{"meta":{"query_hash":"416db036aa93","filters":{"venue":"Proceedings of the VLDB Endowment"},"cohort_total":329,"direct_labels_cover":0,"predictions_cover":329,"exported":329,"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/416db036aa93","api":"https://metacan.xera.ac/api/v1/cohort?venue=Proceedings+of+the+VLDB+Endowment"},"results":[{"id":"W1509825211","doi":"10.14778/2733085.2733099","title":"Generating top-k packages via preference elicitation","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Recommender system; Preference elicitation; Ranking (information retrieval); Function (biology); Preference; Rank (graph theory); Information retrieval; Variety (cybernetics); Learning to rank; Preference learning; Machine learning; Data mining; Artificial intelligence; Mathematics","score_opus":0.01761425993665048,"score_gpt":0.21582282076499093,"score_spread":0.19820856082834046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1509825211","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":[0.02666762,0.00038364544,0.9643417,0.000426911,0.00005709253,0.0005581886,0.001187642,0.0039786147,0.0023985389],"genre_scores_gemma":[0.16687642,0.0002836219,0.823087,0.00031374014,0.00008941913,0.0007406676,0.0047445986,0.00043284745,0.0034316748],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9940918,0.0023474584,0.00049274263,0.0012127407,0.0015436746,0.00031159524],"domain_scores_gemma":[0.9816551,0.010766596,0.000882314,0.003455147,0.002766119,0.00047458755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054607713,0.0024298944,0.0031432065,0.003350758,0.0015039708,0.0019975016,0.0038376737,0.0023518775,0.007231285],"category_scores_gemma":[0.027782915,0.0011976366,0.002585514,0.0053527295,0.0010243874,0.004325232,0.0035470661,0.0025834218,0.004808981],"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":[0.0014929227,0.0008791267,0.008956404,0.0012807335,0.00033391773,0.00039279554,0.0014388309,0.10616814,0.010401131,0.021759849,0.038723964,0.8081721],"study_design_scores_gemma":[0.00021755714,0.00030390284,0.0013665376,0.0000922474,0.0001344349,0.0004244465,0.00043580902,0.9248314,0.0070997034,0.054846223,0.010126392,0.00012127299],"about_ca_topic_score_codex":0.003431275,"about_ca_topic_score_gemma":0.009142269,"teacher_disagreement_score":0.007231285,"about_ca_system_score_codex":0.0011418979,"about_ca_system_score_gemma":0.0026039418,"threshold_uncertainty_score":0.028879642},"labels":[],"label_agreement":null},{"id":"W1539153042","doi":"10.14778/2831360.2831367","title":"S-Store","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Science Foundation","keywords":"Online transaction processing; Computer science; Stream processing; Transaction processing; Distributed transaction; Database; Database transaction; Correctness; Operating system; Programming language","score_opus":0.02531497744140868,"score_gpt":0.22789163883845423,"score_spread":0.20257666139704555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1539153042","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":[0.027662832,0.0030029637,0.3359903,0.0020863006,0.0014913538,0.0014781007,0.0349041,0.43129197,0.16209215],"genre_scores_gemma":[0.27307147,0.004173631,0.30961448,0.004615123,0.0011017997,0.0012215995,0.14417894,0.045960087,0.21606281],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99775463,0.000178919,0.00027338148,0.0004907149,0.00097177684,0.00033070578],"domain_scores_gemma":[0.9937782,0.00093644595,0.00029595202,0.0025323234,0.001979575,0.00047748405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016781858,0.0016451676,0.0012639939,0.0018851837,0.0012404746,0.0057783173,0.006863098,0.001519828,0.07205056],"category_scores_gemma":[0.00720215,0.0012476494,0.0012970676,0.0024547414,0.0010240742,0.009809266,0.0062142,0.0025710845,0.045749217],"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":[0.0031810165,0.0005456772,0.0068908758,0.0019173797,0.00035602125,0.00070042355,0.0007807652,0.0063967165,0.01978284,0.08858516,0.58031815,0.290545],"study_design_scores_gemma":[0.00044908028,0.0004507771,0.0018833578,0.00024178863,0.0002097395,0.0010877039,0.00040575798,0.055743035,0.039884303,0.04018129,0.85923725,0.00022591728],"about_ca_topic_score_codex":0.0048176586,"about_ca_topic_score_gemma":0.0049220314,"teacher_disagreement_score":0.07205056,"about_ca_system_score_codex":0.0011348482,"about_ca_system_score_gemma":0.0029119982,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W1545879303","doi":"10.14778/2732219.2732221","title":"More is simpler","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":69,"is_retracted":false,"has_abstract":true,"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; Leverage (statistics); Similarity (geometry); Computation; Theoretical computer science; Cluster analysis; Similarity measure; Artificial intelligence; Algorithm","score_opus":0.007098670208723978,"score_gpt":0.2316398325702858,"score_spread":0.22454116236156182,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1545879303","genre_codex":"other","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":[0.0068021645,0.007875624,0.10079852,0.09610492,0.010222321,0.00030192366,0.0018867367,0.00380033,0.7722074],"genre_scores_gemma":[0.16759361,0.012841973,0.08290876,0.05036479,0.008208821,0.00055760937,0.0036251063,0.0055561457,0.6683431],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939772,0.0014695416,0.00033902176,0.0017228969,0.0019692807,0.0005219832],"domain_scores_gemma":[0.9923179,0.0019660403,0.00047419837,0.003063273,0.0014387721,0.0007397174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004833234,0.0017825874,0.0013726164,0.0027264694,0.0041361437,0.010464513,0.002834327,0.004021485,0.26175097],"category_scores_gemma":[0.019687796,0.0006820217,0.0014706153,0.0027619675,0.0060283225,0.024360886,0.008333485,0.0050358768,0.10230602],"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":[0.00012865769,0.00010325088,0.00075756427,0.00045053597,0.000057173726,0.00025812068,0.0013838933,0.0008297442,0.0009832969,0.5948128,0.21066466,0.1895703],"study_design_scores_gemma":[0.000023665858,0.000033367432,0.00028614126,0.00017849718,0.000022005386,0.00021639495,0.00056419434,0.0007085851,0.00032008643,0.15774533,0.83987236,0.000029389761],"about_ca_topic_score_codex":0.004317719,"about_ca_topic_score_gemma":0.004596435,"teacher_disagreement_score":0.26175097,"about_ca_system_score_codex":0.0031222606,"about_ca_system_score_gemma":0.0030796488,"threshold_uncertainty_score":0.8756442},"labels":[],"label_agreement":null},{"id":"W1656389077","doi":"10.14778/2735479.2735485","title":"Rapid sampling for visualizations with ordering guarantees","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":102,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute of General Medical Sciences","keywords":"Computer science; Bar chart; Visualization; Focus (optics); Sampling (signal processing); Chart; Theoretical computer science; Property (philosophy); Algorithm; Data