{"meta":{"query_hash":"c4fafddebffc","filters":{"venue":"The VLDB Journal"},"cohort_total":59,"direct_labels_cover":0,"predictions_cover":59,"exported":59,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/c4fafddebffc","api":"https://metacan.xera.ac/api/v1/cohort?venue=The+VLDB+Journal"},"results":[{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.03874119627521677,"score_gpt":0.23404084669650793,"score_spread":0.19529965042129116,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1974932383","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25669616,0.006606243,0.68890494,0.0017985628,0.0008846679,0.00033452222,0.0015329267,0.026852395,0.016389543],"genre_scores_gemma":[0.82102734,0.00081101584,0.16929458,0.00037356946,0.00019446142,0.000101651895,0.0016670208,0.00055457506,0.0059757694],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985096,0.0002729014,0.00013448246,0.00019287562,0.00059392885,0.00029619268],"domain_scores_gemma":[0.99589145,0.00065323577,0.00021213284,0.0020951512,0.0008660267,0.00028194068],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020302914,0.0007935461,0.0010071717,0.0009861566,0.0013944601,0.0029699013,0.0046858834,0.0009739334,0.0030159599],"category_scores_gemma":[0.0046704374,0.0006873924,0.0004915883,0.002683265,0.0009247199,0.0040688156,0.002288194,0.0015669349,0.00079229334],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0054309196,0.0009791035,0.016728764,0.00052062864,0.00050877203,0.00076120207,0.0005012627,0.3125937,0.05275954,0.077780835,0.0915332,0.43990204],"study_design_scores_gemma":[0.00020514407,0.00021529404,0.0010101206,0.000020236415,0.000108633816,0.00021435737,0.00013059266,0.95847,0.014169549,0.018347273,0.007053705,0.000055154836],"about_ca_topic_score_codex":0.009194011,"about_ca_topic_score_gemma":0.021958452,"teacher_disagreement_score":0.009194011,"about_ca_system_score_codex":0.0013343139,"about_ca_system_score_gemma":0.0032214306,"threshold_uncertainty_score":0.018280983},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Schema (genetic algorithms); Rewriting; Conceptual schema; Theoretical computer science; Semantics (computer science); Information retrieval; Programming language","score_opus":0.05595234486683632,"score_gpt":0.2594263878383082,"score_spread":0.20347404297147187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1977320861","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13308804,0.0006537027,0.84907544,0.002090947,0.0004448183,0.00023239636,0.0010409673,0.0052226717,0.008151084],"genre_scores_gemma":[0.74641997,0.0007839896,0.23969409,0.0011605365,0.00045633063,0.00021618238,0.0025876383,0.0022303306,0.0064510256],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97954583,0.0072160144,0.002198812,0.0039053292,0.005715683,0.0014183336],"domain_scores_gemma":[0.89628303,0.031249942,0.004375931,0.059263263,0.0078027067,0.0010250647],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013001611,0.0011294866,0.0025903918,0.002770867,0.0022696801,0.008142119,0.004986068,0.003247301,0.0033163733],"category_scores_gemma":[0.06665141,0.0024098516,0.0022749354,0.0065543465,0.0040118303,0.021302035,0.008825484,0.006619938,0.0013159912],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024640765,0.0006779814,0.014082756,0.0009997725,0.001072475,0.0025622037,0.0039679985,0.085083514,0.037762903,0.41063958,0.013976589,0.42671016],"study_design_scores_gemma":[0.00027160725,0.00021078889,0.00206447,0.00013748195,0.0009361424,0.0014253742,0.0012369897,0.16419643,0.065189905,0.74642664,0.017769031,0.00013520085],"about_ca_topic_score_codex":0.0025749956,"about_ca_topic_score_gemma":0.0019844472,"teacher_disagreement_score":0.013001611,"about_ca_system_score_codex":0.0011956526,"about_ca_system_score_gemma":0.003784974,"threshold_uncertainty_score":0.06875986},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; 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","score_opus":0.01980670494016319,"score_gpt":0.2688788118537077,"score_spread":0.24907210691354453,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1988668397","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017008528,0.007450273,0.9620278,0.0022558053,0.00021629011,0.00014846728,0.0017987369,0.0037515324,0.005342662],"genre_scores_gemma":[0.3203615,0.012326392,0.6507199,0.00053749594,0.0007260865,0.00040962858,0.006692308,0.00085629564,0.0073704296],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.995899,0.0015210522,0.00061132107,0.00053811754,0.0012395278,0.00019092529],"domain_scores_gemma":[0.9939814,0.0033880812,0.0005122092,0.0012474614,0.0006882609,0.00018248093],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0062094964,0.0008504153,0.0011465591,0.0033310333,0.0008235184,0.007952701,0.0035226904,0.00188221,0.0030003067],"category_scores_gemma":[0.0148278875,0.0010706129,0.001386147,0.007056882,0.0017812762,0.012049731,0.0013911374,0.0016035532,0.0013076545],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003808387,0.00028042294,0.0056759226,0.0008557687,0.0001243745,0.000439146,0.0010801698,0.11784529,0.00411377,0.5937322,0.020431349,0.25504076],"study_design_scores_gemma":[0.000045407145,0.00009833643,0.00096645684,0.0001606011,0.00012432146,0.0006240569,0.00033326467,0.6637654,0.004334326,0.28496847,0.04450912,0.000070272785],"about_ca_topic_score_codex":0.011803828,"about_ca_topic_score_gemma":0.0075916336,"teacher_disagreement_score":0.011803828,"about_ca_system_score_codex":0.0020443061,"about_ca_system_score_gemma":0.0022571438,"threshold_uncertainty_score":0.032839358},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; 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","score_opus":0.47988074415434634,"score_gpt":0.47707972753851563,"score_spread":0.0028010166158307093,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1995541812","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033633895,0.0003936307,0.95449805,0.0005678157,0.00009633327,0.00020148454,0.001916215,0.0075104674,0.0011820814],"genre_scores_gemma":[0.18324225,0.000406788,0.80930066,0.00014844302,0.00009252705,0.00022997949,0.0046754754,0.0009120769,0.0009918391],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9881392,0.002553105,0.0012740368,0.0021081758,0.005480946,0.00044451372],"domain_scores_gemma":[0.93718404,0.034241896,0.0029996645,0.019050362,0.005741648,0.00078225485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011814,0.001015899,0.0021311915,0.005695032,0.0014786258,0.006886034,0.006783619,0.0022362608,0.0035795625],"category_scores_gemma":[0.07379757,0.0025108473,0.0029797207,0.00944797,0.0014225324,0.011774803,0.007296313,0.0022798423,0.0015519845],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010342164,0.00043809295,0.016092433,0.0010479343,0.0007365667,0.0022956277,0.0019051025,0.16309053,0.0124956705,0.079477556,0.022162318,0.6992239],"study_design_scores_gemma":[0.00010793951,0.00016016632,0.0016092645,0.0001122813,0.0003219376,0.0017488278,0.00056356966,0.82751906,0.02251518,0.12825261,0.016961262,0.00012782245],"about_ca_topic_score_codex":0.002422068,"about_ca_topic_score_gemma":0.0034706858,"teacher_disagreement_score":0.011814,"about_ca_system_score_codex":0.001218541,"about_ca_system_score_gemma":0.0030402478,"threshold_uncertainty_score":0.06247914},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; 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","score_opus":0.020176654432361135,"score_gpt":0.26893277961267065,"score_spread":0.2487561251803095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1999968379","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052246884,0.0007239908,0.90245265,0.00076687796,0.0006949221,0.00020437443,0.0006936922,0.033837155,0.008379376],"genre_scores_gemma":[0.42196855,0.00034519492,0.56570524,0.0003026213,0.0004156815,0.00022213891,0.001879816,0.0018680898,0.0072926795],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9966425,0.00078180694,0.0003015049,0.0005887366,0.0013388112,0.0003465448],"domain_scores_gemma":[0.9837992,0.0055707395,0.000496161,0.006867094,0.00231849,0.00094824075],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043046186,0.0012525951,0.0022546174,0.002089137,0.0016196512,0.003337681,0.0029954354,0.0014735726,0.01048976],"category_scores_gemma":[0.018938025,0.0015393625,0.0009351251,0.0033979185,0.0012382346,0.006386018,0.0044755023,0.0022327937,0.0022083342],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008996285,0.0010104608,0.005625395,0.00050151005,0.00039044512,0.0014511453,0.0012863266,0.10324994,0.0805121,0.099446505,0.05910219,0.6384276],"study_design_scores_gemma":[0.00034541613,0.00028252238,0.0005153343,0.00004502426,0.00007915028,0.00039048295,0.00012575509,0.9110613,0.025496865,0.043921914,0.01765995,0.00007632569],"about_ca_topic_score_codex":0.0024266925,"about_ca_topic_score_gemma":0.0023682772,"teacher_disagreement_score":0.01048976,"about_ca_system_score_codex":0.0013606164,"about_ca_system_score_gemma":0.0017330601,"threshold_uncertainty_score":0.035091758},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Online transaction processing; Transaction processing; Transaction processing system; Scalability; Distributed transaction; Distributed computing; Database transaction; Concurrency; Thread (computing); Compensating transaction; Operating system; Database","score_opus":0.010700441848596995,"score_gpt":0.2297111019525497,"score_spread":0.2190106601039527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2002220286","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1827174,0.002779995,0.7932744,0.0024803123,0.00070859387,0.00028336985,0.00014938887,0.005859603,0.011746889],"genre_scores_gemma":[0.833163,0.0006302186,0.15682341,0.0007536038,0.00042904916,0.0002257582,0.00020049422,0.00084043125,0.006933896],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.989915,0.0035112286,0.00070665136,0.00080027524,0.0038246855,0.0012420975],"domain_scores_gemma":[0.9593445,0.016217167,0.002244409,0.017136479,0.004272508,0.000784993],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050586015,0.0008664032,0.0012831112,0.0013141874,0.0030625197,0.0028554518,0.0046349643,0.0019687638,0.004516142],"category_scores_gemma":[0.02680689,0.0012954039,0.0007330774,0.0025428196,0.0032507295,0.008164631,0.004316665,0.003966322,0.0013718226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028226261,0.0011173789,0.008083401,0.0008209459,0.00015075563,0.00055311417,0.0018818404,0.12415482,0.028460614,0.17732315,0.01733802,0.6372934],"study_design_scores_gemma":[0.000300485,0.0006723965,0.0011926524,0.00012844923,0.00024257314,0.00042693884,0.0006449567,0.6871874,0.05421243,0.23705265,0.017837266,0.00010180201],"about_ca_topic_score_codex":0.0019371207,"about_ca_topic_score_gemma":0.0034191941,"teacher_disagreement_score":0.0050586015,"about_ca_system_score_codex":0.0013057404,"about_ca_system_score_gemma":0.0043217037,"threshold_uncertainty_score":0.02675277},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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)","score_opus":0.029799517877208973,"score_gpt":0.2431261256838491,"score_spread":0.21332660780664012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2006326991","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13395894,0.0040721004,0.83197397,0.00052235246,0.0004382265,0.000553487,0.0020836492,0.009410014,0.016987178],"genre_scores_gemma":[0.32676342,0.002777432,0.63348377,0.00050434604,0.00035956196,0.00027100169,0.008938349,0.001577201,0.025324961],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989386,0.000092940434,0.00006111713,0.00023258131,0.00053966936,0.0001350417],"domain_scores_gemma":[0.99906284,0.00014695931,0.00007211367,0.00044465222,0.0002088249,0.0000646551],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036743138,0.0012566868,0.0027902394,0.004004979,0.0011084799,0.0024226785,0.0023721275,0.0019054317,0.010663493],"category_scores_gemma":[0.0025470715,0.0008038718,0.0016893221,0.0058268085,0.0007705652,0.0025715407,0.002794491,0.0010110292,0.008870185],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010307726,0.00023046303,0.0009857308,0.00042054272,0.00015050588,0.00058994564,0.00018075247,0.013199075,0.21243562,0.009003935,0.01983851,0.7419342],"study_design_scores_gemma":[0.00035745776,0.0008037992,0.004907127,0.00010095809,0.00042853592,0.006689903,0.00074777706,0.6472642,0.25497672,0.038614847,0.04495137,0.00015739673],"about_ca_topic_score_codex":0.0025808515,"about_ca_topic_score_gemma":0.004292394,"teacher_disagreement_score":0.010663493,"about_ca_system_score_codex":0.0008087635,"about_ca_system_score_gemma":0.00092474546,"threshold_uncertainty_score":0.035672903},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.04986115690612704,"score_gpt":0.32384594359330166,"score_spread":0.27398478668717463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012848182","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16858363,0.0030971472,0.8067501,0.0046329326,0.00048925355,0.00036472038,0.005317983,0.0055292714,0.0052349167],"genre_scores_gemma":[0.6873715,0.0013158369,0.29714623,0.0006839785,0.00090490154,0.00014594123,0.008768324,0.0005362898,0.00312701],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98735166,0.0033253543,0.0013591655,0.0017074202,0.0052174386,0.0010389147],"domain_scores_gemma":[0.9317869,0.05248009,0.0025351318,0.0075056003,0.0045707333,0.0011215868],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008457639,0.0016923775,0.004651676,0.0030702369,0.002000741,0.008426222,0.0045914548,0.00370967,0.0046893433],"category_scores_gemma":[0.054085717,0.0013751837,0.0017379492,0.007149325,0.0014943484,0.017354956,0.0040271785,0.0032918903,0.0016107337],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033913173,0.0010921528,0.013101767,0.0027254391,0.0006572658,0.0026059775,0.0016166144,0.39401138,0.020698128,0.07969754,0.050311036,0.43009138],"study_design_scores_gemma":[0.000096647076,0.00019117535,0.00050887995,0.000052467385,0.00010251575,0.00048714294,0.00043582983,0.8932379,0.0052572573,0.09556155,0.004028921,0.00003970831],"about_ca_topic_score_codex":0.0047969674,"about_ca_topic_score_gemma":0.007261544,"teacher_disagreement_score":0.008457639,"about_ca_system_score_codex":0.0012163868,"about_ca_system_score_gemma":0.0025015671,"threshold_uncertainty_score":0.044728816},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Reachability; Graph