001204274,0.000601887,0.001119705],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000323575,"about_ca_system_score_gemma":0.0006268462,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001488841,"about_ca_topic_score_gemma":0.002593106,"domain_scores_codex":[0.9992087,0.000133289,0.00005980219,0.0001393687,0.0003417246,0.0001170997],"domain_scores_gemma":[0.996612,0.001816886,0.0002627646,0.0007149627,0.0004648532,0.0001283826],"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":[0.001133608,0.0001959678,0.01881465,0.0002390976,0.0001090613,0.000413215,0.0001163405,0.07108977,0.06565055,0.01179903,0.003096979,0.8273417],"study_design_scores_gemma":[0.00002064124,0.00008449229,0.009883365,0.00001586883,0.00004656679,0.0007437708,0.00009802698,0.9378821,0.04223669,0.006845319,0.002104762,0.0000383318],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1274479,0.001240831,0.8679289,0.0002794506,0.00006943396,0.00003256906,0.0005910651,0.0007558807,0.001653955],"genre_scores_gemma":[0.6536906,0.001425756,0.337769,0.0001507876,0.0002370064,0.0000690936,0.001930933,0.0001032469,0.004623554],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.001846734,"threshold_uncertainty_score":0.006177902,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2006326991","doi":"10.1007/s007780100059","title":"Efficient retrieval of similar shapes","year":2002,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; University of Alberta","funders":"","keywords":"Search engine indexing; Normalization (sociology); Computer science; Pattern recognition (psychology); Artificial intelligence; Similarity (geometry); Invariant (physics); Image retrieval; Computer vision; Mathematics; Algorithm; Image (mathematics)","authors":[{"name":"Davood Rafiei","is_ca":true},{"name":"Alberto O. Mendelzon","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02979951787720897,"gpt":0.2431261256838491,"spread":0.2133266078066401,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003674314,0.001256687,0.002790239,0.004004979,0.00110848,0.002422679,0.002372127,0.001905432,0.01066349],"category_scores_gemma":[0.002547072,0.0008038718,0.001689322,0.005826809,0.0007705652,0.002571541,0.002794491,0.001011029,0.008870185],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008087635,"about_ca_system_score_gemma":0.0009247455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002580852,"about_ca_topic_score_gemma":0.004292394,"domain_scores_codex":[0.9989386,0.00009294043,0.00006111713,0.0002325813,0.0005396694,0.0001350417],"domain_scores_gemma":[0.9990628,0.0001469593,0.00007211367,0.0004446522,0.0002088249,0.0000646551],"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":[0.001030773,0.000230463,0.0009857308,0.0004205427,0.0001505059,0.0005899456,0.0001807525,0.01319908,0.2124356,0.009003935,0.01983851,0.7419342],"study_design_scores_gemma":[0.0003574578,0.0008037992,0.004907127,0.0001009581,0.0004285359,0.006689903,0.0007477771,0.6472642,0.2549767,0.03861485,0.04495137,0.0001573967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1339589,0.0040721,0.831974,0.0005223525,0.0004382265,0.000553487,0.002083649,0.009410014,0.01698718],"genre_scores_gemma":[0.3267634,0.002777432,0.6334838,0.000504346,0.000359562,0.0002710017,0.008938349,0.001577201,0.02532496],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01066349,"threshold_uncertainty_score":0.0356729,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2035006988","doi":"10.1007/s00778-006-0017-y","title":"Query processing of multi-way stream window joins","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Joins; Computer science; Join (topology); Data stream mining; Window (computing); Sliding window protocol; Data stream; Database; Data mining; Aggregate (composite); Distributed computing; Real-time computing; Programming language","authors":[{"name":"Moustafa A. Hammad","is_ca":true},{"name":"Walid G. Aref","is_ca":false},{"name":"Ahmed