mining; Mathematics; Statistics; Computer vision","score_opus":0.07172555247471732,"score_gpt":0.31408397678660194,"score_spread":0.24235842431188462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1656389077","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":[0.008890252,0.00027543167,0.9871329,0.00031362937,0.000038942584,0.00009527728,0.00017396142,0.0021361164,0.00094349164],"genre_scores_gemma":[0.21971178,0.00041234386,0.7756688,0.00024456347,0.00009830163,0.00043119176,0.00095465715,0.00096636044,0.0015121107],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939621,0.0025673562,0.00041379355,0.0008388458,0.0019226804,0.0002951667],"domain_scores_gemma":[0.96959645,0.018312985,0.001699146,0.0066467053,0.003047841,0.00069687405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006365862,0.0013261682,0.00129078,0.0018866578,0.0011449248,0.0034972865,0.0017113334,0.0016768171,0.007357129],"category_scores_gemma":[0.06189564,0.0010725622,0.0011226502,0.0019401339,0.0016847526,0.0049540605,0.0035153367,0.0025817684,0.002221414],"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":[0.0013788164,0.00019710742,0.006275727,0.0007865188,0.00012470197,0.0004260461,0.0016697529,0.28800246,0.03286411,0.20665623,0.018134363,0.4434842],"study_design_scores_gemma":[0.00012861853,0.00012637692,0.0005944074,0.00006409738,0.000018385936,0.00021346424,0.0001958249,0.80386484,0.009892378,0.17420411,0.010663487,0.000033944703],"about_ca_topic_score_codex":0.001875183,"about_ca_topic_score_gemma":0.0024273135,"teacher_disagreement_score":0.007357129,"about_ca_system_score_codex":0.0013474046,"about_ca_system_score_gemma":0.001490287,"threshold_uncertainty_score":0.033666313},"labels":[],"label_agreement":null},{"id":"W1786535492","doi":"10.14778/2824032.2824111","title":"Gain control over your integration evaluations","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"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; Data integration; Generality; Scalability; Information integration; Schema (genetic algorithms); System integration; Metadata; Reuse; Schema evolution; Data mining; Information retrieval; Database schema; Database; World Wide Web; Database design","score_opus":0.056228122232949214,"score_gpt":0.30801182340339606,"score_spread":0.25178370117044685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1786535492","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":[0.075501926,0.0022843038,0.71824515,0.021824123,0.0030698897,0.0019461545,0.002752355,0.040869787,0.13350628],"genre_scores_gemma":[0.6144798,0.0005733819,0.33437303,0.00552293,0.0012503166,0.0026911558,0.0025671278,0.018992985,0.019549279],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.8949157,0.055318765,0.0068593114,0.012774334,0.027001645,0.0031302613],"domain_scores_gemma":[0.627287,0.18423232,0.011584573,0.1190653,0.051128108,0.006702751],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.10638132,0.0025205356,0.002459399,0.0029746238,0.0017481452,0.012176208,0.0035528166,0.003157414,0.023019178],"category_scores_gemma":[0.47708917,0.0014017867,0.0011883507,0.0026558482,0.0024422677,0.018815063,0.010255502,0.006465422,0.011669663],"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":[0.003019834,0.00074845634,0.0134849185,0.00084860926,0.00039927984,0.00034764255,0.0019946753,0.013315176,0.0104397405,0.103384264,0.12835647,0.723661],"study_design_scores_gemma":[0.0015869576,0.00092455605,0.0115912,0.0015466665,0.00045891092,0.00070285355,0.002129319,0.314061,0.045361947,0.3192088,0.30188146,0.00054629816],"about_ca_topic_score_codex":0.0013333053,"about_ca_topic_score_gemma":0.001436534,"teacher_disagreement_score":0.10638132,"about_ca_system_score_codex":0.002603085,"about_ca_system_score_gemma":0.0034402006,"threshold_uncertainty_score":0.5626049},"labels":[],"label_agreement":null},{"id":"W1828805609","doi":"10.14778/2733085.2733086","title":"Show me the money","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Heuristics; Scalability; Profitability index; Revenue; Exploit; Maximization; Revenue model; Mathematical optimization; Database; Economics; Computer security","score_opus":0.012291976906149613,"score_gpt":0.2133033050567286,"score_spread":0.20101132815057898,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1828805609","genre_codex":"other","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":[0.0078655295,0.00906173,0.00906782,0.088197865,0.018223898,0.00019270614,0.004241359,0.0028689716,0.86028004],"genre_scores_gemma":[0.050912324,0.005784738,0.005911427,0.025240295,0.0035527276,0.0001163723,0.002474328,0.0012483522,0.9047594],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99895144,0.00026999658,0.000047839785,0.00018987848,0.00037327158,0.00016751444],"domain_scores_gemma":[0.99797744,0.0003605918,0.0001735759,0.00033106242,0.00053866237,0.00061867293],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00096676894,0.0008764802,0.00068000134,0.0010412927,0.0026056203,0.006740739,0.0012211232,0.0028156783,0.49669352],"category_scores_gemma":[0.00828778,0.0003112383,0.0005741272,0.001344375,0.0010482186,0.0066156248,0.0033925083,0.0024086803,0.27074242],"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":[0.00009553772,0.000030837142,0.000976986,0.00012326143,0.000023704322,0.00035185285,0.00028907563,0.00010532095,0.00033083648,0.013733154,0.8741097,0.109829776],"study_design_scores_gemma":[0.000007929177,0.000019590698,0.0004359434,0.000070754526,0.000009232948,0.0004890567,0.0002981824,0.00012026089,0.00013348152,0.0038563337,0.99454606,0.000013183],"about_ca_topic_score_codex":0.0024110475,"about_ca_topic_score_gemma":0.0036353064,"teacher_disagreement_score":0.49669352,"about_ca_system_score_codex":0.0010466669,"about_ca_system_score_gemma":0.00093730254,"threshold_uncertainty_score":0.71790564},"labels":[],"label_agreement":null},{"id":"W1875516392","doi":"10.14778/2752939.2752950","title":"Viral marketing meets social advertising","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Optimization and Search Problems","field":"Computer Science","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Viral marketing; Leverage (statistics); Click-through rate; Computer science; Advertising; Online advertising; Regret; Context (archaeology); Host (biology); Social network (sociolinguistics); Social media; Display advertising; Business; World Wide Web; The Internet; Artificial