database; Graph; Factor-critical graph; Theoretical computer science; Web search query; Data mining; Information retrieval; Line graph; Search engine; Voltage graph","score_opus":0.03072226293727882,"score_gpt":0.25766782772861585,"score_spread":0.22694556479133704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2014871820","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34951276,0.0036500718,0.61875135,0.0059360024,0.00026413903,0.00045873807,0.007378902,0.010546626,0.0035013866],"genre_scores_gemma":[0.6912882,0.0012624451,0.29055774,0.00080188905,0.0002860343,0.00016034194,0.012320364,0.0005903595,0.0027326024],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.995567,0.0010197579,0.0005335532,0.0013295831,0.0012634756,0.00028650195],"domain_scores_gemma":[0.98455566,0.00989353,0.001134136,0.0030773552,0.000910357,0.00042907428],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018422508,0.0013125669,0.003172606,0.0043951753,0.0012187632,0.00486026,0.0036305941,0.0030884908,0.0026757251],"category_scores_gemma":[0.016625063,0.0012586752,0.0016734191,0.0075049708,0.0012119821,0.016231354,0.0036378028,0.0024097404,0.0010908253],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030325863,0.002043365,0.028870488,0.0022560179,0.0010062588,0.0012180138,0.0014565564,0.2765237,0.02825109,0.0500709,0.050787818,0.5544832],"study_design_scores_gemma":[0.00015327049,0.00017120538,0.001330664,0.000035803358,0.00017704697,0.00044386208,0.0007721808,0.83567864,0.005777818,0.15219639,0.0032234397,0.00003972288],"about_ca_topic_score_codex":0.004410752,"about_ca_topic_score_gemma":0.007966775,"teacher_disagreement_score":0.00486026,"about_ca_system_score_codex":0.0012049072,"about_ca_system_score_gemma":0.0014013929,"threshold_uncertainty_score":0.009742856},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Dependency (UML); Functional dependency; Data mining; Sampling (signal processing); Class (philosophy); Space (punctuation); Variety (cybernetics); Semantics (computer science); Artificial intelligence; Relational database; Programming language","score_opus":0.30258014556785606,"score_gpt":0.4077661213846789,"score_spread":0.10518597581682282,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2021255304","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6686205,0.0004916348,0.3253549,0.00056091434,0.00014150928,0.00024125386,0.0012925217,0.0018257515,0.0014710546],"genre_scores_gemma":[0.9538024,0.00008114153,0.04148501,0.00014722286,0.00005922643,0.00012558614,0.0029486415,0.00019324712,0.0011574405],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99150085,0.0038537104,0.00056593976,0.0017166586,0.0015708097,0.0007920372],"domain_scores_gemma":[0.8830783,0.08502354,0.0049302857,0.019526681,0.005538608,0.0019026343],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010535745,0.0008267699,0.0018141969,0.0020200307,0.00068760046,0.0011631215,0.0036523633,0.0019327428,0.0031611235],"category_scores_gemma":[0.08984218,0.000849051,0.0014346315,0.0016567705,0.0014307336,0.002351098,0.0021318304,0.0026005704,0.00057097],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005039938,0.0012816082,0.22757772,0.0007922354,0.00078074326,0.0018505696,0.0014590302,0.37239835,0.00941426,0.033465374,0.017068949,0.32887125],"study_design_scores_gemma":[0.00013773656,0.0002794929,0.008612437,0.000052063766,0.00013541726,0.0003793274,0.0003094088,0.9500205,0.0034510365,0.035258636,0.0013320808,0.000031848533],"about_ca_topic_score_codex":0.00518581,"about_ca_topic_score_gemma":0.007995241,"teacher_disagreement_score":0.010535745,"about_ca_system_score_codex":0.00074382406,"about_ca_system_score_gemma":0.0021854548,"threshold_uncertainty_score":0.055719018},"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.023451583210226296,"score_gpt":0.22292722325460765,"score_spread":0.19947564004438134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2024336226","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008013211,0.0005881589,0.9888158,0.00025701817,0.00006814592,0.000027334965,0.00024013226,0.0005421258,0.0014480939],"genre_scores_gemma":[0.4229415,0.0019457036,0.56336635,0.0005369673,0.0006358157,0.00021717571,0.0021247345,0.0007606627,0.0074710706],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99895346,0.0002712279,0.000050180155,0.00019490722,0.00042391164,0.00010633052],"domain_scores_gemma":[0.9958961,0.002351341,0.00024268393,0.0006529803,0.000651457,0.00020548505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013886852,0.00042422107,0.0012904119,0.0021037993,0.00063211337,0.0019057882,0.0017718382,0.0010319719,0.006366499],"category_scores_gemma":[0.0070232595,0.00059064286,0.0006776342,0.0022365702,0.0010089965,0.004397087,0.0019381212,0.0017505167,0.0014640111],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007357134,0.00013160614,0.0021949697,0.00038191845,0.00011609937,0.00016680613,0.00016134011,0.12949763,0.015868545,0.3271496,0.012955957,0.51063985],"study_design_scores_gemma":[0.000054030475,0.00009874269,0.0010885107,0.000031812957,0.00003216869,0.00015174624,0.00004149525,0.76585686,0.0049499893,0.22130325,0.0063303183,0.000061004004],"about_ca_topic_score_codex":0.0016941306,"about_ca_topic_score_gemma":0.0012527622,"teacher_disagreement_score":0.006366499,"about_ca_system_score_codex":0.0007442914,"about_ca_system_score_gemma":0.00086419436,"threshold_uncertainty_score":0.02129805},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.059896002714666226,"score_gpt":0.26536957458185456,"score_spread":0.20547357186718834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2026485377","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08839287,0.0011088651,0.8028378,0.00132935,0.00023119418,0.0006216089,0.0012440339,0.08112134,0.023112923],"genre_scores_gemma":[0.3168527,0.0011269987,0.6528541,0.0004948357,0.00018083402,0.00051342935,0.008023914,0.0055590416,0.014394166],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9948502,0.00094911514,0.0007356157,0.0005393928,0.002587179,0.00033850377],"domain_scores_gemma":[0.9892827,0.003245854,0.0006665177,0.005232651,0.0011261887,0.00044598305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004102397,0.0010765612,0.0015358069,0.0067258333,0.0027090556,0.00693424,0.0028478696,0.0016100652,0.0061064577],"category_scores_gemma":[0.014703744,0.0015083302,0.001357951,0.01015719,0.0017234973,0.0121906875,0.008094511,0.001826221,0.0036386738],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013531387,0.00065647776,0.024043638,0.0008697966,0.0003143213,0.0016470123,0.004372249,0.008461107,0.028969942,0.07657515,0.06245535,0.79028183],"study_design_scores_gemma":[0.00047655008,0.0005964045,0.012523165,0.00064256485,0.0006170788,0.0051651876,0.005599478,0.36908078,0.10981565,0.21831119,0.27675486,0.00041709462],"about_ca_topic_score_codex":0.0045977114,"about_ca_topic_score_gemma":0.008527393,"teacher_disagreement_score":0.00693424,"about_ca_system_score_codex":0.0007165212,"about_ca_system_score_gemma":0.0021059697,"threshold_uncertainty_score":0.021695793},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.05671574245012417,"score_gpt":0.29787606197723465,"score_spread":0.24116031952711048,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2026808466","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14567873,0.009827557,0.82456666,0.0055107255,0.0006015592,0.0002959726,0.0047082407,0.0028998305,0.005910596],"genre_scores_gemma":[0.7564956,0.0036768364,0.22271395,0.0006690067,0.0015730246,0.00018425041,0.008271473,0.00041170383,0.006004146],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99091154,0.002909408,0.00091298047,0.0010705506,0.0036422561,0.0005532038],"domain_scores_gemma":[0.97748417,0.016532991,0.0010487298,0.0026723703,0.0016830913,0.0005786863],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00537132,0.0012133244,0.0033434615,0.0041478346,0.0011467645,0.0055527734,0.002323704,0.0022199843,0.0039913794],"category_scores_gemma":[0.025816819,0.0007750656,0.0011490899,0.007318537,0.0011054329,0.0085172765,0.0020058956,0.0021636682,0.0013419776],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002624922,0.0005547741,0.009920379,0.002154478,0.0006211415,0.0014459379,0.0006204713,0.31488052,0.012539878,0.11517542,0.060696434,0.4787657],"study_design_scores_gemma":[0.00007718947,0.00024320906,0.00091258954,0.00006372145,0.00013646983,0.000604653,0.0002309345,0.83349586,0.004217181,0.15223831,0.0077325525,0.000047374535],"about_ca_topic_score_codex":0.002119123,"about_ca_topic_score_gemma":0.002882676,"teacher_disagreement_score":0.0055527734,"about_ca_system_score_codex":0.0014085786,"about_ca_system_score_gemma":0.0011416429,"threshold_uncertainty_score":0.02840656},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.13732934415176967,"score_gpt":0.35816510893963266,"score_spread":0.220835764787863,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2027705590","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10190032,0.0029175265,0.8792844,0.001441366,0.00037243543,0.0005017735,0.002557948,0.0026470546,0.0083770845],"genre_scores_gemma":[0.52217907,0.0018878435,0.46424368,0.00031546247,0.00013240748,0.00045671375,0.004435394,0.00024547,0.0061039305],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982161,0.00068768306,0.00015639157,0.00033322608,0.00051433145,0.00009227305],"domain_scores_gemma":[0.99497616,0.0024556906,0.00026619292,0.0009115022,0.0012148838,0.00017554474],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001569522,0.0009395137,0.0007952247,0.003338238,0.00066907844,0.001956584,0.0013484483,0.0015569445,0.008550234],"category_scores_gemma":[0.013102292,0.0009382434,0.00107509,0.002475245,0.0006605476,0.0035596208,0.001231397,0.0008642491,0.0019259682],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014342812,0.00064720854,0.0205264,0.0015768015,0.0002922962,0.00090488186,0.0007380189,0.13073581,0.01799577,0.049955867,0.06349501,0.7116977],"study_design_scores_gemma":[0.0002563069,0.00084156735,0.003155101,0.00024018384,0.00025650594,0.0007282255,0.0010281195,0.87480843,0.012102515,0.07128881,0.035215963,0.000078323355],"about_ca_topic_score_codex":0.0018270549,"about_ca_topic_score_gemma":0.0054848716,"teacher_disagreement_score":0.008550234,"about_ca_system_score_codex":0.00074879394,"about_ca_system_score_gemma":0.0010618824,"threshold_uncertainty_score":0.028603375},"labels":[],"label_agreement":null},{"id":"W2029487844","doi":"10.1007/s00778-010-0184-8","title":"Efficient evaluation of generalized tree-pattern queries on XML streams","year":2010,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"XPath; Computer science; XML; Scalability; Query language; Fragment (logic); Spatial query; Theoretical computer science; Query optimization; Tree (set theory); Focus (optics); Node (physics); Time complexity; Data mining; XML database; Information retrieval; Algorithm; Web query classification; Web search query; Database; Search