K. Elmagarmid","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01945096438599307,"gpt":0.2412950354553263,"spread":0.2218440710693332,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002607685,0.000689307,0.001931231,0.001556756,0.00126007,0.00336873,0.001470706,0.0008026635,0.003050573],"category_scores_gemma":[0.005215438,0.0005887022,0.0006154604,0.0027275,0.0004983344,0.003812851,0.001956798,0.0008608769,0.0006960512],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004564847,"about_ca_system_score_gemma":0.0008036967,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001375432,"about_ca_topic_score_gemma":0.001938983,"domain_scores_codex":[0.9982241,0.0002988744,0.0002076386,0.0002970872,0.0007452429,0.0002269798],"domain_scores_gemma":[0.9972784,0.00123722,0.0001911205,0.0006561905,0.0004107706,0.0002263354],"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":[0.01308214,0.0008810909,0.01980982,0.0006654497,0.0005714099,0.001293085,0.001455867,0.08008959,0.161432,0.03351938,0.01985696,0.6673433],"study_design_scores_gemma":[0.0003915028,0.0009022259,0.004426768,0.0000296283,0.0001821651,0.0007000703,0.0006348787,0.892761,0.06272402,0.02759729,0.009592785,0.00005769134],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4424849,0.00291421,0.5424862,0.0006114672,0.0003880504,0.0002339029,0.001090259,0.005055828,0.004735194],"genre_scores_gemma":[0.8141633,0.0007515874,0.1785797,0.0001003628,0.000294288,0.0001425562,0.001686016,0.0003090618,0.00397324],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00336873,"threshold_uncertainty_score":0.01379097,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4387098977","doi":"10.1007/s00778-023-00814-z","title":"A survey on transactional stream processing","year":2023,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Banting and Best Diabetes Centre, University of Toronto; Singapore University of Technology and Design; National Research Foundation; Deutsche Forschungsgemeinschaft; Bundesministerium für Bildung und Forschung; Info-communications Media Development Authority; National Research Foundation Singapore; Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie; Technische Universität Berlin","keywords":"Transactional leadership; Transactional analysis; Computer science; Perspective (graphical); Stream processing; Data science; Transaction processing; Database transaction; Distributed computing; Database; Artificial intelligence; Management; Economics; Psychology","authors":[{"name":"Shuhao Zhang","is_ca":false},{"name":"Juan Soto","is_ca":false},{"name":"Volker Markl","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04847366620191614,"gpt":0.2954808845975803,"spread":0.2470072183956642,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002138055,0.0009378541,0.000983917,0.004061101,0.000626496,0.00355887,0.0018648,0.00120136,0.007193884],"category_scores_gemma":[0.004273574,0.0007748754,0.0007990787,0.01378295,0.0007986816,0.006126662,0.001342584,0.002028301,0.002767008],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009390747,"about_ca_system_score_gemma":0.001824404,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001529587,"about_ca_topic_score_gemma":0.0007388945,"domain_scores_codex":[0.9977636,0.0004211985,0.0002971717,0.0003423887,0.001005457,0.0001701778],"domain_scores_gemma":[0.9962425,0.001954081,0.0001971392,0.0003453469,0.001088405,0.0001725063],"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":[0.0001959077,0.0001829146,0.002123195,0.006775458,0.0001135809,0.0002406165,0.0003087945,0.008410766,0.002609213,0.07204111,0.04917137,0.857827],"study_design_scores_gemma":[0.00002988213,0.000273453,0.00201707,0.002491748,0.00008441632,0.001239007,0.0003237585,0.02592468,0.002756143,0.05129959,0.9134897,0.00007053261],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.008918214,0.7758262,0.1561585,0.003554551,0.001819728,0.0002634033,0.0007852322,0.001160502,0.05151368],"genre_scores_gemma":[0.05120351,0.8520576,0.08086892,0.001713605,0.003448811,0.0001987588,0.001770241,0.0003311789,0.008407349],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.007193884,"threshold_uncertainty_score":0.02406597,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2157044362","doi":"10.1007/s00778-008-0128-8","title":"Top-k typicality queries and efficient query answering methods on large databases","year":2009,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Discriminative model; Set (abstract data type); Tree (set theory); Information retrieval; Measure (data warehouse); Quality (philosophy); Series (stratigraphy); Data mining; Artificial intelligence; Mathematics","authors":[{"name":"Hua Ming","is_ca":true},{"name":"Jian Pei","is_ca":true},{"name":"Ada W. C. Fu","is_ca":false},{"name":"Xuemin Lin","is_ca":false},{"name":"Ho-fung