intelligence","score_opus":0.025921147798563318,"score_gpt":0.25511025738176607,"score_spread":0.22918910958320277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1875516392","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":[0.03559246,0.0062463954,0.83401126,0.016080882,0.0010260851,0.00076569844,0.0012565725,0.0011020191,0.10391862],"genre_scores_gemma":[0.7498981,0.007162332,0.1862942,0.0036437411,0.0034534868,0.0009976227,0.0010617796,0.0006606945,0.046828035],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99545324,0.0019848815,0.00015864994,0.0010053682,0.00080134603,0.0005965238],"domain_scores_gemma":[0.9853967,0.011344717,0.00082288566,0.0010180809,0.000709547,0.0007081963],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035474033,0.0032623222,0.003941571,0.0011312832,0.0021111097,0.0071483566,0.002689214,0.006686685,0.022976922],"category_scores_gemma":[0.018014919,0.0012841155,0.0016395248,0.00206267,0.0027156956,0.008650465,0.0032737292,0.007013941,0.004527421],"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":[0.00039887056,0.0004955638,0.0010348056,0.0010644149,0.0001967571,0.0003747022,0.00031531585,0.24122716,0.0026848533,0.6351656,0.032508254,0.084533766],"study_design_scores_gemma":[0.000103485974,0.00018212011,0.0002697467,0.00006716784,0.00004209153,0.00026086185,0.000080487574,0.508438,0.0006213409,0.47280893,0.017086275,0.00003949716],"about_ca_topic_score_codex":0.0029423258,"about_ca_topic_score_gemma":0.002642736,"teacher_disagreement_score":0.022976922,"about_ca_system_score_codex":0.004023502,"about_ca_system_score_gemma":0.002997267,"threshold_uncertainty_score":0.076865494},"labels":[],"label_agreement":null},{"id":"W1967167578","doi":"10.14778/1453856.1453980","title":"Discovering data quality rules","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":268,"is_retracted":false,"has_abstract":true,"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; Data quality; Consistency (knowledge bases); Context (archaeology); Quality (philosophy); Data mining; Data integrity; Set (abstract data type); Data consistency; Process (computing); Database; Artificial intelligence; Programming language; Engineering","score_opus":0.49777141101266814,"score_gpt":0.451736219472416,"score_spread":0.04603519154025215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967167578","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":[0.09791879,0.0024853484,0.8449972,0.0059248866,0.0002697595,0.0021822895,0.026526932,0.011324405,0.008370271],"genre_scores_gemma":[0.21355699,0.0008566868,0.75219357,0.00095190626,0.00010547928,0.0007087578,0.029345874,0.0007992009,0.0014815],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9650422,0.0049808673,0.0052169454,0.005815096,0.01771588,0.0012289743],"domain_scores_gemma":[0.8815446,0.070961155,0.009725274,0.014737435,0.021213263,0.0018182913],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015119768,0.0018133466,0.0021014563,0.011942493,0.001823355,0.0073477565,0.004192876,0.0023776633,0.0022707505],"category_scores_gemma":[0.10615004,0.0015721194,0.0039983294,0.005548572,0.0018340886,0.008104751,0.0046804603,0.0039412472,0.0011317234],"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":[0.0005732684,0.0009830564,0.16744779,0.0036369914,0.0010304921,0.0034843583,0.0027495872,0.052523207,0.015778063,0.06386509,0.035902638,0.65202546],"study_design_scores_gemma":[0.00019053917,0.0003615384,0.024179332,0.001836159,0.00078768167,0.0028443085,0.002675211,0.6406669,0.0491715,0.15067647,0.12632138,0.00028893835],"about_ca_topic_score_codex":0.008356454,"about_ca_topic_score_gemma":0.011226863,"teacher_disagreement_score":0.015119768,"about_ca_system_score_codex":0.0028084274,"about_ca_system_score_gemma":0.0065222457,"threshold_uncertainty_score":0.079961896},"labels":[],"label_agreement":null},{"id":"W1967348488","doi":"10.14778/1920841.1920978","title":"Identifying, attributing and describing spatial bursts","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Terabyte; Scalability; Computer science; Task (project management); Social media; Information retrieval; Scale (ratio); Data science; Data mining; World Wide Web; Database; Geography; Cartography","score_opus":0.02560125426761936,"score_gpt":0.2604515705919254,"score_spread":0.23485031632430606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967348488","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":[0.56061107,0.0017956778,0.41492668,0.0011028193,0.00015167073,0.000483767,0.013524048,0.0039744154,0.00342991],"genre_scores_gemma":[0.7557122,0.0011262156,0.22802484,0.00011832237,0.00022331596,0.0003144527,0.01270484,0.00023282455,0.0015429313],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985782,0.00021419562,0.0002527905,0.0003922191,0.0004315046,0.00013104346],"domain_scores_gemma":[0.9922272,0.0036201493,0.0016951385,0.0011117477,0.0010618027,0.00028400344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015688691,0.0010439517,0.0007646883,0.010933805,0.0006846615,0.0024888192,0.0011712332,0.0010503023,0.0005028618],"category_scores_gemma":[0.008214758,0.00050459825,0.00058493525,0.007631849,0.0005804858,0.0033582156,0.0019518195,0.000779626,0.00045903237],"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":[0.00067109044,0.00046537773,0.450899,0.00077632855,0.0002982401,0.0018250559,0.0044110394,0.058928996,0.028732048,0.011775994,0.012183026,0.42903388],"study_design_scores_gemma":[0.00005269243,0.00024012124,0.13560632,0.00015390902,0.00021343955,0.0015735744,0.0057515935,0.76936734,0.027125329,0.032432683,0.02735558,0.00012744621],"about_ca_topic_score_codex":0.006783954,"about_ca_topic_score_gemma":0.0102327345,"teacher_disagreement_score":0.010933805,"about_ca_system_score_codex":0.0008888782,"about_ca_system_score_gemma":0.0011145805,"threshold_uncertainty_score":0.013488948},"labels":[],"label_agreement":null},{"id":"W1967654850","doi":"10.14778/1687627.1687642","title":"Measure-driven keyword-query expansion","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"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; Exploit; Pruning; Information retrieval; Query expansion; Web search query; Context (archaeology); Set (abstract data type); Process (computing); Domain (mathematical analysis); Measure (data warehouse); Focus (optics); Query optimization; Data mining; Word (group theory); Search engine; Mathematics","score_opus":0.012523221015441705,"score_gpt":0.21224724329768602,"score_spread":0.1997240222822443,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967654850","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":[0.05700483,0.00062108744,0.93738055,0.00021703732,0.00002723077,0.0003342178,0.0003773574,0.0021253692,0.0019123307],"genre_scores_gemma":[0.4322627,0.0002971892,0.5631849,0.00020237023,0.00006438388,0.00030963853,0.0013052908,0.0002005766,0.0021730089],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965932,0.001110728,0.00027596572,0.0006044864,0.0011831922,0.00023242655],"domain_scores_gemma":[0.994218,0.0033182385,0.00041440266,0.0009573521,0.000966171,0.00012588855],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027734411,0.00095277664,0.0024637056,0.0020163064,0.0006842415,0.0013754902,0.0025318782,0.0012141435,0.0022160849],"category_scores_gemma":[0.012136029,0.0005287729,0.0011661052,0.0033137202,0.00087081874,0.002724686,0.0024395776,0.0010481557,0.0010295224],"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":[0.001182579,0.0005762913,0.005710719,0.00075316895,0.00019031692,0.0004945683,0.00087842334,0.30837682,0.04444588,0.02566034,0.00971801,0.6020129],"study_design_scores_gemma":[0.000040001967,0.000114475224,0.000634531,0.000011307547,0.000029793608,0.00032412584,0.00010251467,0.97998637,0.0074210055,0.010049246,0.001259642,0.000026997208],"about_ca_topic_score_codex":0.0035818713,"about_ca_topic_score_gemma":0.00531679,"teacher_disagreement_score":0.0035818713,"about_ca_system_score_codex":0.0010919226,"about_ca_system_score_gemma":0.001294963,"threshold_uncertainty_score":0.014667511},"labels":[],"label_agreement":null},{"id":"W1972924401","doi":"10.14778/1920841.1920947","title":"Building