engine; Mathematics","score_opus":0.02671680487953133,"score_gpt":0.2935799402125482,"score_spread":0.26686313533301687,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2029487844","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4580931,0.0018364522,0.51040405,0.0007565229,0.00025618883,0.00042904928,0.0022300237,0.02071383,0.00528082],"genre_scores_gemma":[0.79427904,0.00044005577,0.1959776,0.000219876,0.0001657669,0.00015830777,0.0036216371,0.0011555803,0.0039821523],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958715,0.00077417574,0.00037883333,0.0005125627,0.0020047773,0.00045810867],"domain_scores_gemma":[0.99434257,0.0030291048,0.00034185167,0.0010461418,0.0009890298,0.00025122013],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024268818,0.0012408054,0.0023985489,0.0014573704,0.00069444114,0.0029651031,0.0020632716,0.0011252402,0.003595418],"category_scores_gemma":[0.007969531,0.0005748042,0.00082604314,0.0026737994,0.00077724596,0.0042919153,0.0022730138,0.00090458465,0.0008253916],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.010309126,0.00061944366,0.0142402295,0.0010628671,0.00044966486,0.0016745393,0.0009585936,0.12881376,0.10375737,0.029076828,0.029679341,0.67935824],"study_design_scores_gemma":[0.0003347284,0.0003425186,0.0019008224,0.000027643946,0.000098006116,0.00036453927,0.00030090712,0.932673,0.038644392,0.021587238,0.0036868672,0.000039429757],"about_ca_topic_score_codex":0.004315688,"about_ca_topic_score_gemma":0.0063166,"teacher_disagreement_score":0.004315688,"about_ca_system_score_codex":0.0012699014,"about_ca_system_score_gemma":0.001485303,"threshold_uncertainty_score":0.012834728},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.019450964385993072,"score_gpt":0.24129503545532632,"score_spread":0.22184407106933324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2035006988","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44248486,0.0029142096,0.5424862,0.0006114672,0.0003880504,0.00023390295,0.001090259,0.0050558276,0.0047351937],"genre_scores_gemma":[0.81416327,0.0007515874,0.17857966,0.000100362806,0.00029428795,0.0001425562,0.0016860162,0.00030906184,0.00397324],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982241,0.00029887442,0.00020763862,0.00029708724,0.0007452429,0.00022697976],"domain_scores_gemma":[0.9972784,0.0012372199,0.00019112047,0.0006561905,0.0004107706,0.00022633543],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026076846,0.000689307,0.0019312312,0.0015567562,0.0012600704,0.00336873,0.0014707056,0.0008026635,0.0030505732],"category_scores_gemma":[0.005215438,0.0005887022,0.00061546045,0.0027275002,0.0004983344,0.003812851,0.0019567981,0.00086087687,0.0006960512],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0130821355,0.00088109093,0.019809823,0.00066544965,0.00057140994,0.0012930846,0.0014558674,0.08008959,0.16143195,0.033519376,0.019856963,0.6673433],"study_design_scores_gemma":[0.0003915028,0.0009022259,0.0044267676,0.000029628301,0.00018216505,0.00070007035,0.0006348787,0.892761,0.06272402,0.027597286,0.0095927855,0.000057691344],"about_ca_topic_score_codex":0.0013754318,"about_ca_topic_score_gemma":0.0019389829,"teacher_disagreement_score":0.00336873,"about_ca_system_score_codex":0.00045648473,"about_ca_system_score_gemma":0.00080369675,"threshold_uncertainty_score":0.013790965},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02968092915324465,"score_gpt":0.24349118944614273,"score_spread":0.21381026029289807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2040374677","genre_codex":"software","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015473803,0.0011891719,0.3010935,0.0009803518,0.0007352425,0.0005334383,0.02184816,0.60033053,0.05781583],"genre_scores_gemma":[0.19683675,0.0022569122,0.39191404,0.0029009588,0.0005460267,0.0010649745,0.14005914,0.106766865,0.15765427],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989785,0.00011435006,0.00014141723,0.00018629052,0.0004653207,0.00011420033],"domain_scores_gemma":[0.9983242,0.00031779514,0.00007606765,0.00068349985,0.00039661667,0.00020179339],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001711137,0.00091522874,0.0008096745,0.0016800401,0.00081985036,0.0029518509,0.0033356862,0.0011155099,0.041982993],"category_scores_gemma":[0.0041506425,0.0009285269,0.0006353547,0.0017573026,0.00067813107,0.0051895455,0.0026393612,0.0015912486,0.026769951],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013673331,0.00041467423,0.0027853032,0.0007249621,0.00013691175,0.000646078,0.0005836682,0.002747043,0.028063811,0.035970483,0.629412,0.29714775],"study_design_scores_gemma":[0.0007072585,0.0002653515,0.0030055721,0.00018764487,0.00019598717,0.0014446724,0.0004483227,0.049103215,0.041971777,0.04970949,0.8527276,0.00023311705],"about_ca_topic_score_codex":0.0028970826,"about_ca_topic_score_gemma":0.003936544,"teacher_disagreement_score":0.041982993,"about_ca_system_score_codex":0.00047863182,"about_ca_system_score_gemma":0.0013399178,"threshold_uncertainty_score":0.14044708},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.017376135837386203,"score_gpt":0.24773613173478778,"score_spread":0.23035999589740158,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2046474027","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5713867,0.0023271185,0.40812883,0.0010149264,0.000328427,0.0002206025,0.0007075186,0.013702738,0.0021830986],"genre_scores_gemma":[0.94798315,0.00022408592,0.05054151,0.000093398565,0.000048436013,0.000030491889,0.00043325606,0.0001653124,0.0004802577],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99321884,0.0010143804,0.0006081504,0.0010924307,0.0035104323,0.00055583037],"domain_scores_gemma":[0.9679357,0.0102036195,0.007712883,0.007344904,0.0060231425,0.000779603],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035761984,0.0006783207,0.0010071377,0.0030069877,0.00080768333,0.002079629,0.0020933005,0.0013557348,0.0007172665],"category_scores_gemma":[0.020715555,0.00069086533,0.0005507961,0.0021866725,0.0008296196,0.0024057934,0.0016750578,0.0017211379,0.00033842452],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016310168,0.00067264726,0.4160343,0.0005142884,0.0003467167,0.0010872522,0.00076457235,0.059183374,0.08937742,0.011401771,0.008194437,0.41079214],"study_design_scores_gemma":[0.00008068988,0.0003392546,0.04792956,0.000067374895,0.00023079326,0.0018716864,0.00035873835,0.8731759,0.058293078,0.014487258,0.0030818027,0.00008382606],"about_ca_topic_score_codex":0.0034066525,"about_ca_topic_score_gemma":0.0058659227,"teacher_disagreement_score":0.0035761984,"about_ca_system_score_codex":0.00091901317,"about_ca_system_score_gemma":0.0023572473,"threshold_uncertainty_score":0.01891303},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.009606939193727806,"score_gpt":0.23790982533144633,"score_spread":0.22830288613771854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2061240327","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040062345,0.0011808838,0.99328893,0.00017159774,0.00007993594,0.00001994094,0.00006131666,0.00088320667,0.00030794617],"genre_scores_gemma":[0.13967614,0.002535899,0.8541574,0.00009179443,0.0003632563,0.00012421941,0.0005627967,0.00027945108,0.0022090233],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9974463,0.0005715375,0.00020046195,0.00044726356,0.0012377782,0.0000966532],"domain_scores_gemma":[0.9940037,0.002686643,0.00066082465,0.00068706437,0.0017876143,0.00017419153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020910713,0.0014098233,0.00232931,0.0038231777,0.00097032683,0.002238905,0.0032631396,0.0025373877,0.0015349429],"category_scores_gemma":[0.011467537,0.0006284287,0.0007691493,0.007980758,0.001465115,0.0030767734,0.0023344506,0.002692567,0.0015022898],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030024874,0.00016227679,0.003354048,0.00025917616,0.00010254101,0.00012678654,0.00015831023,0.10367799,0.005953399,0.026071036,0.0071506696,0.8526835],"study_design_scores_gemma":[0.00003844649,0.00007562499,0.00081282743,0.000031830674,0.000041286457,0.00035684736,0.00008732695,0.9452694,0.0063670217,0.04053836,0.006332429,0.00004853649],"about_ca_topic_score_codex":0.002487089,"about_ca_topic_score_gemma":0.0015702187,"teacher_disagreement_score":0.0038231777,"about_ca_system_score_codex":0.0007716829,"about_ca_system_score_gemma":0.0007714336,"threshold_uncertainty_score":0.011058748},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; 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","score_opus":0.010352431843665732,"score_gpt":0.22936693184287188,"score_spread":0.21901449999920614,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064580935","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.056309525,0.0019071333,0.867203,0.00254368,0.0007876902,0.0004269332,0.007390821,0.044232056,0.019199098],"genre_scores_gemma":[0.4196128,0.0028207062,0.5065785,0.0014087165,0.00073694385,0.0004413061,0.031239608,0.005327012,0.031834424],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99629384,0.0006468805,0.0004999244,0.0006088804,0.001751194,0.00019926566],"domain_scores_gemma":[0.99545175,0.0020027594,0.00018783392,0.0013873357,0.0007694774,0.00020085176],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036332824,0.0012061204,0.0015915827,0.0027979144,0.0011371651,0.0065316367,0.0022404578,0.001973054,0.0102072805],"category_scores_gemma":[0.013146792,0.00091859564,0.0012877604,0.0035167914,0.0011176803,0.009381784,0.0032969401,0.0018984699,0.0035972875],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002492207,0.00057538936,0.012927976,0.0010704205,0.00038264805,0.0013228157,0.001637983,0.03220011,0.02944381,0.23246281,0.09486795,0.59061587],"study_design_scores_gemma":[0.0003122075,0.00023525824,0.001594074,0.00024204687,0.0002849004,0.0014593849,0.00098274,0.5070152,0.0790376,0.24127679,0.16742958,0.00013017569],"about_ca_topic_score_codex":0.0025575708,"about_ca_topic_score_gemma":0.002188037,"teacher_disagreement_score":0.0102072805,"about_ca_system_score_codex":0.0012041637,"about_ca_system_score_gemma":0.0010828401,"threshold_uncertainty_score":0.034146726},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.01871367238197218,"score_gpt":0.2452819636881325,"score_spread":0.2265682913061603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2065714066","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5108377,0.0062597757,0.45306292,0.0030767578,0.00039485164,0.0004570504,0.0021819498,0.0062534413,0.01747551],"genre_scores_gemma":[0.84324616,0.0006774544,0.15075774,0.0001252694,0.00011718738,0.00010038424,0.0019949963,0.0005094907,0.0024712286],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9900511,0.0040320465,0.00053707504,0.0010529477,0.0033717537,0.00095502],"domain_scores_gemma":[0.9805335,0.014823851,0.0006311806,0.0014573605,0.0020306641,0.0005233619],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0061213155,0.0013907241,0.0023825467,0.0018173221,0.0009320736,0.002972526,0.0026832726,0.0016256046,0.0046550217],"category_scores_gemma":[0.026885184,0.0006165013,0.0007899499,0.0031543016,0.001317867,0.004186526,0.0016333953,0.0011788596,0.0006676768],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0037478204,0.0007408289,0.005166725,0.00060297095,0.00015841024,0.00016955374,0.00018871987,0.72889125,0.007308987,0.020211522,0.012654789,0.22015841],"study_design_scores_gemma":[0.000078749756,0.00011885915,0.0004863992,0.000006346985,0.000021174905,0.00004067344,0.000054842603,0.98996335,0.0018119243,0.0068546985,0.00055496534,0.0000080722075],"about_ca_topic_score_codex":0.012597493,"about_ca_topic_score_gemma":0.011944941,"teacher_disagreement_score":0.012597493,"about_ca_system_score_codex":0.0034607418,"about_ca_system_score_gemma":0.0042011864,"threshold_uncertainty_score":0.03237301},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02434631267444179,"score_gpt":0.2217052823758772,"score_spread":0.1973589697014354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2068157134","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.045243844,0.0008133336,0.91946256,0.0023595088,0.0001743238,0.0004324718,0.0020999196,0.004549313,0.024864826],"genre_scores_gemma":[0.32503852,0.0006863857,0.6233127,0.0010965841,0.00031596454,0.000467711,0.0069009652,0.001764973,0.04041621],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9969784,0.0005933564,0.00019628736,0.0006492062,0.0011958856,0.0003868044],"domain_scores_gemma":[0.9970939,0.0010511606,0.0001341071,0.0010675918,0.0005310765,0.000122149],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013569826,0.0009073167,0.0015924954,0.0014081454,0.00088059914,0.002378471,0.0023335025,0.0011972619,0.017205302],"category_scores_gemma":[0.0058462843,0.00050491554,0.001429913,0.0026742457,0.000725089,0.003728926,0.0024318532,0.002081223,0.004724224],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009952836,0.00062505505,0.0026708478,0.00058710674,0.00021702048,0.00022419705,0.00023088626,0.038999163,0.018517213,0.14419985,0.0741691,0.7185643],"study_design_scores_gemma":[0.00021113799,0.00038426032,0.0014667166,0.00005030237,0.0002626363,0.0011197644,0.0002900104,0.64634645,0.037333544,0.28347582,0.02901756,0.00004182628],"about_ca_topic_score_codex":0.0015441948,"about_ca_topic_score_gemma":0.0027780132,"teacher_disagreement_score":0.017205302,"about_ca_system_score_codex":0.00098216,"about_ca_system_score_gemma":0.0019259216,"threshold_uncertainty_score":0.057557464},"labels":[],"label_agreement":null},{"id":"W2077020471","doi":"10.1007/s00778-005-0179-z","title":"Special issue: Best papers of VLDB 2004","year":2005,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University of Waterloo","funders":"","keywords":"Very large database; Computer science; Data science; Data