Leung","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03418460099070763,"gpt":0.3495786809841878,"spread":0.3153940799934802,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008500651,0.001527301,0.004818978,0.00334122,0.0024818,0.006715759,0.007021165,0.003233871,0.005305656],"category_scores_gemma":[0.04078459,0.002013889,0.002561387,0.009267921,0.003048939,0.01916537,0.005584576,0.003811864,0.002040759],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002216254,"about_ca_system_score_gemma":0.002344378,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00274866,"about_ca_topic_score_gemma":0.003151962,"domain_scores_codex":[0.9869449,0.003939845,0.001567791,0.002749191,0.003823037,0.0009751743],"domain_scores_gemma":[0.964893,0.02176647,0.00140174,0.008451535,0.002760776,0.0007264594],"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":[0.004140123,0.0004978483,0.00950204,0.002076219,0.0006286331,0.0005512723,0.001694934,0.1967379,0.01605605,0.2537173,0.05391936,0.4604784],"study_design_scores_gemma":[0.000147047,0.0001177571,0.0005880591,0.00004036884,0.0001389733,0.0005980825,0.0003120322,0.5930769,0.004404963,0.3957084,0.0048062,0.00006128094],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0323919,0.002384635,0.9565058,0.001988192,0.0001420678,0.0001011085,0.001177637,0.003597562,0.001711157],"genre_scores_gemma":[0.4254284,0.001441173,0.563081,0.0007538561,0.0007796686,0.0002725379,0.003677942,0.001061245,0.003504214],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008500651,"threshold_uncertainty_score":0.04495627,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2024336226","doi":"10.1007/s00778-006-0015-0","title":"Histograms based on the minimum description length principle","year":2006,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Saint Mary's University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Histogram; Computer science; Data mining; Set (abstract data type); Representation (politics); Overhead (engineering); Online analytical processing; Data set; Algorithm; Pattern recognition (psychology); Artificial intelligence; Image (mathematics); Data warehouse","authors":[{"name":"Hai Wang","is_ca":true},{"name":"Kenneth C. Sevcik","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0234515832102263,"gpt":0.2229272232546076,"spread":0.1994756400443813,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001388685,0.0004242211,0.001290412,0.002103799,0.0006321134,0.001905788,0.001771838,0.001031972,0.006366499],"category_scores_gemma":[0.00702326,0.0005906429,0.0006776342,0.00223657,0.001008997,0.004397087,0.001938121,0.001750517,0.001464011],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007442914,"about_ca_system_score_gemma":0.0008641944,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001694131,"about_ca_topic_score_gemma":0.001252762,"domain_scores_codex":[0.9989535,0.0002712279,0.00005018016,0.0001949072,0.0004239116,0.0001063305],"domain_scores_gemma":[0.9958961,0.002351341,0.0002426839,0.0006529803,0.000651457,0.000205485],"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":[0.0007357134,0.0001316061,0.00219497,0.0003819184,0.0001160994,0.0001668061,0.0001613401,0.1294976,0.01586854,0.3271496,0.01295596,0.5106398],"study_design_scores_gemma":[0.00005403048,0.00009874269,0.001088511,0.00003181296,0.00003216869,0.0001517462,0.00004149525,0.7658569,0.004949989,0.2213033,0.006330318,0.000061004],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008013211,0.0005881589,0.9888158,0.0002570182,0.00006814592,0.00002733497,0.0002401323,0.0005421258,0.001448094],"genre_scores_gemma":[0.4229415,0.001945704,0.5633664,0.0005369673,0.0006358157,0.0002171757,0.002124734,0.0007606627,0.007471071],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006366499,"threshold_uncertainty_score":0.02129805,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3171876400","doi":"10.1007/s00778-021-00679-0","title":"A fractional memory-efficient approach for online continuous-time influence maximization","year":2021,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":12,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Maximization; Computer science; Mathematical optimization; Mathematics","authors":[{"name":"Glenn S. Bevilacqua","is_ca":true},{"name":"Laks