ranked mashups of unstructured sources with uncertain information","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Mashup; Computer science; Ranking (information retrieval); Information retrieval; Rank (graph theory); Probabilistic logic; Information extraction; Semantics (computer science); Database; World Wide Web; Data mining; Web service; Artificial intelligence; Web modeling","score_opus":0.00427333096314046,"score_gpt":0.19202117821881115,"score_spread":0.1877478472556707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1972924401","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":[0.024237052,0.00015179979,0.9717838,0.00021563869,0.000019547519,0.00013662405,0.000295886,0.0017793247,0.0013802178],"genre_scores_gemma":[0.2378239,0.00027862913,0.7576227,0.00013834864,0.00006152817,0.00024863554,0.0013815673,0.00037276238,0.0020719313],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99678636,0.00094451476,0.00022365234,0.0006373064,0.0011926505,0.00021549068],"domain_scores_gemma":[0.99457955,0.0028237356,0.00048892037,0.001318713,0.0005619757,0.00022708901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027312343,0.0010949888,0.0010676103,0.0015574655,0.0012063414,0.0025262006,0.0017419992,0.0009075924,0.0015057792],"category_scores_gemma":[0.010407234,0.0008647961,0.001200409,0.0023550526,0.0011404254,0.005428982,0.0038529632,0.0014753584,0.00074027386],"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":[0.0011629236,0.00051600364,0.005620344,0.0009827641,0.0006406984,0.002572374,0.0032812299,0.4397057,0.07857477,0.13966972,0.013482036,0.31379148],"study_design_scores_gemma":[0.000062795356,0.00014025315,0.00054434146,0.000039779872,0.00011928483,0.00025554118,0.0006132044,0.8559306,0.032505155,0.095473774,0.014250877,0.00006442395],"about_ca_topic_score_codex":0.002048713,"about_ca_topic_score_gemma":0.003776427,"teacher_disagreement_score":0.0027312343,"about_ca_system_score_codex":0.00064361695,"about_ca_system_score_gemma":0.0010153252,"threshold_uncertainty_score":0.014444292},"labels":[],"label_agreement":null},{"id":"W1980185632","doi":"10.14778/2350229.2350241","title":"Fundamentals of order dependencies","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Functional dependency; Lexicographical order; Axiom; Dependency theory (database theory); Tuple; Computer science; Inference; Dependency (UML); Set (abstract data type); Query optimization; Theoretical computer science; Mathematics; Data mining; Artificial intelligence; Discrete mathematics; Relational database; Programming language; Combinatorics","score_opus":0.015635755890120978,"score_gpt":0.22162146837522673,"score_spread":0.20598571248510575,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1980185632","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":[0.008927558,0.006125013,0.9011064,0.0048802136,0.0010478097,0.00021899983,0.002230905,0.00088880997,0.074574254],"genre_scores_gemma":[0.279914,0.015019823,0.65918493,0.0042169928,0.0035118188,0.0010708921,0.003837719,0.0008717074,0.03237215],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.994282,0.0012267185,0.00074984547,0.0012684814,0.0019738583,0.00049912214],"domain_scores_gemma":[0.99017626,0.0057252753,0.0005882928,0.0016193341,0.0016501345,0.00024064412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048755934,0.0012849596,0.0010253678,0.003275057,0.0029505685,0.0044853473,0.0024598646,0.0022559906,0.009872815],"category_scores_gemma":[0.01207822,0.0013843088,0.0024940227,0.004229862,0.0072171506,0.012880397,0.004340912,0.005852221,0.00369793],"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":[0.000014314301,0.00001471411,0.00023406083,0.00011084408,0.000012622884,0.000101242615,0.0002147517,0.0016236231,0.00040281788,0.9773286,0.0038158458,0.016126627],"study_design_scores_gemma":[0.000012581411,0.000016179058,0.00014522653,0.00005568294,0.00001832637,0.00024109842,0.000048239504,0.0038569197,0.000734066,0.9276745,0.06717586,0.000021365973],"about_ca_topic_score_codex":0.005617953,"about_ca_topic_score_gemma":0.0034436975,"teacher_disagreement_score":0.009872815,"about_ca_system_score_codex":0.0027054497,"about_ca_system_score_gemma":0.0027499553,"threshold_uncertainty_score":0.033027887},"labels":[],"label_agreement":null},{"id":"W1981001739","doi":"10.14778/1454159.1454200","title":"Semandaq","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"","keywords":"Computer science; Relational database; SQL; Quality (philosophy); Database; Data mining; User interface; Interface (matter); Data quality; Programming language; Engineering; Operating system; Metric (unit)","score_opus":0.19753170774097992,"score_gpt":0.36462387965471404,"score_spread":0.16709217191373413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981001739","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":[0.0059783524,0.00083827914,0.68141776,0.0011316633,0.0003828267,0.0004044065,0.015882572,0.2836559,0.010308139],"genre_scores_gemma":[0.09953402,0.001125266,0.77640474,0.0017305467,0.00021245293,0.0007050423,0.060979154,0.044859312,0.014449507],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99367934,0.0014303188,0.00078876526,0.0013330238,0.0024537486,0.00031484818],"domain_scores_gemma":[0.9818186,0.006126603,0.0009862324,0.007115166,0.0035002541,0.00045316917],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008801186,0.0013176543,0.0013069933,0.0026475566,0.0010359116,0.006549304,0.004904649,0.0013822885,0.023924958],"category_scores_gemma":[0.022697631,0.0015493806,0.0021239128,0.0025557228,0.001481745,0.009941374,0.006264762,0.0025107001,0.009237615],"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":[0.0016289379,0.00031968846,0.009097397,0.0028615207,0.00040225653,0.00070522714,0.0017591246,0.009857164,0.019501662,0.09905809,0.37756753,0.47724143],"study_design_scores_gemma":[0.00036681333,0.00020169589,0.002613459,0.00026499608,0.00014407697,0.0009254937,0.00036769806,0.08870943,0.02532764,0.07789332,0.8029523,0.00023322857],"about_ca_topic_score_codex":0.0057245796,"about_ca_topic_score_gemma":0.0046405727,"teacher_disagreement_score":0.023924958,"about_ca_system_score_codex":0.0014472468,"about_ca_system_score_gemma":0.0031547274,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W1981578383","doi":"10.14778/2536336.2536345","title":"Discovering linkage points over web