mining","score_opus":0.014589702295020011,"score_gpt":0.2587781872719452,"score_spread":0.24418848497692516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077020471","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021549922,0.07485902,0.008321708,0.06555553,0.71407217,0.0005572876,0.021022966,0.0032234713,0.11023279],"genre_scores_gemma":[0.008721948,0.058674376,0.0065004476,0.012327414,0.27241293,0.00031653172,0.04004293,0.0024774561,0.59852594],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976109,0.00032915428,0.0002844732,0.00037090929,0.001155444,0.00024907358],"domain_scores_gemma":[0.98408115,0.002626786,0.00080467813,0.0012956235,0.007856855,0.0033349576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033222116,0.0023950043,0.0043371837,0.010889002,0.0020562361,0.014219285,0.0029410867,0.0027878697,0.1888429],"category_scores_gemma":[0.012046314,0.0008465702,0.0013775011,0.011314165,0.00083075167,0.0051495316,0.0019778165,0.0028413665,0.12121795],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006614758,0.000021354912,0.00009013212,0.00015941316,0.000009846363,0.000025114143,0.0000040095347,0.00006974428,0.00013408248,0.0004149133,0.97924316,0.019762157],"study_design_scores_gemma":[0.000036365516,0.00003679166,0.00056666415,0.00018615497,0.000037017362,0.00010200655,0.000017998529,0.00033178052,0.00037411303,0.0014419741,0.9968541,0.000015048147],"about_ca_topic_score_codex":0.003996015,"about_ca_topic_score_gemma":0.006116969,"teacher_disagreement_score":0.1888429,"about_ca_system_score_codex":0.00366304,"about_ca_system_score_gemma":0.003953946,"threshold_uncertainty_score":0.6317424},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.044471574677033665,"score_gpt":0.2691169814537689,"score_spread":0.22464540677673522,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2085472312","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02691211,0.00090111356,0.9624556,0.0023643458,0.00022785673,0.00011618155,0.0005190169,0.0006778636,0.0058258385],"genre_scores_gemma":[0.86010647,0.0016321732,0.12528104,0.0008728536,0.00075457647,0.0002846365,0.0010777019,0.00025549205,0.00973513],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9777625,0.009048078,0.0014727117,0.0029558563,0.007153261,0.001607482],"domain_scores_gemma":[0.9024111,0.047255702,0.004624547,0.040435985,0.004074038,0.001198631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012966242,0.0009974707,0.0029507072,0.0026673414,0.0021775456,0.0071116784,0.003374378,0.0021810636,0.004190474],"category_scores_gemma":[0.066319786,0.0013889406,0.001714032,0.0071279085,0.003798183,0.011822326,0.009424089,0.004169993,0.0017360359],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014966499,0.0002935768,0.00511448,0.00043261098,0.00029536596,0.0009070004,0.0006753049,0.11129498,0.0064049694,0.68493056,0.012038645,0.17611594],"study_design_scores_gemma":[0.000118273725,0.00012288732,0.00047191026,0.00007208764,0.00012634773,0.0008036238,0.00008881495,0.47487733,0.0104183145,0.5046828,0.008164916,0.00005266606],"about_ca_topic_score_codex":0.0005916122,"about_ca_topic_score_gemma":0.0007321251,"teacher_disagreement_score":0.012966242,"about_ca_system_score_codex":0.0026235215,"about_ca_system_score_gemma":0.004045967,"threshold_uncertainty_score":0.06857288},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; 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","score_opus":0.08655948616922705,"score_gpt":0.28821611297236155,"score_spread":0.2016566268031345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2086395677","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72043467,0.0017528486,0.24368863,0.00094274775,0.0006094374,0.0002993814,0.00066831324,0.018971585,0.01263237],"genre_scores_gemma":[0.95880103,0.00012195652,0.038339566,0.00009190742,0.0001373195,0.000053666725,0.0005850564,0.00047354688,0.001395902],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974306,0.0007958046,0.0001865725,0.00032359487,0.00078173535,0.00048164625],"domain_scores_gemma":[0.99339503,0.0023788286,0.0005509715,0.0017061596,0.0011620172,0.00080691214],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019992096,0.0010376243,0.000960589,0.001026904,0.001039119,0.0018448435,0.0028677206,0.0005770706,0.002662172],"category_scores_gemma":[0.0080983825,0.0006900636,0.00041305143,0.0014221133,0.00036072533,0.0014760372,0.0013627842,0.00094069866,0.0008115894],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011424339,0.0023598988,0.023058187,0.0005415319,0.00038449978,0.00100653,0.0017039904,0.26339033,0.2231225,0.010653922,0.04473553,0.4176187],"study_design_scores_gemma":[0.00014260312,0.00048778526,0.004331651,0.000011437146,0.00009833279,0.00018002777,0.00021982494,0.9587099,0.027297089,0.0026872733,0.005780621,0.00005346302],"about_ca_topic_score_codex":0.0030076755,"about_ca_topic_score_gemma":0.0064385813,"teacher_disagreement_score":0.0030076755,"about_ca_system_score_codex":0.000745464,"about_ca_system_score_gemma":0.0022791256,"threshold_uncertainty_score":0.01057297},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02986945958720896,"score_gpt":0.28454405881404765,"score_spread":0.2546745992268387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2089760975","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19061737,0.0020822543,0.77086574,0.0025834015,0.00065592723,0.00017824875,0.00066506595,0.006080424,0.026271597],"genre_scores_gemma":[0.7595986,0.0017670416,0.21117009,0.0008205549,0.00056480634,0.00024882628,0.0014980092,0.0024654085,0.021866696],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99549156,0.0011289204,0.0004489597,0.000871233,0.0013526118,0.0007067946],"domain_scores_gemma":[0.9896192,0.004869935,0.0007428275,0.0029544325,0.0010494305,0.00076411647],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006009543,0.00044762317,0.0015850156,0.0017610824,0.0028758144,0.0150961755,0.0019995635,0.0030050871,0.0031068265],"category_scores_gemma":[0.015039048,0.001656742,0.0013991859,0.0025807042,0.0050395885,0.022679867,0.0041513424,0.0038666867,0.0011311345],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001874251,0.00007430781,0.0034237432,0.000085483895,0.000025843125,0.00030144156,0.0029117512,0.0013022927,0.0015734632,0.94864213,0.00336044,0.038111676],"study_design_scores_gemma":[0.00008211709,0.00008163509,0.0008343121,0.00010101962,0.00016145289,0.0005127035,0.0012203876,0.016339907,0.006954972,0.91777897,0.055866003,0.00006647584],"about_ca_topic_score_codex":0.003113034,"about_ca_topic_score_gemma":0.0029866688,"teacher_disagreement_score":0.0150961755,"about_ca_system_score_codex":0.0018259387,"about_ca_system_score_gemma":0.002179459,"threshold_uncertainty_score":0.03178191},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; 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","score_opus":0.03430227086175861,"score_gpt":0.26878400891824294,"score_spread":0.23448173805648434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2092324613","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033363132,0.0030606273,0.9804765,0.0006342405,0.00023079515,0.000092455375,0.00033775563,0.0047589205,0.0070723137],"genre_scores_gemma":[0.10462277,0.004929546,0.8683001,0.0007808017,0.00093359576,0.0003516893,0.0019071298,0.001864398,0.016309945],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99543875,0.0008560827,0.0005256697,0.00072048017,0.0020235314,0.00043549473],"domain_scores_gemma":[0.9947647,0.002037353,0.0002332329,0.0019192114,0.0009383738,0.00010709499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024668844,0.0013682703,0.0018086237,0.0029300442,0.00203488,0.0053291684,0.0029009902,0.0011948927,0.011534825],"category_scores_gemma":[0.009459683,0.0012049025,0.0019936801,0.0069329003,0.0017547373,0.009747375,0.0033060263,0.0034541965,0.006127321],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048386081,0.00012737073,0.0008632146,0.00091000594,0.00019271807,0.00023084237,0.0010022096,0.010671038,0.0155230565,0.3672738,0.049470432,0.55325145],"study_design_scores_gemma":[0.00014167937,0.00019984301,0.00052650966,0.00019379794,0.0003040991,0.0008597743,0.00046223417,0.13355885,0.029189147,0.67695284,0.15750052,0.00011069983],"about_ca_topic_score_codex":0.0024982851,"about_ca_topic_score_gemma":0.0021155071,"teacher_disagreement_score":0.011534825,"about_ca_system_score_codex":0.0010956445,"about_ca_system_score_gemma":0.0011589554,"threshold_uncertainty_score":0.03858787},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Database; Generator (circuit theory); Set (abstract data type); Feature (linguistics); Test case; Test suite; Test (biology); Data mining; Programming language; Power (physics); Machine learning","score_opus":0.0426218733597992,"score_gpt":0.3124691026543456,"score_spread":0.26984722929454635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099525917","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.053926412,0.00019673392,0.9164899,0.0006941945,0.00010286814,0.0010846247,0.0013967745,0.021263305,0.0048451405],"genre_scores_gemma":[0.29708588,0.000114535236,0.6957729,0.00037606087,0.00004877413,0.0008969759,0.0020695208,0.001531177,0.002104207],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9720072,0.009471626,0.0033174262,0.003379909,0.009620407,0.002203445],"domain_scores_gemma":[0.94463074,0.02869226,0.0027124623,0.015394362,0.007135031,0.0014351928],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017568536,0.002010628,0.0012444503,0.00456622,0.0017077324,0.006557925,0.007544273,0.004018087,0.0055464073],"category_scores_gemma":[0.055266224,0.001865669,0.003094841,0.0019095533,0.0039326493,0.015298515,0.0058092205,0.004642787,0.0013990159],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020552091,0.0054680067,0.050066832,0.0020616713,0.00085731404,0.0016455075,0.003530375,0.096004404,0.07654272,0.34258214,0.026281292,0.39290455],"study_design_scores_gemma":[0.0003871399,0.0017036223,0.0058425767,0.00040112287,0.00042822666,0.00096498034,0.0014244942,0.7316642,0.07261649,0.15375902,0.030484367,0.00032389784],"about_ca_topic_score_codex":0.009674661,"about_ca_topic_score_gemma":0.007729436,"teacher_disagreement_score":0.017568536,"about_ca_system_score_codex":0.0021761057,"about_ca_system_score_gemma":0.0039778696,"threshold_uncertainty_score":0.092912436},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.057340703357166435,"score_gpt":0.3061374292867121,"score_spread":0.2487967259295457,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100257623","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017991383,0.00009473358,0.97568476,0.0011781057,0.00007087906,0.00015181524,0.000050618815,0.0002269584,0.0045507387],"genre_scores_gemma":[0.7691895,0.00023687592,0.22269277,0.00047227836,0.00015405213,0.00043899033,0.00012537304,0.00016246914,0.006527773],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9784581,0.01237374,0.00084695517,0.0021296556,0.004565814,0.0016257546],"domain_scores_gemma":[0.9592054,0.027787007,0.0017844558,0.0072098877,0.002639523,0.0013736384],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014914026,0.0012960859,0.0022976112,0.0012716909,0.0032751067,0.009136052,0.0043512788,0.0040441556,0.0029401863],"category_scores_gemma":[0.037609465,0.0014765539,0.0023728735,0.0015796392,0.0066475235,0.015275417,0.010266185,0.0065316893,0.0009796727],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044161602,0.00013476454,0.000632867,0.00010302284,0.00012038721,0.00029228898,0.0006694301,0.059425075,0.0024451173,0.9135897,0.0021096913,0.02003607],"study_design_scores_gemma":[0.00006408378,0.000050839644,0.00004836168,0.000015196619,0.00004180093,0.00009576361,0.00009708409,0.37133855,0.0016856012,0.62442666,0.002107316,0.00002869491],"about_ca_topic_score_codex":0.0018411715,"about_ca_topic_score_gemma":0.0013367286,"teacher_disagreement_score":0.014914026,"about_ca_system_score_codex":0.002574654,"about_ca_system_score_gemma":0.004648136,"threshold_uncertainty_score":0.07887381},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"","keywords":"Computer science; Wavelet; Query optimization; Data mining; Tuple; Histogram; Algorithm; Artificial intelligence; Mathematics; Image (mathematics)","score_opus":0.0412518917975479,"score_gpt":0.2701677799750619,"score_spread":0.22891588817751402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110557355","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022355795,0.00068945996,0.974838,0.00027325953,0.00012115079,0.000029796667,0.000090928326,0.00071146514,0.00089010736],"genre_scores_gemma":[0.42440933,0.0011854653,0.56917,0.0002105767,0.00037578974,0.00014521516,0.0008243181,0.00026693684,0.0034123156],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99653023,0.00091922,0.00029131278,0.0004194638,0.0015174467,0.0003223026],"domain_scores_gemma":[0.9946956,0.002182281,0.0002738486,0.0017452196,0.00093295757,0.0001701513],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002698724,0.0007566524,0.0024333296,0.0019333527,0.00080616045,0.0029193496,0.0014089892,0.0010848101,0.0027945547],"category_scores_gemma":[0.012947684,0.0007078713,0.0011208508,0.0040124995,0.0010219909,0.0040418915,0.0021986857,0.0020015684,0.0011878127],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018782353,0.00027786853,0.0030081454,0.0002992157,0.00027901892,0.0002583879,0.0003515654,0.25006658,0.034418013,0.113969214,0.012597296,0.5825965],"study_design_scores_gemma":[0.000028691284,0.00006702182,0.00019830924,0.0000073175224,0.000020802765,0.00007342384,0.00004600806,0.9685407,0.0033226814,0.026220975,0.0014624267,0.000011567985],"about_ca_topic_score_codex":0.0024025308,"about_ca_topic_score_gemma":0.0017799778,"teacher_disagreement_score":0.0029193496,"about_ca_system_score_codex":0.0008298015,"about_ca_system_score_gemma":0.001185185,"threshold_uncertainty_score":0.014272392},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.012074452013339535,"score_gpt":0.20327558979719282,"score_spread":0.19120113778385328,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121041488","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010695391,0.0007946964,0.63075984,0.00055894075,0.00021327219,0.00044872632,0.019353226,0.32917285,0.008003093],"genre_scores_gemma":[0.15968014,0.001690638,0.66299134,0.0021510446,0.00014811114,0.0006794217,0.1133544,0.039465837,0.01983907],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982718,0.00023162637,0.00020527282,0.0003370269,0.00082804746,0.00012620226],"domain_scores_gemma":[0.9983897,0.0005805668,0.00009212132,0.0005095821,0.0003362913,0.000091709684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016111932,0.0013683804,0.0015123902,0.0024830147,0.00065959105,0.0024945235,0.0034881982,0.0011899015,0.013706576],"category_scores_gemma":[0.0040408475,0.0010530084,0.001654714,0.0028760869,0.00076810445,0.0048405374,0.0028099942,0.0015807728,0.0064803227],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021567894,0.00067181385,0.0051432294,0.0030892445,0.0010660749,0.00072144694,0.00074672076,0.035445336,0.0642688,0.07019129,0.37372825,0.442771],"study_design_scores_gemma":[0.00074566493,0.00035427578,0.0027085315,0.00024795253,0.00051820866,0.00094188,0.0004834453,0.5010182,0.07639446,0.11062019,0.30558982,0.00037723823],"about_ca_topic_score_codex":0.008267613,"about_ca_topic_score_gemma":0.012660597,"teacher_disagreement_score":0.013706576,"about_ca_system_score_codex":0.0009718442,"about_ca_system_score_gemma":0.0014437593,"threshold_uncertainty_score":0.04585302},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.04612929082107779,"score_gpt":0.2739027125305562,"score_spread":0.2277734217094784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123080243","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06355721,0.0008864255,0.91413784,0.0016228833,0.00022221544,0.00045911514,0.0009818865,0.010725826,0.0074065207],"genre_scores_gemma":[0.51872003,0.0012496229,0.46529642,0.00082439446,0.00014645247,0.0004916566,0.004052925,0.0017399135,0.0074785696],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.992761,0.0029679702,0.00097629486,0.0008142027,0.0020212217,0.00045932995],"domain_scores_gemma":[0.9916053,0.004748578,0.00029886764,0.0022181275,0.0009218721,0.0002072663],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008666717,0.0006861103,0.0011798358,0.0019819944,0.0015291993,0.0061955648,0.0028188266,0.0019948815,0.003615539],"category_scores_gemma":[0.015462165,0.0011479482,0.0014382552,0.003282381,0.0018555155,0.008977483,0.0043622265,0.0019556577,0.0014194092],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021957965,0.000952195,0.008024548,0.001313444,0.00034906328,0.004070913,0.005851466,0.07184397,0.039745275,0.27234504,0.034413014,0.55889535],"study_design_scores_gemma":[0.00034207586,0.0002913464,0.0022723665,0.0002167929,0.0003449768,0.0016597371,0.0034959419,0.60882986,0.11046946,0.22234957,0.049541797,0.00018611719],"about_ca_topic_score_codex":0.004554855,"about_ca_topic_score_gemma":0.002707478,"teacher_disagreement_score":0.008666717,"about_ca_system_score_codex":0.0010436667,"about_ca_system_score_gemma":0.0018066276,"threshold_uncertainty_score":0.04583454},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.04801050073090831,"score_gpt":0.3160797680561167,"score_spread":0.2680692673252084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126431245","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063389577,0.0019542282,0.97920144,0.00089791196,0.00021511008,0.000079900565,0.0003003806,0.0015231587,0.009488952],"genre_scores_gemma":[0.31471542,0.0036274523,0.65375775,0.0011498927,0.002264948,0.00067938,0.0021533633,0.0019801646,0.019671755],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99328953,0.0022342661,0.0004683542,0.0010047501,0.002230083,0.00077310443],"domain_scores_gemma":[0.9749116,0.016975673,0.0010672213,0.0040946384,0.002336722,0.0006141624],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058178934,0.0024986477,0.0021530841,0.005575067,0.0025338058,0.00698099,0.005808397,0.0020489225,0.013877985],"category_scores_gemma":[0.038814344,0.0013354854,0.0028446068,0.0073616947,0.0051719025,0.017230734,0.0055537107,0.0067913216,0.004388889],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031379831,0.00015826907,0.0009700437,0.00032523889,0.00008653073,0.00004961489,0.00022508496,0.02553238,0.0008970145,0.85293186,0.011886465,0.106623635],"study_design_scores_gemma":[0.000019391284,0.00001893421,0.00016515146,0.00004202341,0.000037356618,0.000030408915,0.000035125406,0.092816055,0.000847549,0.9024708,0.0034977805,0.000019297917],"about_ca_topic_score_codex":0.0029166138,"about_ca_topic_score_gemma":0.0029885834,"teacher_disagreement_score":0.013877985,"about_ca_system_score_codex":0.0034800463,"about_ca_system_score_gemma":0.0030176626,"threshold_uncertainty_score":0.046426475},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Semantics (computer science); Tuple; Key (lock); Range (aeronautics); Variation (astronomy); Window (computing); Programming language; Theoretical computer science; World Wide Web; Operating system; Engineering","score_opus":0.019849684280367718,"score_gpt":0.24530657120357555,"score_spread":0.22545688692320784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129088288","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.089366876,0.00021591096,0.9022364,0.000544658,0.00007740187,0.00010239991,0.00048168458,0.0021888867,0.0047857673],"genre_scores_gemma":[0.8408599,0.00041978242,0.15283118,0.00014525816,0.000079847014,0.00012741945,0.00065936014,0.0005459359,0.004331274],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99820614,0.00043523358,0.00018695169,0.00028300536,0.0006142219,0.00027441152],"domain_scores_gemma":[0.9961868,0.0017594147,0.00042659306,0.0008833627,0.0006065163,0.00013744163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026274533,0.0007554023,0.00062664854,0.0010265382,0.0006783857,0.0047507156,0.002361227,0.0013053344,0.002184543],"category_scores_gemma":[0.008479436,0.0008807061,0.0012167013,0.0010806384,0.0019065102,0.006652239,0.0014875067,0.0016320149,0.0005171864],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032072776,0.000109794586,0.0026968273,0.00008796752,0.00005130572,0.0003138934,0.0003753256,0.69861484,0.0032545582,0.27998903,0.0012229562,0.012962742],"study_design_scores_gemma":[0.00001740435,0.00001108843,0.000050246985,0.000007523701,0.000020966085,0.000022831447,0.000023017936,0.9400804,0.0022491438,0.05631048,0.0011992564,0.000007768394],"about_ca_topic_score_codex":0.009280956,"about_ca_topic_score_gemma":0.0068055107,"teacher_disagreement_score":0.009280956,"about_ca_system_score_codex":0.0017882771,"about_ca_system_score_gemma":0.0024968062,"threshold_uncertainty_score":0.018453836},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Scalability; Middleware (distributed applications); Database; Architecture; Satellite; Distributed computing","score_opus":0.012294103049097956,"score_gpt":0.2241104797065471,"score_spread":0.21181637665744915,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136058416","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13317426,0.002428932,0.82426375,0.0012740508,0.0003755764,0.00015022577,0.0005476364,0.016859302,0.020926358],"genre_scores_gemma":[0.54792625,0.0022953898,0.43643188,0.0005325383,0.0003139777,0.00010366968,0.0014381289,0.0014133997,0.00954474],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987649,0.0002770021,0.00019162451,0.00017384444,0.00045437028,0.0001384277],"domain_scores_gemma":[0.99598986,0.00083403295,0.00018331215,0.0023471075,0.0004827012,0.00016301183],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018474226,0.00044048994,0.0007182898,0.0011063673,0.0007095409,0.0025794965,0.0017051145,0.00087960577,0.0035699406],"category_scores_gemma":[0.0057190075,0.00081015524,0.0008806658,0.0024111555,0.00086378393,0.0062612155,0.0041190973,0.0014213042,0.0012016834],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00084585155,0.00038792662,0.017197903,0.0010192113,0.00037064365,0.0014454324,0.0027449667,0.058212407,0.05354891,0.18571596,0.021475842,0.65703493],"study_design_scores_gemma":[0.0002537331,0.00043808963,0.0029469028,0.00029950115,0.0005892937,0.0019063542,0.00095242466,0.29784268,0.06632508,0.24859482,0.3796961,0.00015504833],"about_ca_topic_score_codex":0.0023325447,"about_ca_topic_score_gemma":0.0029791726,"teacher_disagreement_score":0.0035699406,"about_ca_system_score_codex":0.0003935452,"about_ca_system_score_gemma":0.0005474991,"threshold_uncertainty_score":0.011942625},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; 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","score_opus":0.010908030588630723,"score_gpt":0.2453907915984678,"score_spread":0.2344827610098371,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141724398","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07426152,0.0032004528,0.91365445,0.0013961497,0.000089896275,0.0001939711,0.00054549065,0.0028661848,0.0037918577],"genre_scores_gemma":[0.5193705,0.0016108444,0.4709459,0.0004394113,0.00020996903,0.0002340623,0.0017308096,0.0006280487,0.0048304326],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9920763,0.002825018,0.00085380627,0.0008178975,0.002856656,0.00057040784],"domain_scores_gemma":[0.9919504,0.005921576,0.00039994888,0.0007658566,0.00076725293,0.00019508522],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057584196,0.00077326,0.0022906784,0.0014092516,0.00082716235,0.0037445151,0.002465655,0.0011750584,0.0025671904],"category_scores_gemma":[0.011878716,0.0009052192,0.0010910622,0.0037075875,0.0011757397,0.004623153,0.0018600782,0.0014502599,0.00059947046],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002366523,0.0007564998,0.004721431,0.0012276646,0.0003110024,0.0006014063,0.00068351725,0.47872564,0.016563386,0.09702289,0.01998481,0.37703517],"study_design_scores_gemma":[0.0001190887,0.0001382344,0.0004603981,0.00002353141,0.000070889284,0.00018959238,0.00018894006,0.94159514,0.007931672,0.046109997,0.003148252,0.000024287943],"about_ca_topic_score_codex":0.0039854944,"about_ca_topic_score_gemma":0.0037701086,"teacher_disagreement_score":0.0057584196,"about_ca_system_score_codex":0.0015027511,"about_ca_system_score_gemma":0.0015439366,"threshold_uncertainty_score":0.030453801},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Scalability; Thread (computing); Heap (data structure); Multi-core processor; Distributed computing; Parallel computing; Database transaction; Operating system; Programming language","score_opus":0.01529434217783311,"score_gpt":0.22644222987697848,"score_spread":0.2111478876991454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141833902","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45148236,0.0018308578,0.5090193,0.0012049967,0.0006647793,0.00044343516,0.0009833635,0.014192886,0.020177972],"genre_scores_gemma":[0.9311538,0.00012989901,0.06520716,0.00011795933,0.000073898984,0.00008985013,0.0006235191,0.00022697759,0.0023768398],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99858654,0.00023683008,0.00010219533,0.00032080413,0.00038898678,0.0003646812],"domain_scores_gemma":[0.99754286,0.0005738781,0.00014599509,0.0010104239,0.00048616185,0.00024068501],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009503695,0.0010233875,0.00096512167,0.0008141769,0.0016633691,0.0025821037,0.0029202006,0.0006474412,0.0032174627],"category_scores_gemma":[0.0034688108,0.00053186895,0.00033630952,0.0013097516,0.0007249746,0.003361661,0.0031860254,0.00083486433,0.00083384593],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006392353,0.0011395612,0.008866444,0.00042917192,0.0002968229,0.00094981235,0.000771019,0.40179142,0.087203585,0.016126469,0.028995631,0.44703776],"study_design_scores_gemma":[0.00030692937,0.000408421,0.0017633098,0.000023606834,0.00012738217,0.00032247565,0.00053871307,0.95823395,0.018280279,0.015755573,0.0041851774,0.000054313194],"about_ca_topic_score_codex":0.0040780357,"about_ca_topic_score_gemma":0.008466125,"teacher_disagreement_score":0.0040780357,"about_ca_system_score_codex":0.0010530179,"about_ca_system_score_gemma":0.0019560803,"threshold_uncertainty_score":0.010763466},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.04003804845750331,"score_gpt":0.2958259358899701,"score_spread":0.2557878874324668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143682328","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23191631,0.0037001858,0.72542375,0.00212552,0.00042744496,0.00023260979,0.0025119295,0.021904675,0.01175753],"genre_scores_gemma":[0.7734467,0.0010220394,0.21719071,0.00034510173,0.00024879028,0.0000756262,0.002636575,0.0010555355,0.0039789495],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9910232,0.0015058549,0.001461646,0.0013549813,0.0034739445,0.0011803731],"domain_scores_gemma":[0.9746523,0.011983865,0.0012878597,0.008777367,0.0023721366,0.0009265894],