V. S. Lakshmanan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01091987405908378,"gpt":0.2582131317671943,"spread":0.2472932577081105,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001676571,0.001717877,0.002095831,0.001351428,0.0008002879,0.001879838,0.002595353,0.002209807,0.005645781],"category_scores_gemma":[0.01007466,0.0008567839,0.001302619,0.001750431,0.001759977,0.002489925,0.002388593,0.002441121,0.0007506197],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001635218,"about_ca_system_score_gemma":0.001852472,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006246966,"about_ca_topic_score_gemma":0.008151116,"domain_scores_codex":[0.999235,0.0002460037,0.00003400011,0.0001565437,0.0002112803,0.000117165],"domain_scores_gemma":[0.9945557,0.004592307,0.0001792069,0.0002257067,0.0002985057,0.0001485013],"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":[0.0001028335,0.00009366921,0.0002713628,0.0001977431,0.00007146272,0.0001138492,0.0001003381,0.8486264,0.002148358,0.09142429,0.003223593,0.05362611],"study_design_scores_gemma":[0.000003151661,0.000004667423,0.00001325323,0.000003625527,0.000004163746,0.000006212856,0.000002942323,0.9888181,0.0001296037,0.01080011,0.00021143,0.000002638105],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002362933,0.0002399275,0.995257,0.0001528459,0.00004985687,0.00002052767,0.00004734944,0.0001036017,0.001766017],"genre_scores_gemma":[0.427318,0.001213376,0.5573747,0.0004996004,0.0005183368,0.0004645094,0.0004130591,0.0004353712,0.01176314],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006246966,"threshold_uncertainty_score":0.01888704,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4385408288","doi":"10.1007/s00778-023-00805-0","title":"Accelerating directed densest subgraph queries with software and hardware approaches","year":2023,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Complexity and Algorithms in Graphs","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"Basic and Applied Basic Research Foundation of Guangdong Province; Shenzhen Science and Technology Innovation Program; National Natural Science Foundation of China; Chinese University of Hong Kong, Shenzhen","keywords":"Computer science; Speedup; Scalability; Parameterized complexity; Bipartite graph; Computation; Matching (statistics); Approximation algorithm; Linear programming; Theoretical computer science; Graph; Algorithm; Parallel computing; Mathematics","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":"Xiaolin Han","is_ca":false},{"name":"Xiaodong Li","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08005984562599212,"gpt":0.2340146645626377,"spread":0.1539548189366456,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008544116,0.001241475,0.001043776,0.001936379,0.0008161243,0.002622954,0.002646494,0.001299907,0.007845215],"category_scores_gemma":[0.00731454,0.0006017605,0.001106372,0.003240645,0.001065993,0.006005454,0.002742193,0.001487235,0.001648336],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001841696,"about_ca_system_score_gemma":0.00296667,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007010965,"about_ca_topic_score_gemma":0.01849857,"domain_scores_codex":[0.9974148,0.0004848933,0.0001470231,0.0005161988,0.00100393,0.00043308],"domain_scores_gemma":[0.9941327,0.002613923,0.0002986289,0.002129719,0.000632718,0.0001923289],"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":[0.001550609,0.0008144849,0.00670225,0.0007172442,0.00022285,0.000262134,0.0003503669,0.1717995,0.04822946,0.06952446,0.04710183,0.6527249],"study_design_scores_gemma":[0.0001855237,0.0002760362,0.0007450313,0.00002584014,0.00008156097,0.0001818778,0.0001840125,0.9073796,0.01393001,0.0703106,0.006667481,0.00003243486],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1320519,0.001753127,0.820596,0.00239528,0.0005577668,0.000300017,0.0009638707,0.02270293,0.01867904],"genre_scores_gemma":[0.5452585,0.0004325418,0.4461052,0.0005482458,0.0002422913,0.0001860886,0.001579368,0.001001853,0.004645848],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007845215,"threshold_uncertainty_score":0.02624488,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2027705590","doi":"10.1007/s00778-012-0298-2","title":"Recommending XML physical designs for XML databases","year":2012,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"IBM (Canada); University of Waterloo","funders":"","keywords":"Computer science; XML