data","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":41,"is_retracted":false,"has_abstract":true,"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 matching; Schema (genetic algorithms); Data integration; Information retrieval; Data mining; Linked data; Linkage (software); Star schema; Database schema; Semi-structured model; Semantic Web; Database design","score_opus":0.1633565029719125,"score_gpt":0.3784563459656941,"score_spread":0.2150998429937816,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981578383","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":[0.28949118,0.001773605,0.6986631,0.0007150124,0.000043504595,0.0007180125,0.002406189,0.0034916846,0.00269773],"genre_scores_gemma":[0.41379312,0.00094834575,0.58015054,0.00009001524,0.000041293737,0.00025770807,0.0038075852,0.00018252083,0.00072891527],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9907555,0.0021227004,0.0010091561,0.0015660843,0.004121761,0.00042487198],"domain_scores_gemma":[0.98439085,0.008619614,0.0021413057,0.001899274,0.002524655,0.00042437864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0076551847,0.0007929338,0.0016529519,0.018619763,0.0021604132,0.0050965534,0.0019565672,0.0021541088,0.0011050018],"category_scores_gemma":[0.0347103,0.0008109922,0.0015285908,0.016456425,0.0009282562,0.009267304,0.0047444184,0.0013509542,0.00067070394],"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":[0.0005510496,0.00084188255,0.10967456,0.0010816917,0.0006717798,0.0016395339,0.0031808547,0.07494428,0.012623176,0.03942991,0.0053371247,0.7500242],"study_design_scores_gemma":[0.00009022862,0.0002982656,0.021031523,0.0002288184,0.00032827276,0.0012493717,0.003720334,0.7906955,0.028199196,0.14077057,0.0132629955,0.000124915],"about_ca_topic_score_codex":0.003481897,"about_ca_topic_score_gemma":0.0037066275,"teacher_disagreement_score":0.018619763,"about_ca_system_score_codex":0.0010935799,"about_ca_system_score_gemma":0.0018752561,"threshold_uncertainty_score":0.040484965},"labels":[],"label_agreement":null},{"id":"W1982177147","doi":"10.14778/2002974.2002976","title":"gStore","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":263,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"SPARQL; RDF; Computer science; Information retrieval; Named graph; RDF Schema; Linked data; RDF query language; Scalability; RDF/XML; Pruning; Database; Semantic Web; Web search query; Web query classification; Search engine","score_opus":0.03354416418142819,"score_gpt":0.19815301334233806,"score_spread":0.16460884916090987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1982177147","genre_codex":"other","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":[0.008428562,0.0024663706,0.2025956,0.002551394,0.0020783355,0.00066714175,0.06884974,0.3455576,0.36680526],"genre_scores_gemma":[0.07981606,0.002431221,0.20349263,0.0037691076,0.0005277935,0.0009835127,0.30162504,0.050609123,0.35674548],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99881554,0.00013577167,0.00009864494,0.00034318454,0.00043169546,0.00017518323],"domain_scores_gemma":[0.9989249,0.00014059167,0.000052552856,0.0005193547,0.00026204184,0.00010047681],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00076451927,0.0013257931,0.0012239767,0.0015654557,0.0011444603,0.0034436043,0.002953996,0.0015029187,0.23063797],"category_scores_gemma":[0.0025507302,0.000959144,0.001425723,0.002620814,0.0004822961,0.0045073316,0.0033973064,0.0019733084,0.18472062],"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":[0.00044972013,0.000150872,0.001271919,0.0007304072,0.000118695876,0.00024782904,0.00016214646,0.002182583,0.0070225066,0.023492102,0.72792584,0.23624532],"study_design_scores_gemma":[0.00012163163,0.00006012377,0.0006603679,0.000060392947,0.000034674435,0.0003059394,0.000070359216,0.0115190325,0.0084865615,0.017971072,0.9606709,0.000038963175],"about_ca_topic_score_codex":0.0030693277,"about_ca_topic_score_gemma":0.004725919,"teacher_disagreement_score":0.23063797,"about_ca_system_score_codex":0.0009796979,"about_ca_system_score_gemma":0.0015264023,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W1986041862","doi":"10.14778/1920841.1920982","title":"An access cost-aware approach for object retrieval over multiple sources","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"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; Overhead (engineering); Probabilistic logic; Object (grammar); Source code; Selection (genetic algorithm); Data mining; Information retrieval; Data source; Database; Artificial intelligence; Programming language","score_opus":0.024767759603829508,"score_gpt":0.279069790039691,"score_spread":0.2543020304358615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1986041862","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":[0.011602663,0.00039846747,0.9861456,0.00021231279,0.000021244596,0.00010898451,0.00009745097,0.0006288899,0.0007844221],"genre_scores_gemma":[0.2986858,0.00059046934,0.6960201,0.00018639822,0.000182547,0.00031788534,0.0005868316,0.00023905333,0.0031908266],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961377,0.0007155513,0.00028942194,0.0006456359,0.0019584065,0.00025337993],"domain_scores_gemma":[0.9928005,0.002949278,0.0005798573,0.001965022,0.0014970471,0.00020842628],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028619163,0.0010559638,0.0022174362,0.0045147724,0.0012621406,0.0029663045,0.0037889064,0.0015330104,0.0022786236],"category_scores_gemma":[0.011820064,0.0007677681,0.0014289335,0.006404118,0.0009971847,0.006970945,0.003256321,0.0014054392,0.0007588859],"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":[0.0006988843,0.00050610455,0.004183117,0.0003937241,0.00030966508,0.00038135977,0.000568464,0.23961498,0.031368665,0.056859888,0.0069395434,0.6581755],"study_design_scores_gemma":[0.000057330388,0.00015776455,0.0007560142,0.000013634053,0.00011024038,0.00043047112,0.000114516915,0.9551309,0.009184999,0.03017065,0.0038197995,0.000053766133],"about_ca_topic_score_codex":0.003236054,"about_ca_topic_score_gemma":0.004235825,"teacher_disagreement_score":0.0045147724,"about_ca_system_score_codex":0.0013806531,"about_ca_system_score_gemma":0.0018311954,"threshold_uncertainty_score":0.015135467},"labels":[],"label_agreement":null},{"id":"W1991635064","doi":"10.14778/2047485.2047492","title":"A data-based approach to social influence maximization","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":426,"is_retracted":false,"has_abstract":true,"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; Submodular set function; Computer science; Scalability; Perspective (graphical); Set (abstract data type); Function (biology); Expectation–maximization algorithm; Social graph; Social network (sociolinguistics); Mathematical optimization; Theoretical computer science; Artificial intelligence; Social media; Mathematics; Maximum likelihood; World Wide