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005228457,0.001087648,0.0026009292,0.0014441856,0.0024318404,0.008435749,0.0050384793,0.0023221492,0.005461701],"category_scores_gemma":[0.024543,0.0014960065,0.0012096213,0.0034806414,0.002048505,0.017407231,0.0064024376,0.0023789587,0.0036280267],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009171318,0.0012449906,0.033378214,0.0027905835,0.0007308449,0.0022295604,0.0043582246,0.083032586,0.079179205,0.10421686,0.057633623,0.622034],"study_design_scores_gemma":[0.0005810113,0.0006774622,0.0030556954,0.00018851865,0.0005458107,0.0022114315,0.0019335044,0.6167038,0.068529695,0.27914062,0.026170032,0.0002623973],"about_ca_topic_score_codex":0.003360765,"about_ca_topic_score_gemma":0.004614528,"teacher_disagreement_score":0.008435749,"about_ca_system_score_codex":0.0009942695,"about_ca_system_score_gemma":0.0021836564,"threshold_uncertainty_score":0.027651072},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.046977370759670206,"score_gpt":0.2579352204925931,"score_spread":0.2109578497329229,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145957360","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024518851,0.00034081185,0.9735702,0.00016787805,0.000032509037,0.00004076932,0.00009572914,0.00021234326,0.0010208796],"genre_scores_gemma":[0.28076366,0.00038940294,0.7153053,0.00015338295,0.00008752726,0.0001750451,0.0005488069,0.00021525925,0.0023615195],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9973219,0.0011756208,0.00014274915,0.0004019975,0.0008214145,0.00013628507],"domain_scores_gemma":[0.995,0.002784934,0.0003327577,0.00065398245,0.0010332512,0.00019519814],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035908802,0.0009867304,0.0017824688,0.0012548012,0.0008232119,0.0016772497,0.0025091723,0.0016674481,0.001827541],"category_scores_gemma":[0.015311781,0.0005442859,0.0007234583,0.0015086605,0.0012679114,0.0027388195,0.002670983,0.0011762651,0.0006806678],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008845334,0.0002836921,0.0027408453,0.00056415814,0.00013896724,0.00020954212,0.00035389335,0.57099867,0.012304662,0.07155029,0.00767636,0.33229437],"study_design_scores_gemma":[0.000012575376,0.000049397597,0.00020654792,0.000010271958,0.000008167142,0.0000649469,0.00003339589,0.9796363,0.0015352311,0.01794044,0.0004899972,0.000012787584],"about_ca_topic_score_codex":0.0039375667,"about_ca_topic_score_gemma":0.004985914,"teacher_disagreement_score":0.0039375667,"about_ca_system_score_codex":0.0009535821,"about_ca_system_score_gemma":0.0011842219,"threshold_uncertainty_score":0.018990636},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02152338838573285,"score_gpt":0.24920143496609998,"score_spread":0.22767804658036714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147440220","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1274479,0.0012408312,0.86792886,0.00027945056,0.00006943396,0.00003256906,0.0005910651,0.0007558807,0.0016539547],"genre_scores_gemma":[0.6536906,0.0014257557,0.337769,0.00015078756,0.00023700637,0.0000690936,0.001930933,0.00010324687,0.0046235537],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99920875,0.000133289,0.000059802187,0.00013936871,0.00034172458,0.00011709973],"domain_scores_gemma":[0.996612,0.0018168864,0.0002627646,0.0007149627,0.00046485325,0.00012838261],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011392876,0.00038683743,0.00088625285,0.0014824662,0.00037735436,0.00083671557,0.00088229525,0.0008137988,0.0018467341],"category_scores_gemma":[0.0055311816,0.00026514535,0.00043967378,0.0018548719,0.0003719233,0.0015237486,0.0012042744,0.000601887,0.0011197046],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011336076,0.00019596783,0.01881465,0.00023909759,0.00010906129,0.000413215,0.00011634053,0.071089774,0.06565055,0.011799026,0.0030969793,0.82734174],"study_design_scores_gemma":[0.000020641235,0.00008449229,0.009883365,0.000015868834,0.000046566787,0.0007437708,0.00009802698,0.9378821,0.04223669,0.0068453187,0.0021047615,0.0000383318],"about_ca_topic_score_codex":0.001488841,"about_ca_topic_score_gemma":0.002593106,"teacher_disagreement_score":0.0018467341,"about_ca_system_score_codex":0.00032357496,"about_ca_system_score_gemma":0.0006268462,"threshold_uncertainty_score":0.006177902},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.015818699428412988,"score_gpt":0.24787564485482708,"score_spread":0.23205694542641409,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156892372","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07652448,0.0021810078,0.9089671,0.0018172422,0.00030828762,0.00040504945,0.0013325211,0.0018785557,0.006585797],"genre_scores_gemma":[0.80174255,0.000989435,0.18960807,0.0005458335,0.00032009167,0.000290124,0.0014990738,0.000345593,0.004659254],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99253535,0.0021005196,0.00066299096,0.0013422925,0.0025370775,0.00082172186],"domain_scores_gemma":[0.98101383,0.011929716,0.0008573743,0.0035230694,0.001817405,0.00085863937],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052558016,0.0012679754,0.0042971615,0.001927595,0.0019737545,0.0044767815,0.006446841,0.0024188212,0.0045313193],"category_scores_gemma":[0.028441435,0.0011542948,0.0014158437,0.0049483967,0.0017084009,0.008687461,0.0056463405,0.0022149403,0.0012214826],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0044720992,0.0005500252,0.004568313,0.0012160271,0.00030957186,0.00032941764,0.000650089,0.6389386,0.013419838,0.09765514,0.020209271,0.2176817],"study_design_scores_gemma":[0.00013312159,0.00017521497,0.00031850202,0.000023365335,0.00008163668,0.0002834029,0.00011627092,0.9083197,0.0023713345,0.086307704,0.0018364489,0.0000332671],"about_ca_topic_score_codex":0.00154627,"about_ca_topic_score_gemma":0.0026683144,"teacher_disagreement_score":0.006446841,"about_ca_system_score_codex":0.002524327,"about_ca_system_score_gemma":0.0022184344,"threshold_uncertainty_score":0.027795672},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Discriminative model; Set (abstract data type); Tree (set theory); Information retrieval; Measure (data warehouse); Quality (philosophy); Series (stratigraphy); Data mining; Artificial intelligence; Mathematics","score_opus":0.03418460099070763,"score_gpt":0.3495786809841878,"score_spread":0.3153940799934802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2157044362","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0323919,0.0023846347,0.9565058,0.0019881919,0.00014206785,0.00010110847,0.0011776371,0.0035975624,0.0017111569],"genre_scores_gemma":[0.4254284,0.001441173,0.563081,0.0007538561,0.0007796686,0.00027253793,0.0036779416,0.0010612451,0.0035042139],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98694485,0.003939845,0.0015677905,0.002749191,0.0038230375,0.0009751743],"domain_scores_gemma":[0.964893,0.021766467,0.0014017401,0.008451535,0.0027607759,0.0007264594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0085006505,0.0015273009,0.004818978,0.00334122,0.0024817996,0.0067157587,0.0070211645,0.0032338707,0.0053056558],"category_scores_gemma":[0.040784586,0.0020138887,0.0025613874,0.009267921,0.0030489392,0.019165365,0.005584576,0.0038118642,0.0020407594],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041401233,0.00049784826,0.00950204,0.0020762193,0.00062863313,0.0005512723,0.0016949342,0.19673789,0.016056048,0.2537173,0.053919364,0.46047837],"study_design_scores_gemma":[0.00014704697,0.00011775712,0.00058805913,0.000040368835,0.00013897334,0.0005980825,0.00031203218,0.5930769,0.0044049625,0.39570838,0.0048061996,0.00006128094],"about_ca_topic_score_codex":0.0027486596,"about_ca_topic_score_gemma":0.0031519618,"teacher_disagreement_score":0.0085006505,"about_ca_system_score_codex":0.0022162537,"about_ca_system_score_gemma":0.0023443778,"threshold_uncertainty_score":0.044956267},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.1166235600128447,"score_gpt":0.4294959857511475,"score_spread":0.3128724257383028,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170688652","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.087731756,0.0016429276,0.89731705,0.00069616997,0.00019119296,0.000377973,0.0012845567,0.003974736,0.0067835557],"genre_scores_gemma":[0.3245328,0.0007950961,0.66948104,0.0001539444,0.00010726123,0.00025709302,0.0029981444,0.0005318079,0.0011428744],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9953908,0.001990187,0.0006650296,0.0006860617,0.0011453221,0.00012253955],"domain_scores_gemma":[0.9822104,0.011597368,0.0007956791,0.002304488,0.0028222553,0.00026973317],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004904095,0.00054207304,0.001111243,0.010697845,0.0013743155,0.004032878,0.0015485914,0.0010182017,0.0039217896],"category_scores_gemma":[0.030561551,0.0004899811,0.000765036,0.011822916,0.0014636952,0.005064588,0.0020828717,0.0008781508,0.000922015],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00066971936,0.0004790099,0.009518186,0.0007994192,0.0003343703,0.0003659847,0.00244703,0.030786587,0.025090657,0.08084552,0.00869566,0.8399679],"study_design_scores_gemma":[0.00018800901,0.0002686151,0.0067073326,0.00021070293,0.0002709236,0.00094886386,0.0024550518,0.73573184,0.03135468,0.19494063,0.026786812,0.00013654852],"about_ca_topic_score_codex":0.0063467505,"about_ca_topic_score_gemma":0.006656075,"teacher_disagreement_score":0.010697845,"about_ca_system_score_codex":0.0014222182,"about_ca_system_score_gemma":0.002580957,"threshold_uncertainty_score":0.02593565},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Scalability; Multi-core processor; Serialization; Operating system; Parallel computing; Database","score_opus":0.04511224232357768,"score_gpt":0.24973414704582975,"score_spread":0.20462190472225206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2172197636","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9782657,0.0020333386,0.010683595,0.00053340447,0.00019588899,0.00004194151,0.00033310356,0.0031599207,0.0047529903],"genre_scores_gemma":[0.9958111,0.00017137779,0.0026296454,0.000049393624,0.000035104244,0.000015511181,0.00018738545,0.00006505865,0.0010355515],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99716705,0.0004607956,0.0001768786,0.00046451617,0.0010124436,0.0007183834],"domain_scores_gemma":[0.98764265,0.006438448,0.0007625024,0.0024347843,0.0019053535,0.00081638445],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002332588,0.0008444802,0.00080024905,0.0012364108,0.00090476905,0.001653621,0.0015956592,0.0006628449,0.004637957],"category_scores_gemma":[0.010958115,0.0004839129,0.00025800272,0.0013305531,0.0010332976,0.0042171995,0.001343323,0.0011599312,0.0003651628],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.012938208,0.0022687325,0.046393827,0.0012144543,0.0003599546,0.0008254221,0.0010993382,0.44677347,0.15581764,0.012377153,0.02336645,0.29656532],"study_design_scores_gemma":[0.00034425332,0.0009536664,0.010788786,0.000059011796,0.00016741014,0.0002547102,0.00062833703,0.9268247,0.052782375,0.0055147046,0.001599621,0.00008241854],"about_ca_topic_score_codex":0.0048701228,"about_ca_topic_score_gemma":0.006298512,"teacher_disagreement_score":0.0048701228,"about_ca_system_score_codex":0.0012520106,"about_ca_system_score_gemma":0.0020168044,"threshold_uncertainty_score":0.015515506},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Serialization; Concurrency control; Serializability; Commit; Transactional memory; Distributed transaction; Online transaction processing; Database transaction","score_opus":0.018270925180170403,"score_gpt":0.27674830958537444,"score_spread":0.258477384405204,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2386452338","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10558655,0.0004201545,0.86240673,0.0007853316,0.00040908498,0.0002991787,0.0002206264,0.01722453,0.012647817],"genre_scores_gemma":[0.7097064,0.0003176954,0.27130136,0.0004653296,0.00022103479,0.0002560393,0.00060096954,0.0018262271,0.0153049],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9947259,0.0010000414,0.00042707226,0.0008942952,0.002168217,0.00078437873],"domain_scores_gemma":[0.98430204,0.0033077095,0.0007012943,0.0100377295,0.0012507563,0.00040041667],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004282639,0.001183663,0.0011441654,0.00088226586,0.0013042879,0.0038390209,0.00282738,0.0010068446,0.0067019234],"category_scores_gemma":[0.013254509,0.001168532,0.0008838439,0.0015952217,0.0022242828,0.0053798477,0.0043801465,0.002408984,0.0020127485],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038681205,0.00076467596,0.005819903,0.0006029899,0.0002458489,0.0005565429,0.0011357394,0.09781514,0.14040563,0.16876197,0.021944763,0.5580787],"study_design_scores_gemma":[0.0009028547,0.00065565953,0.0013628001,0.00011698286,0.00039593494,0.0004161755,0.00041377984,0.48317978,0.1474164,0.32817388,0.036818035,0.0001477126],"about_ca_topic_score_codex":0.002168544,"about_ca_topic_score_gemma":0.0030153622,"teacher_disagreement_score":0.0067019234,"about_ca_system_score_codex":0.0007705898,"about_ca_system_score_gemma":0.0030858328,"threshold_uncertainty_score":0.02264905},"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.02115508178165599,"score_gpt":0.263779211117652,"score_spread":0.24262412933599603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2755658662","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.051134534,0.5899597,0.25860384,0.058074847,0.0012367122,0.00011520617,0.0018430515,0.0013665102,0.03766572],"genre_scores_gemma":[0.30216563,0.5546473,0.12019776,0.008320748,0.006998458,0.0001373027,0.0028668102,0.0005662181,0.004099707],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978035,0.00058306433,0.000100893754,0.0007293304,0.0006286118,0.0001545708],"domain_scores_gemma":[0.96814716,0.025942622,0.0011529954,0.0024630039,0.0016408637,0.0006534325],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003521405,0.0008484069,0.0015252125,0.003811397,0.0010943662,0.004307706,0.0026868416,0.0021509258,0.0038704684],"category_scores_gemma":[0.0199426,0.0010335068,0.0007591307,0.007085608,0.0036709548,0.014969991,0.0024975247,0.0029257261,0.0012778296],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000164054,0.00017945888,0.0126058245,0.004672566,0.00019318289,0.00026728405,0.0004383849,0.018742736,0.0019274188,0.2863099,0.04613047,0.62836874],"study_design_scores_gemma":[0.000021561282,0.0000994178,0.0060193962,0.00088910694,0.00009043835,0.0014783939,0.0007428388,0.04954101,0.0016104669,0.71150386,0.22793522,0.000068347224],"about_ca_topic_score_codex":0.0023606038,"about_ca_topic_score_gemma":0.0028264483,"teacher_disagreement_score":0.004307706,"about_ca_system_score_codex":0.0015874488,"about_ca_system_score_gemma":0.0014598825,"threshold_uncertainty_score":0.018623173},"labels":[],"label_agreement":null},{"id":"W2948421209","doi":"10.1007/s00778-019-00543-2","title":"Thematic