database; Database; Materialized view; Efficient XML Interchange; XML Schema Editor; XML validation; XML; Relational database; Streaming XML; Information retrieval; Database design; XML Schema (W3C); XML framework; View; XML Encryption; World Wide Web","authors":[{"name":"Iman Elghandour","is_ca":false},{"name":"Ashraf Aboulnaga","is_ca":true},{"name":"Daniel C. Zilio","is_ca":true},{"name":"Calisto Zuzarte","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1373293441517697,"gpt":0.3581651089396327,"spread":0.220835764787863,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001569522,0.0009395137,0.0007952247,0.003338238,0.0006690784,0.001956584,0.001348448,0.001556944,0.008550234],"category_scores_gemma":[0.01310229,0.0009382434,0.00107509,0.002475245,0.0006605476,0.003559621,0.001231397,0.0008642491,0.001925968],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007487939,"about_ca_system_score_gemma":0.001061882,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001827055,"about_ca_topic_score_gemma":0.005484872,"domain_scores_codex":[0.9982161,0.0006876831,0.0001563916,0.0003332261,0.0005143314,0.00009227305],"domain_scores_gemma":[0.9949762,0.002455691,0.0002661929,0.0009115022,0.001214884,0.0001755447],"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":[0.001434281,0.0006472085,0.0205264,0.001576802,0.0002922962,0.0009048819,0.0007380189,0.1307358,0.01799577,0.04995587,0.06349501,0.7116977],"study_design_scores_gemma":[0.0002563069,0.0008415674,0.003155101,0.0002401838,0.0002565059,0.0007282255,0.001028119,0.8748084,0.01210252,0.07128881,0.03521596,0.00007832336],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1019003,0.002917527,0.8792844,0.001441366,0.0003724354,0.0005017735,0.002557948,0.002647055,0.008377085],"genre_scores_gemma":[0.5221791,0.001887843,0.4642437,0.0003154625,0.0001324075,0.0004567138,0.004435394,0.00024547,0.006103931],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008550234,"threshold_uncertainty_score":0.02860337,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2026485377","doi":"10.1007/s007780200061","title":"Locating and accessing data repositories with WebSemantics","year":2002,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Canada Research Chairs; University of Toronto","funders":"","keywords":"Metadata; Computer science; World Wide Web; Data element; Interoperability; Data discovery; Data sharing; Metadata repository; Data mapping; Linked data; Semantics (computer science); Data publishing; Information retrieval; Data science; Semantic Web; Publishing; Database","authors":[{"name":"George A. Mihaila","is_ca":true},{"name":"Louiqa Raschid","is_ca":false},{"name":"Anthony Tomasic","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05989600271466623,"gpt":0.2653695745818546,"spread":0.2054735718671883,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004102397,0.001076561,0.001535807,0.006725833,0.002709056,0.00693424,0.00284787,0.001610065,0.006106458],"category_scores_gemma":[0.01470374,0.00150833,0.001357951,0.01015719,0.001723497,0.01219069,0.008094511,0.001826221,0.003638674],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007165212,"about_ca_system_score_gemma":0.00210597,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004597711,"about_ca_topic_score_gemma":0.008527393,"domain_scores_codex":[0.9948502,0.0009491151,0.0007356157,0.0005393928,0.002587179,0.0003385038],"domain_scores_gemma":[0.9892827,0.003245854,0.0006665177,0.005232651,0.001126189,0.0004459831],"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":[0.001353139,0.0006564778,0.02404364,0.0008697966,0.0003143213,0.001647012,0.004372249,0.008461107,0.02896994,0.07657515,0.06245535,0.7902818],"study_design_scores_gemma":[0.0004765501,0.0005964045,0.01252316,0.0006425648,0.0006170788,0.005165188,0.005599478,0.3690808,0.1098156,0.2183112,0.2767549,0.0004170946],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08839287,0.001108865,0.8028378,0.00132935,0.0002311942,0.0006216089,0.001244034,0.08112134,0.02311292],"genre_scores_gemma":[0.3168527,0.001126999,0.6528541,0.0004948357,0.000180834,0.0005134293,0.008023914,0.005559042,0.01439417],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00693424,"threshold_uncertainty_score":0.02169579,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2089760975","doi":"10.1007/s00778-004-0122-8","title":"Views as