Web; Statistics","score_opus":0.056720518850135716,"score_gpt":0.26189961063934397,"score_spread":0.20517909178920826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991635064","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":[0.0048381253,0.00026479195,0.9925609,0.00048304928,0.000028674865,0.00009538738,0.00033931804,0.00020291886,0.0011868044],"genre_scores_gemma":[0.4564472,0.00117393,0.53448725,0.00049637357,0.00054440193,0.0008956415,0.001760273,0.00024898772,0.003946037],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965501,0.0012989673,0.0001898503,0.0009792532,0.00079906784,0.00018280443],"domain_scores_gemma":[0.98713976,0.0096063465,0.0008665477,0.00118428,0.00090348924,0.00029952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004309889,0.0019202571,0.0026284894,0.0026807727,0.00096552493,0.002829244,0.004028468,0.0027476521,0.0026755114],"category_scores_gemma":[0.022091039,0.0011041078,0.0018506125,0.0043221614,0.002263358,0.0051015024,0.0029673046,0.0038873984,0.0007443896],"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":[0.00012966551,0.00017342428,0.003062385,0.00033052504,0.00020306313,0.00016083666,0.00025735892,0.77988267,0.0015816392,0.13653482,0.005212234,0.072471425],"study_design_scores_gemma":[0.000011268304,0.000018291606,0.00014747893,0.000014170965,0.000012913803,0.000036753743,0.00001659135,0.93152034,0.0005871168,0.0663912,0.0012348779,0.000008946231],"about_ca_topic_score_codex":0.0038817415,"about_ca_topic_score_gemma":0.0041902927,"teacher_disagreement_score":0.004309889,"about_ca_system_score_codex":0.002906474,"about_ca_system_score_gemma":0.0015999059,"threshold_uncertainty_score":0.022793174},"labels":[],"label_agreement":null},{"id":"W2000482994","doi":"10.14778/1978665.1978666","title":"Similarity join size estimation using locality sensitive hashing","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Locality-sensitive hashing; Joins; Nearest neighbor search; Similarity (geometry); Join (topology); Computer science; Generalization; Data mining; Hash function; Range (aeronautics); Sampling (signal processing); Set (abstract data type); Pattern recognition (psychology); Algorithm; Artificial intelligence; Mathematics; Hash table","score_opus":0.05016504780970029,"score_gpt":0.27647349014964706,"score_spread":0.22630844233994676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2000482994","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":[0.042573906,0.00033568512,0.9552586,0.00010230898,0.000031454696,0.000089465124,0.0000785637,0.0009345375,0.00059547625],"genre_scores_gemma":[0.53954154,0.00022478879,0.45831797,0.00007185613,0.00011422843,0.00013576806,0.00052368006,0.00015961115,0.00091057824],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958404,0.00091472676,0.0002810998,0.0006017093,0.0022243734,0.00013770617],"domain_scores_gemma":[0.9910466,0.0047172178,0.0011080363,0.0016189818,0.0013242603,0.00018483817],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00287611,0.00045611436,0.0010964955,0.0018706778,0.0006545824,0.0013414308,0.001660332,0.0008683379,0.00088925054],"category_scores_gemma":[0.014922176,0.0003801673,0.00052579853,0.001808541,0.00069502345,0.0031674618,0.0016270129,0.00076004566,0.00055829127],"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":[0.0006756032,0.000312864,0.0122169275,0.00028366523,0.00013901565,0.00018196797,0.0005774482,0.13000971,0.074121185,0.019125275,0.0049291654,0.75742716],"study_design_scores_gemma":[0.00003085837,0.00020604025,0.002706671,0.000011931423,0.000022101887,0.00041072338,0.00018249873,0.93973136,0.041970197,0.01207782,0.0026014887,0.000048346006],"about_ca_topic_score_codex":0.0010486231,"about_ca_topic_score_gemma":0.00093502627,"teacher_disagreement_score":0.00287611,"about_ca_system_score_codex":0.00057792733,"about_ca_system_score_gemma":0.0007878806,"threshold_uncertainty_score":0.01521045},"labels":[],"label_agreement":null},{"id":"W2000516574","doi":"10.14778/1920841.1920919","title":"Small domain randomization","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"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; Table (database); Partition (number theory); Randomization; Domain (mathematical analysis); Key (lock); Algorithm; Data mining; Theoretical computer science; Mathematics; Combinatorics","score_opus":0.01725406313730675,"score_gpt":0.22441536667375253,"score_spread":0.2071613035364458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2000516574","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":[0.0134130865,0.0004614633,0.9806662,0.00042048938,0.00012273673,0.00028931416,0.00033513008,0.0009648893,0.0033265671],"genre_scores_gemma":[0.46465948,0.0009336972,0.5221093,0.000877455,0.0003026052,0.0012621382,0.0012505206,0.00041521605,0.008189488],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9941088,0.0026006403,0.0004293079,0.0011021828,0.0014201183,0.0003389296],"domain_scores_gemma":[0.9803254,0.008085177,0.0011433009,0.008879115,0.0011887127,0.0003783342],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004747272,0.0007028041,0.0014509141,0.0012172724,0.0011730236,0.002037215,0.0025301834,0.0013333757,0.00570101],"category_scores_gemma":[0.02081015,0.00048768325,0.0014063315,0.0023536594,0.0022639376,0.0049949563,0.0036010013,0.0022693132,0.0019273254],"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":[0.00207696,0.0005760816,0.0049995817,0.00069167756,0.0003799194,0.00085808494,0.0006245428,0.14655119,0.04461044,0.42656326,0.017341342,0.35472688],"study_design_scores_gemma":[0.00043161123,0.0007750173,0.0011839967,0.00013734962,0.0001803513,0.0018051933,0.00020554263,0.46598867,0.045970015,0.4279112,0.05525582,0.00015525843],"about_ca_topic_score_codex":0.0004196975,"about_ca_topic_score_gemma":0.00040739132,"teacher_disagreement_score":0.00570101,"about_ca_system_score_codex":0.00077930075,"about_ca_system_score_gemma":0.0015668179,"threshold_uncertainty_score":0.025106251},"labels":[],"label_agreement":null},{"id":"W2004580781","doi":"10.14778/1920841.1921038","title":"Peer coordination through distributed triggers","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Peer-to-Peer Network Technologies","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University of Ottawa","funders":"","keywords":"Computer science; Consistency (knowledge bases); Distributed computing; Distributed database; Peer-to-peer; Set (abstract data type); Distributed management; Semantics (computer science); Eventual consistency; Database; Data consistency; Programming language; Consistency