issue on data management for graphs","year":2019,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Thematic map; Data science; Computer science; Data management; Geography; Cartography; Database","score_opus":0.05563453672975735,"score_gpt":0.3060517895038253,"score_spread":0.2504172527740679,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2948421209","genre_codex":"editorial","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033892663,0.10124123,0.10897337,0.136869,0.5027969,0.0012207853,0.0065560183,0.0029090324,0.1360444],"genre_scores_gemma":[0.024667066,0.1367857,0.06101179,0.024219062,0.3748801,0.0010051981,0.025047107,0.0042861323,0.34809792],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99572194,0.0011423443,0.00027000884,0.0007285797,0.0018300341,0.00030714646],"domain_scores_gemma":[0.98380846,0.0054797744,0.000659294,0.0023510656,0.0054555824,0.0022457773],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007886367,0.001121647,0.0020570348,0.0060815183,0.0022559564,0.009470595,0.0031184857,0.0032448359,0.071071215],"category_scores_gemma":[0.017122248,0.0005440097,0.0014832788,0.009960803,0.0017227329,0.012327957,0.004318998,0.003983,0.02671761],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040240535,0.00006647722,0.00026583183,0.0012353329,0.00003414623,0.000028820088,0.00008742034,0.0005030547,0.0009087656,0.029134663,0.81848997,0.14920531],"study_design_scores_gemma":[0.000010624384,0.000056287583,0.0005008,0.00072438153,0.000017755254,0.00010929619,0.00012857412,0.00109443,0.00044645116,0.02246286,0.9744313,0.00001727431],"about_ca_topic_score_codex":0.001692628,"about_ca_topic_score_gemma":0.0019380031,"teacher_disagreement_score":0.071071215,"about_ca_system_score_codex":0.002679558,"about_ca_system_score_gemma":0.0046047196,"threshold_uncertainty_score":0.23775685},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Maximization; Computer science; Mathematical optimization; Mathematics","score_opus":0.010919874059083779,"score_gpt":0.2582131317671943,"score_spread":0.2472932577081105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3171876400","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023629325,0.00023992745,0.99525696,0.00015284585,0.00004985687,0.00002052767,0.000047349436,0.00010360171,0.0017660167],"genre_scores_gemma":[0.42731798,0.0012133756,0.55737466,0.00049960037,0.0005183368,0.00046450936,0.00041305905,0.00043537124,0.011763138],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99923503,0.00024600368,0.000034000106,0.00015654373,0.00021128035,0.00011716503],"domain_scores_gemma":[0.9945557,0.0045923074,0.00017920687,0.00022570672,0.0002985057,0.00014850126],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016765705,0.0017178769,0.002095831,0.0013514278,0.00080028793,0.0018798377,0.0025953527,0.0022098066,0.005645781],"category_scores_gemma":[0.010074662,0.00085678394,0.0013026189,0.0017504309,0.0017599771,0.0024899251,0.0023885928,0.0024411215,0.0007506197],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010283352,0.00009366921,0.00027136283,0.00019774307,0.000071462724,0.00011384923,0.00010033811,0.8486264,0.0021483577,0.091424294,0.0032235926,0.05362611],"study_design_scores_gemma":[0.0000031516613,0.000004667423,0.000013253232,0.0000036255274,0.000004163746,0.0000062128556,0.0000029423231,0.9888181,0.00012960374,0.0108001055,0.00021142997,0.0000026381051],"about_ca_topic_score_codex":0.0062469663,"about_ca_topic_score_gemma":0.008151116,"teacher_disagreement_score":0.0062469663,"about_ca_system_score_codex":0.0016352176,"about_ca_system_score_gemma":0.0018524721,"threshold_uncertainty_score":0.018887043},"labels":[],"label_agreement":null},{"id":"W4322741464","doi":"10.1007/s00778-023-00787-z","title":"Leveraging user itinerary to improve personalized deep matching at Fliggy","year":2023,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Natural Science Foundation of Guangxi Province; National Natural Science Foundation of China","keywords":"Computer science; Matching (statistics); Weighting; Benchmark (surveying); Artificial intelligence; Machine learning; Data mining","score_opus":0.026442433696779455,"score_gpt":0.2688672081487491,"score_spread":0.24242477445196964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4322741464","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22899175,0.007945437,0.7170873,0.002263046,0.0011528389,0.00034045952,0.0066318517,0.017178545,0.018408677],"genre_scores_gemma":[0.7948023,0.000843482,0.16655864,0.00095830805,0.0002985021,0.00009826334,0.005307286,0.0007346468,0.030398574],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99831045,0.00033842685,0.000079562786,0.0005370586,0.00049463386,0.00023990781],"domain_scores_gemma":[0.99793327,0.00062621734,0.00008900814,0.0007552399,0.00041875005,0.00017753235],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014995422,0.0011896227,0.002119713,0.0020475236,0.0013117903,0.0016604222,0.0021669413,0.0023387745,0.011535896],"category_scores_gemma":[0.0065334467,0.0007061121,0.0009590399,0.0023216929,0.0004975036,0.0048537287,0.0020808775,0.002053042,0.006114601],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002129904,0.0018958371,0.02293583,0.00039598398,0.0005991404,0.00035374594,0.0005855229,0.061761763,0.019709056,0.00959817,0.070554435,0.80948055],"study_design_scores_gemma":[0.00006666792,0.00046761823,0.0042735473,0.000052251253,0.0002067273,0.00033023683,0.00018832862,0.950637,0.010390586,0.019096041,0.014191874,0.000099184115],"about_ca_topic_score_codex":0.01599964,"about_ca_topic_score_gemma":0.054066803,"teacher_disagreement_score":0.01599964,"about_ca_system_score_codex":0.00093812414,"about_ca_system_score_gemma":0.0012840312,"threshold_uncertainty_score":0.038591385},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.016390848947298697,"score_gpt":0.22849349027064442,"score_spread":0.21210264132334572,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4364362023","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0816086,0.0013537816,0.7442208,0.0009373897,0.0010246409,0.0002720196,0.0012816906,0.1522357,0.017065296],"genre_scores_gemma":[0.5976306,0.0012221427,0.34794542,0.0006624365,0.00031776848,0.0003116488,0.004291348,0.014285517,0.03333308],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988574,0.00028876108,0.00009258645,0.00017833014,0.00039655683,0.00018636476],"domain_scores_gemma":[0.99758303,0.00048243246,0.00013027781,0.0013792459,0.00021457479,0.0002105488],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020734742,0.0009285569,0.0010197097,0.0008857762,0.00146124,0.0025014835,0.0023964697,0.0011060201,0.009207987],"category_scores_gemma":[0.0034571195,0.000866059,0.00062491116,0.0013120583,0.0015456253,0.0053312723,0.0040754103,0.0017588105,0.0020731033],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0065766275,0.0006631613,0.005906114,0.0012661683,0.00033925922,0.0012538708,0.0024262106,0.047010966,0.046921987,0.15478599,0.15669467,0.576155],"study_design_scores_gemma":[0.0014151471,0.0012395325,0.0020309351,0.00029847302,0.0002909037,0.0010604953,0.0007023344,0.44882905,0.10853247,0.15526429,0.28005645,0.00027994032],"about_ca_topic_score_codex":0.0015902793,"about_ca_topic_score_gemma":0.0017284087,"teacher_disagreement_score":0.009207987,"about_ca_system_score_codex":0.0005625294,"about_ca_system_score_gemma":0.0012287871,"threshold_uncertainty_score":0.0308038},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.08005984562599212,"score_gpt":0.2340146645626377,"score_spread":0.1539548189366456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385408288","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13205193,0.0017531265,0.82059604,0.0023952802,0.0005577668,0.00030001698,0.00096387067,0.022702932,0.018679036],"genre_scores_gemma":[0.54525846,0.00043254183,0.44610518,0.0005482458,0.00024229128,0.00018608858,0.0015793683,0.001001853,0.0046458477],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974148,0.00048489333,0.00014702308,0.00051619875,0.0010039301,0.00043307996],"domain_scores_gemma":[0.9941327,0.0026139228,0.00029862893,0.0021297187,0.00063271803,0.00019232892],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085441157,0.0012414752,0.0010437763,0.0019363794,0.00081612426,0.0026229536,0.002646494,0.0012999072,0.007845215],"category_scores_gemma":[0.0073145404,0.0006017605,0.0011063715,0.0032406452,0.0010659934,0.006005454,0.0027421934,0.0014872347,0.0016483363],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015506095,0.0008144849,0.00670225,0.00071724417,0.00022284995,0.00026213404,0.0003503669,0.17179945,0.048229456,0.06952446,0.047101825,0.6527249],"study_design_scores_gemma":[0.00018552369,0.0002760362,0.0007450313,0.000025840143,0.000081560975,0.00018187777,0.00018401255,0.90737957,0.013930005,0.0703106,0.0066674813,0.00003243486],"about_ca_topic_score_codex":0.007010965,"about_ca_topic_score_gemma":0.01849857,"teacher_disagreement_score":0.007845215,"about_ca_system_score_codex":0.0018416957,"about_ca_system_score_gemma":0.0029666696,"threshold_uncertainty_score":0.026244879},"labels":[],"label_agreement":null},{"id":"W4386099755","doi":"10.1007/s00778-023-00809-w","title":"xDBTagger: explainable natural language interface to databases using keyword mappings and schema graph","year":2023,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Pipeline (software); Schema (genetic algorithms); SQL; Artificial intelligence; Natural language; Natural language processing; Information retrieval; Database; Programming language","score_opus":0.04688202870401947,"score_gpt":0.31181788788836495,"score_spread":0.2649358591843455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386099755","genre_codex":"software","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":"software","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005418815,0.0003288142,0.46128872,0.00055974774,0.00021545125,0.00034287656,0.049044024,0.47732016,0.0054812585],"genre_scores_gemma":[0.12501815,0.0014055549,0.5453085,0.0025686892,0.00018167049,0.0025089723,0.16805433,0.13125004,0.023704149],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99902844,0.000205286,0.00020204418,0.00024470768,0.00024706667,0.00007230552],"domain_scores_gemma":[0.99599683,0.0027886636,0.00018103547,0.00057045935,0.0003294005,0.00013359264],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018661096,0.0019503069,0.0012829052,0.0020871141,0.00050222466,0.0028931885,0.0021786408,0.0017725923,0.052730847],"category_scores_gemma":[0.008149871,0.0013758439,0.001692777,0.0015415685,0.0005316115,0.004879174,0.0042923857,0.0017493347,0.01615648],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020332774,0.0002794673,0.005830723,0.004546542,0.00043067988,0.0019030588,0.0030390527,0.010154159,0.02555507,0.073255114,0.63228714,0.24068576],"study_design_scores_gemma":[0.0008741671,0.00021130366,0.0023503702,0.00093087717,0.0002588569,0.0012129932,0.00065674674,0.1311364,0.047231775,0.07378408,0.74107265,0.00027977867],"about_ca_topic_score_codex":0.0060851797,"about_ca_topic_score_gemma":0.0072100814,"teacher_disagreement_score":0.052730847,"about_ca_system_score_codex":0.0013378756,"about_ca_system_score_gemma":0.001237869,"threshold_uncertainty_score":0.17640227},"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,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.048473666201916135,"score_gpt":0.2954808845975803,"score_spread":0.2470072183956642,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387098977","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