first-class citizens in object-oriented databases","year":2004,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Class hierarchy; Hierarchy; Computer science; Class (philosophy); Schema (genetic algorithms); Object (grammar); Set (abstract data type); Schema evolution; Theoretical computer science; Human–computer interaction; Object-oriented programming; Programming language; Artificial intelligence; Database; Information retrieval; Database schema","authors":[{"name":"Reda Alhajj","is_ca":true},{"name":"Faruk Polat","is_ca":false},{"name":"Cem Y�lmaz","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02986945958720896,"gpt":0.2845440588140477,"spread":0.2546745992268387,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006009543,0.0004476232,0.001585016,0.001761082,0.002875814,0.01509618,0.001999564,0.003005087,0.003106826],"category_scores_gemma":[0.01503905,0.001656742,0.001399186,0.002580704,0.005039589,0.02267987,0.004151342,0.003866687,0.001131134],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001825939,"about_ca_system_score_gemma":0.002179459,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003113034,"about_ca_topic_score_gemma":0.002986669,"domain_scores_codex":[0.9954916,0.00112892,0.0004489597,0.000871233,0.001352612,0.0007067946],"domain_scores_gemma":[0.9896192,0.004869935,0.0007428275,0.002954433,0.00104943,0.0007641165],"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":[0.0001874251,0.00007430781,0.003423743,0.00008548389,0.00002584312,0.0003014416,0.002911751,0.001302293,0.001573463,0.9486421,0.00336044,0.03811168],"study_design_scores_gemma":[0.00008211709,0.00008163509,0.0008343121,0.0001010196,0.0001614529,0.0005127035,0.001220388,0.01633991,0.006954972,0.917779,0.055866,0.00006647584],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1906174,0.002082254,0.7708657,0.002583402,0.0006559272,0.0001782488,0.0006650659,0.006080424,0.0262716],"genre_scores_gemma":[0.7595986,0.001767042,0.2111701,0.0008205549,0.0005648063,0.0002488263,0.001498009,0.002465409,0.0218667],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01509618,"threshold_uncertainty_score":0.03178191,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4391785473","doi":"10.1007/s00778-024-00835-2","title":"Assisted design of data science pipelines","year":2024,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Berlin Center for Machine Learning; Banting and Best Diabetes Centre, University of Toronto; Technische Universität Berlin","keywords":"Pipeline transport; Computer science; Data science; Engineering; Mechanical engineering","authors":[{"name":"Sergey Redyuk","is_ca":false},{"name":"Zoi Kaoudi","is_ca":false},{"name":"Sebastian Schelter","is_ca":false},{"name":"Volker Markl","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1634969110956813,"gpt":0.3746252582239727,"spread":0.2111283471282914,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005937485,0.001584789,0.001342202,0.002162931,0.00106955,0.002019424,0.004105815,0.001453104,0.008305961],"category_scores_gemma":[0.01376023,0.001152607,0.001595175,0.0009189249,0.001269031,0.002827313,0.004150177,0.001865216,0.003088539],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001411263,"about_ca_system_score_gemma":0.003912994,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003856159,"about_ca_topic_score_gemma":0.004054166,"domain_scores_codex":[0.9965822,0.000901247,0.0002758582,0.0009878473,0.0008891906,0.0003635999],"domain_scores_gemma":[0.9917263,0.003275013,0.0007231186,0.001929477,0.001634809,0.0007112386],"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":[0.001447202,0.00113609,0.009201289,0.001198066,0.0002375984,0.0008000667,0.001189056,0.1545352,0.05841273,0.02511346,0.02159364,0.7251356],"study_design_scores_gemma":[0.0001060839,0.000226016,0.0008622737,0.00004154924,0.00004833666,0.0001440187,0.0001348901,0.9341605,0.03320212,0.01009422,0.02092204,0.00005805137],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0195507,0.0002341662,0.9277346,0.0002323541,0.000037165,0.0007013865,0.0002706219,0.04948231,0.001756619],"genre_scores_gemma":[0.2097425,0.0001126632,0.7849005,0.0001728487,0.00002214534,0.0004454643,0.00103327,0.001501471,0.002069079],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008305961,"threshold_uncertainty_score":0.0314008,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}