model","score_opus":0.011311263184876547,"score_gpt":0.23842975324750668,"score_spread":0.22711849006263013,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004580781","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":[0.016830625,0.00044800586,0.91665864,0.0014984417,0.00056526414,0.00027936048,0.00022932455,0.0047494546,0.05874077],"genre_scores_gemma":[0.56904364,0.0007793246,0.385316,0.0007583052,0.0004700012,0.0008041674,0.00059450674,0.0010202292,0.041213814],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9948002,0.0016508385,0.00034393356,0.001081602,0.0016632674,0.00046020435],"domain_scores_gemma":[0.99500847,0.0018128918,0.00037371178,0.0015826889,0.0007863244,0.00043585087],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036293454,0.0005054378,0.0006601674,0.0006651468,0.0016713598,0.0044095167,0.0020606706,0.0014048785,0.009465618],"category_scores_gemma":[0.00852273,0.00043341742,0.00062513124,0.00094647857,0.0021277121,0.004735798,0.005703752,0.0019435133,0.0027459462],"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":[0.0002444215,0.0000887124,0.00094007264,0.00021227912,0.00006200605,0.0007393617,0.002521597,0.013140209,0.011391609,0.899971,0.011955121,0.058733482],"study_design_scores_gemma":[0.00023298657,0.00022837977,0.0005523037,0.00010612996,0.00009374551,0.000825622,0.00068970677,0.102320686,0.022204762,0.52153367,0.3510956,0.00011642532],"about_ca_topic_score_codex":0.0017102682,"about_ca_topic_score_gemma":0.0011295258,"teacher_disagreement_score":0.009465618,"about_ca_system_score_codex":0.0008982153,"about_ca_system_score_gemma":0.0019360359,"threshold_uncertainty_score":0.031665623},"labels":[],"label_agreement":null},{"id":"W2005480575","doi":"10.14778/1920841.1920847","title":"Database replication","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Branco Weiss Fellowship – Society in Science; Eidgenössische Technische Hochschule Zürich; McGill University","keywords":"Computer science; Scalability; Replication (statistics); Distributed computing; Eventual consistency; Database transaction; Consistency (knowledge bases); Fault tolerance; Overhead (engineering); Distributed database; Cloud computing; Database; Transaction processing; Data consistency; Weak consistency; Concurrency control; Strong consistency; Consistency model; Operating system; Artificial intelligence","score_opus":0.009657613492116489,"score_gpt":0.23346822183864122,"score_spread":0.22381060834652472,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2005480575","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":[0.008233492,0.014366679,0.71671194,0.0059702243,0.009025815,0.003111641,0.00854926,0.02856207,0.20546885],"genre_scores_gemma":[0.22715679,0.015782895,0.49655867,0.006242869,0.005889824,0.0023872494,0.030326126,0.0051176534,0.21053791],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9860547,0.0022542402,0.0016062013,0.0023357945,0.0068485583,0.00090058055],"domain_scores_gemma":[0.97513753,0.0020740202,0.000869839,0.014462221,0.0064275526,0.001028782],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0067743505,0.001281474,0.0020268548,0.002715617,0.0028986202,0.009341591,0.0065892227,0.0025565934,0.03983166],"category_scores_gemma":[0.024221975,0.001038776,0.0020170156,0.0040296274,0.0013212903,0.008225502,0.008716768,0.0026845923,0.047846917],"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":[0.0004062108,0.00019257188,0.002042039,0.001641266,0.0002324496,0.00056023605,0.0006385805,0.0054664193,0.011520394,0.18711011,0.3147606,0.4754291],"study_design_scores_gemma":[0.00008849312,0.000118233955,0.00045056085,0.00019942004,0.00006227005,0.0011126631,0.00019967771,0.0066830935,0.005832228,0.047500394,0.9376683,0.000084716106],"about_ca_topic_score_codex":0.0017777905,"about_ca_topic_score_gemma":0.0012632145,"teacher_disagreement_score":0.03983166,"about_ca_system_score_codex":0.0017837019,"about_ca_system_score_gemma":0.0044910703,"threshold_uncertainty_score":0.13325018},"labels":[],"label_agreement":null},{"id":"W2005499394","doi":"10.14778/2168651.2168658","title":"Dense subgraph maintenance under streaming edge weight updates for real-time story identification","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":172,"is_retracted":false,"has_abstract":true,"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; Enhanced Data Rates for GSM Evolution; Identification (biology); Social media; Globe; Scale (ratio); Point (geometry); Data science; Edge device; Range (aeronautics); Social network (sociolinguistics); World Wide Web; Artificial intelligence; Geography; Mathematics; Engineering","score_opus":0.013209024702018898,"score_gpt":0.23841514356549856,"score_spread":0.22520611886347966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2005499394","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":[0.15189083,0.00081494084,0.8400802,0.000516991,0.00005622228,0.0002299817,0.0010928966,0.0041688755,0.0011490996],"genre_scores_gemma":[0.60486597,0.00035701977,0.38903454,0.00014493465,0.00011444364,0.00026421418,0.0034517786,0.00030941545,0.0014576628],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988111,0.00028219342,0.00010371855,0.00036745885,0.00033142872,0.0001040963],"domain_scores_gemma":[0.991293,0.004996122,0.0011464787,0.0015155581,0.0007943123,0.00025446675],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015171234,0.0009236693,0.001450579,0.002850521,0.0007931802,0.0012188345,0.0030249315,0.0012411394,0.0010363597],"category_scores_gemma":[0.017366368,0.00072985416,0.0006752348,0.0032707648,0.0008244656,0.0041648597,0.0016468767,0.0010888033,0.00052569254],"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":[0.00077173987,0.0003907945,0.015224704,0.0003176146,0.00019636838,0.00044005134,0.0006669966,0.3944598,0.01724005,0.008579812,0.008912885,0.55279917],"study_design_scores_gemma":[0.000028873119,0.000053949596,0.0012391957,0.000008827959,0.000029339106,0.0001529856,0.00008766979,0.9860271,0.00385686,0.0075433035,0.0009612164,0.000010631149],"about_ca_topic_score_codex":0.006720096,"about_ca_topic_score_gemma":0.010211146,"teacher_disagreement_score":0.006720096,"about_ca_system_score_codex":0.0009957345,"about_ca_system_score_gemma":0.0010678744,"threshold_uncertainty_score":0.01336199},"labels":[],"label_agreement":null},{"id":"W2006568248","doi":"10.14778/1920841.1921043","title":"Transforming XML documents as schemas evolve","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"DNA and Biological Computing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada); York University","funders":"","keywords":"XML Schema Editor; Computer science; Streaming XML; XML validation; Document Structure Description; Efficient XML Interchange; XML database; XML Schema (W3C); XSLT; Information retrieval; cXML; XML Encryption; RELAX NG; Programming language; Database; XML; World Wide