008918214,0.7758262,0.1561585,0.003554551,0.0018197277,0.00026340326,0.0007852322,0.0011605023,0.05151368],"genre_scores_gemma":[0.051203508,0.85205764,0.08086892,0.0017136047,0.0034488111,0.00019875877,0.0017702408,0.0003311789,0.008407349],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99776363,0.0004211985,0.00029717173,0.00034238872,0.0010054572,0.0001701778],"domain_scores_gemma":[0.9962425,0.0019540808,0.00019713919,0.00034534693,0.0010884046,0.0001725063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021380547,0.0009378541,0.000983917,0.0040611015,0.000626496,0.00355887,0.0018648001,0.0012013603,0.007193884],"category_scores_gemma":[0.004273574,0.0007748754,0.00079907867,0.013782946,0.00079868163,0.006126662,0.0013425837,0.0020283011,0.0027670076],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019590772,0.00018291459,0.0021231945,0.006775458,0.000113580856,0.00024061651,0.00030879452,0.008410766,0.0026092126,0.07204111,0.049171373,0.857827],"study_design_scores_gemma":[0.000029882134,0.00027345304,0.0020170703,0.0024917484,0.000084416315,0.0012390073,0.00032375855,0.02592468,0.0027561435,0.051299587,0.9134897,0.00007053261],"about_ca_topic_score_codex":0.0015295867,"about_ca_topic_score_gemma":0.0007388945,"teacher_disagreement_score":0.007193884,"about_ca_system_score_codex":0.00093907467,"about_ca_system_score_gemma":0.0018244036,"threshold_uncertainty_score":0.024065971},"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,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.16349691109568132,"score_gpt":0.37462525822397275,"score_spread":0.21112834712829143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391785473","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019550703,0.00023416623,0.9277346,0.00023235414,0.000037165,0.00070138654,0.0002706219,0.04948231,0.0017566188],"genre_scores_gemma":[0.20974255,0.0001126632,0.78490055,0.00017284871,0.000022145341,0.00044546433,0.0010332701,0.0015014714,0.0020690793],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965822,0.000901247,0.00027585824,0.0009878473,0.0008891906,0.0003635999],"domain_scores_gemma":[0.9917263,0.0032750128,0.0007231186,0.0019294772,0.0016348094,0.0007112386],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059374855,0.0015847886,0.0013422021,0.002162931,0.0010695502,0.0020194242,0.004105815,0.0014531043,0.008305961],"category_scores_gemma":[0.013760228,0.0011526071,0.0015951754,0.00091892487,0.0012690309,0.0028273128,0.0041501774,0.0018652156,0.003088539],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014472023,0.00113609,0.009201289,0.0011980658,0.0002375984,0.00080006674,0.0011890564,0.15453523,0.058412734,0.025113463,0.021593641,0.72513556],"study_design_scores_gemma":[0.00010608394,0.00022601603,0.00086227374,0.000041549243,0.00004833666,0.00014401872,0.00013489008,0.9341605,0.03320212,0.0100942245,0.020922037,0.000058051366],"about_ca_topic_score_codex":0.0038561586,"about_ca_topic_score_gemma":0.0040541664,"teacher_disagreement_score":0.008305961,"about_ca_system_score_codex":0.0014112628,"about_ca_system_score_gemma":0.003912994,"threshold_uncertainty_score":0.0314008},"labels":[],"label_agreement":null},{"id":"W4398180243","doi":"10.1007/s00778-024-00847-y","title":"Discovering approximate implicit domain orders through order dependencies","year":2024,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; York University; University of Waterloo","funders":"","keywords":"Computer science; Domain (mathematical analysis); Order (exchange); Data mining; Theoretical computer science; Mathematics; Business","score_opus":0.13203993517966403,"score_gpt":0.4151772800383418,"score_spread":0.2831373448586778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398180243","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24866703,0.0016607725,0.7358601,0.000901895,0.00010654081,0.0001728431,0.004804878,0.0022374974,0.0055884677],"genre_scores_gemma":[0.7728777,0.0011580726,0.21538155,0.00015815087,0.00010121596,0.00010442044,0.0072640725,0.00025765877,0.0026971048],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9985202,0.0003166295,0.00013425169,0.0003549543,0.0005086923,0.00016523305],"domain_scores_gemma":[0.9920551,0.0053108977,0.0006354673,0.0010629761,0.00073390437,0.00020159788],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010530535,0.00095862884,0.0011350423,0.0037570372,0.00068131293,0.0023335323,0.0013545983,0.0008527173,0.0031835577],"category_scores_gemma":[0.010416735,0.0009427475,0.0009817961,0.0036917024,0.00082366867,0.00617337,0.0014043461,0.0021052493,0.0010898209],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014350989,0.0008503844,0.05019819,0.00071248534,0.00032101388,0.0011838427,0.00101366,0.40680793,0.008253225,0.072117224,0.016358668,0.4407483],"study_design_scores_gemma":[0.000037889404,0.000073600444,0.0020754663,0.000061949475,0.000056598816,0.00012355282,0.00021022113,0.91937333,0.0020760235,0.07235458,0.0035361664,0.000020654976],"about_ca_topic_score_codex":0.009516074,"about_ca_topic_score_gemma":0.020215727,"teacher_disagreement_score":0.009516074,"about_ca_system_score_codex":0.0009365803,"about_ca_system_score_gemma":0.0023302233,"threshold_uncertainty_score":0.018921316},"labels":[],"label_agreement":null},{"id":"W4400404039","doi":"10.1007/s00778-024-00860-1","title":"Minimum motif-cut: a workload-aware RDF graph partitioning strategy","year":2024,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Motif (music); RDF; Computer science; Workload; Graph; Parallel computing; Theoretical computer science; Artificial intelligence; Operating system; Art; Semantic Web","score_opus":0.01798901487496369,"score_gpt":0.25051093290324294,"score_spread":0.23252191802827926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400404039","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08073127,0.0006002677,0.9048095,0.00077303516,0.00016995684,0.00059603737,0.001311842,0.003634014,0.007374144],"genre_scores_gemma":[0.3717441,0.00028169889,0.6165549,0.00032587702,0.00008760746,0.00033336453,0.00426219,0.00088681956,0.0055234237],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99869764,0.0003026526,0.0000704815,0.0003360729,0.0004116413,0.00018143811],"domain_scores_gemma":[0.997609,0.000773254,0.00011783505,0.0007059043,0.00057542056,0.00021853481],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000912176,0.0012269499,0.0013299666,0.001985919,0.0016366349,0.0014973512,0.003036257,0.0011566778,0.005010044],"category_scores_gemma":[0.0052908137,0.0006117238,0.0009310477,0.0023991936,0.0006264473,0.0032355448,0.002441954,0.0010694925,0.0011146372],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010167832,0.0010021083,0.0050675604,0.000742671,0.00021820562,0.00049696036,0.0008206304,0.1674549,0.054757714,0.043455694,0.038105953,0.68686086],"study_design_scores_gemma":[0.000065685286,0.00016769872,0.0008657371,0.000040214425,0.000081098835,0.0003752419,0.00058942335,0.93804854,0.013003333,0.039202146,0.0075313374,0.000029614024],"about_ca_topic_score_codex":0.0046745953,"about_ca_topic_score_gemma":0.011603229,"teacher_disagreement_score":0.005010044,"about_ca_system_score_codex":0.0010133606,"about_ca_system_score_gemma":0.0019326461,"threshold_uncertainty_score":0.01676023},"labels":[],"label_agreement":null},{"id":"W4405006607","doi":"10.1007/s00778-024-00883-8","title":"A graph pattern mining framework for large graphs on GPU","year":2024,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Graph; CUDA; Theoretical computer science; Data science; Data mining; Parallel computing","score_opus":0.0195451147952551,"score_gpt":0.27693911823719536,"score_spread":0.25739400344194024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405006607","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011567127,0.00051279645,0.9338514,0.00040383617,0.00010557929,0.00021563085,0.0022796355,0.049194742,0.0018691468],"genre_scores_gemma":[0.06400876,0.0002957257,0.9256461,0.00022098048,0.000047188587,0.00021871887,0.004546974,0.001939216,0.0030762241],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992072,0.00009418585,0.00007390519,0.00023613646,0.00030013642,0.00008849098],"domain_scores_gemma":[0.9987626,0.00039237458,0.00007758529,0.00038151396,0.0002766923,0.000109163695],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006983445,0.0014919605,0.0013124595,0.0026813387,0.00105655,0.0025273883,0.0040699677,0.001102126,0.007825301],"category_scores_gemma":[0.0033238954,0.0011523283,0.002238541,0.00384706,0.0004997798,0.0029434452,0.0020318099,0.0018628574,0.0035121092],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060743943,0.0004748355,0.006065844,0.0007796966,0.0004903165,0.00046349593,0.00033942895,0.07095297,0.022113694,0.024073951,0.09719057,0.7764478],"study_design_scores_gemma":[0.00012094945,0.00008104272,0.00097154296,0.000042524076,0.00007723334,0.00024301911,0.00011806005,0.9298173,0.0074367705,0.040060706,0.020995935,0.00003486595],"about_ca_topic_score_codex":0.012698234,"about_ca_topic_score_gemma":0.036070865,"teacher_disagreement_score":0.012698234,"about_ca_system_score_codex":0.00081874616,"about_ca_system_score_gemma":0.0019614357,"threshold_uncertainty_score":0.0261783},"labels":[],"label_agreement":null},{"id":"W4411935250","doi":"10.1007/s00778-025-00933-9","title":"Beyond influence: voting theory for opinion maximization","year":2025,"lang":"en","type":"article","venue":"The VLDB Journal","topic":"Opinion Dynamics and Social Influence","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Key Research and Development Program of Zhejiang Province; Ningbo Municipal Bureau of Science and Technology; Novo Nordisk","keywords":"Voting; Maximization; Utility maximization; Approval voting; Mathematical economics; Political science; Computer science; Psychology; Economics; Social psychology; Disapproval voting; Law; Politics","score_opus":0.006804691820581122,"score_gpt":0.29149749965931665,"score_spread":0.2846928078387355,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411935250","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06825323,0.005002193,0.7918739,0.022357285,0.0007898068,0.00009954263,0.00040283604,0.00016780666,0.111053474],"genre_scores_gemma":[0.9437632,0.0026618182,0.032007255,0.0015439447,0.0024200273,0.00021910192,0.00023161095,0.0001286525,0.017024444],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9966288,0.0017951031,0.00011178712,0.00050826964,0.00057773205,0.00037821877],"domain_scores_gemma":[0.9770563,0.01852276,0.0011232984,0.001064743,0.0015107744,0.0007220989],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0061792876,0.0008182629,0.0021130485,0.0017838954,0.0021445286,0.00553591,0.0023917262,0.0032801789,0.010678473],"category_scores_gemma":[0.035946496,0.0005011031,0.0015326535,0.0022882354,0.0039793584,0.011159547,0.0020661661,0.004027831,0.0011269061],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016846028,0.000013562448,0.00026429768,0.000042322772,0.000021219901,0.000019544941,0.000120699,0.004583245,0.00011576798,0.9870107,0.002384403,0.0054074177],"study_design_scores_gemma":[0.000008373452,0.0000049250903,0.00011407965,0.000010190369,0.0000061873525,0.000009279117,0.000020130785,0.02950971,0.000032962744,0.96928567,0.0009924929,0.000006016568],"about_ca_topic_score_codex":0.0033373798,"about_ca_topic_score_gemma":0.0020669852,"teacher_disagreement_score":0.010678473,"about_ca_system_score_codex":0.002466949,"about_ca_system_score_gemma":0.0012669843,"threshold_uncertainty_score":0.03572303},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.8844173631534311,"score_gpt":0.5434835487839118,"score_spread":0.3409338143695193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W760598031","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20482203,0.4718415,0.2501662,0.02193191,0.00061691535,0.00050951267,0.010690295,0.0019408974,0.03748073],"genre_scores_gemma":[0.4661339,0.3631345,0.14219293,0.004883834,0.00095963036,0.00041791316,0.01806351,0.0006093284,0.003604455],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98078996,0.003948452,0.0025944386,0.002924362,0.009087564,0.0006552319],"domain_scores_gemma":[0.86044216,0.09986163,0.007603403,0.010090188,0.01982508,0.0021775505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018697688,0.00067749346,0.0015072913,0.0150140915,0.0009242385,0.007538349,0.0027443732,0.0012737045,0.0019589588],"category_scores_gemma":[0.07480663,0.0007422493,0.0010518052,0.022260198,0.0017181793,0.017352844,0.0030627295,0.0015985226,0.0010219851],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016783694,0.00017706855,0.0908149,0.0036929497,0.00018980583,0.00007622846,0.0015076757,0.0019367894,0.00225872,0.014154048,0.011427636,0.8735963],"study_design_scores_gemma":[0.000067401,0.00078845577,0.13737592,0.009211984,0.0006429999,0.0038354963,0.015518659,0.03142736,0.019080445,0.061744273,0.7199531,0.00035390182],"about_ca_topic_score_codex":0.0024348474,"about_ca_topic_score_gemma":0.0030642527,"teacher_disagreement_score":0.018697688,"about_ca_system_score_codex":0.0014644377,"about_ca_system_score_gemma":0.002510754,"threshold_uncertainty_score":0.098884046},"labels":[],"label_agreement":null}]}