Web","score_opus":0.006265382445630607,"score_gpt":0.23777636910250835,"score_spread":0.23151098665687775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2006568248","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":[0.021608356,0.000821915,0.93548894,0.0027866373,0.00073949463,0.0005748317,0.002857815,0.013786261,0.02133582],"genre_scores_gemma":[0.08619431,0.0022976918,0.88466954,0.0019497635,0.00021273787,0.0003556314,0.008345917,0.0044228053,0.011551587],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9940685,0.002277417,0.0008570234,0.00063555327,0.0019454182,0.00021611716],"domain_scores_gemma":[0.9848786,0.0060326084,0.0005797811,0.0059583816,0.002322681,0.00022799568],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007771497,0.000701971,0.0005228556,0.0023182218,0.0009708632,0.0063790446,0.002048159,0.0018337152,0.0039359513],"category_scores_gemma":[0.025639573,0.0009589265,0.0012157955,0.004706697,0.0012371883,0.007071348,0.0036544066,0.0031233248,0.0028150626],"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":[0.00029757,0.0003378067,0.0038613211,0.0006838894,0.0001886723,0.0018943504,0.007596706,0.01741024,0.022777105,0.3648923,0.055260204,0.52479976],"study_design_scores_gemma":[0.00008434837,0.00008075472,0.00084999553,0.00031490577,0.00009564633,0.0010600061,0.0016622664,0.04440784,0.03337934,0.19843906,0.71951103,0.00011481512],"about_ca_topic_score_codex":0.0017817722,"about_ca_topic_score_gemma":0.0012159918,"teacher_disagreement_score":0.007771497,"about_ca_system_score_codex":0.0010491566,"about_ca_system_score_gemma":0.0014189024,"threshold_uncertainty_score":0.041100085},"labels":[],"label_agreement":null},{"id":"W2014888296","doi":"10.14778/1920841.1921003","title":"TRAMP","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Provenance; Computer science; Schema (genetic algorithms); Transformation (genetics); Debugging; Tracing; Tramp; Suite; Information retrieval; Programming language; Database","score_opus":0.07102342555161244,"score_gpt":0.3438698995370234,"score_spread":0.27284647398541095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2014888296","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":[0.0070156115,0.000871211,0.675166,0.001122661,0.00036679662,0.00059127423,0.011428821,0.28176847,0.021669101],"genre_scores_gemma":[0.099791326,0.0014024806,0.80966926,0.00072565017,0.00015368748,0.00063272717,0.042388093,0.023750823,0.021486094],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9945286,0.0014199011,0.0005020675,0.0011628873,0.0020819865,0.00030455764],"domain_scores_gemma":[0.9815688,0.0049725855,0.00075501855,0.009629085,0.0027226754,0.00035188184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007516811,0.0013637202,0.00092413765,0.0027730614,0.0012330115,0.0041676215,0.0042427937,0.0012951283,0.020619486],"category_scores_gemma":[0.030518994,0.0012349496,0.0014369129,0.0026818628,0.0009433199,0.008547587,0.0053010373,0.0029433193,0.013885949],"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":[0.0012665596,0.00031998457,0.0057019135,0.0013276755,0.00030540538,0.00066303735,0.0012304028,0.013682993,0.009206719,0.08908156,0.23505518,0.64215857],"study_design_scores_gemma":[0.00024069315,0.00029471645,0.002376125,0.00038706954,0.00013476808,0.001668296,0.00036456226,0.18383278,0.02691809,0.09062997,0.6929849,0.00016810559],"about_ca_topic_score_codex":0.006125817,"about_ca_topic_score_gemma":0.0067365644,"teacher_disagreement_score":0.020619486,"about_ca_system_score_codex":0.0010871631,"about_ca_system_score_gemma":0.0036109604,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2022740398","doi":"10.14778/1880172.1880173","title":"Generating efficient execution plans for vertically partitioned XML databases","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"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; Query plan; Scalability; Database; Query optimization; XML database; Sargable; Distributed database; XML; Online aggregation; Relational database; Distributed computing; Information retrieval; Web search query; Search engine; Operating system","score_opus":0.0174271781966315,"score_gpt":0.25235560063649676,"score_spread":0.23492842243986525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2022740398","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":[0.13633391,0.00029033067,0.85711056,0.000311742,0.000020726015,0.00031444378,0.0003685017,0.0026879637,0.0025618013],"genre_scores_gemma":[0.4294305,0.00018609212,0.56708044,0.000054913795,0.000016631795,0.000323542,0.0010214814,0.00039805076,0.0014884068],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990274,0.00030229814,0.000072885254,0.00014680155,0.00032346358,0.00012714954],"domain_scores_gemma":[0.9982217,0.00093797443,0.00021076215,0.00032100358,0.00022886266,0.0000797733],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012150742,0.00057277887,0.00037623366,0.00066852866,0.0005906887,0.0008605595,0.0007935427,0.0003827284,0.0014536434],"category_scores_gemma":[0.0037079505,0.00042291344,0.0005683462,0.00079701963,0.00058561086,0.0012739216,0.0011319348,0.00073952967,0.00026252551],"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":[0.00060768373,0.00032523225,0.008031659,0.00029639885,0.00008712615,0.00053792546,0.00074655964,0.5947072,0.052482765,0.052048355,0.00802385,0.28210527],"study_design_scores_gemma":[0.00007568761,0.0000978436,0.0008561987,0.000017022414,0.00002994678,0.00007342355,0.00021665113,0.956482,0.018001806,0.020793136,0.003337697,0.000018532452],"about_ca_topic_score_codex":0.00476395,"about_ca_topic_score_gemma":0.008089596,"teacher_disagreement_score":0.00476395,"about_ca_system_score_codex":0.00091860397,"about_ca_system_score_gemma":0.0013660417,"threshold_uncertainty_score":0.009472489},"labels":[],"label_agreement":null},{"id":"W2033486966","doi":"10.14778/2095686.2095695","title":"Mining flipping correlations from large datasets with taxonomies","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Abstraction; Computer science; Range (aeronautics); Correlation; Data mining; Abstraction layer; Pattern recognition (psychology); Algorithm; Artificial intelligence; Theoretical computer science; Mathematics; Programming language","score_opus":0.029882836731726446,"score_gpt":0.21506378199320536,"score_spread":0.1851809452614789,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2033486966","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":[0.44331685,0.0034226095,0.53247297,0.0014617359,0.00022545665,0.0005371852,0.012051997,0.0033470034,0.0031642257],"genre_scores_gemma":[0.65563893,0.00080761896,0.3268916,0.00035227928,0.0001779879,0.00050220534,0.014771214,0.00018931487,0.0006688255],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99463814,0.0011804828,0.00075678417,0.0014130449,0.0016982189,0.00031328478],"domain_scores_gemma":[0.9691158,0.019024232,0.0045267697,0.0045862254,0.002112228,0.0006349121],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003354864,0.0012051315,0.0014932496,0.008330451,0.0012772243,0.0020216762,0.0014204044,0.0012281652,0.0014049199],"category_scores_gemma":[0.036938246,0.0007782503,0.0013157041,0.011049361,0.0011119621,0.004292661,0.0024500228,0.0018448541,0.0007479838],"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":[0.0009361564,0.0005311644,0.33186653,0.0020503704,0.0011265171,0.0059060273,0.0023682553,0.04372829,0.020935688,0.01867781,0.017807653,0.55406547],"study_design_scores_gemma":[0.00021802483,0.0005911396,0.12879652,0.0006830272,0.0006116425,0.008595135,0.00349083,0.55833715,0.017562319,0.24719535,0.033669528,0.0002493786],"about_ca_topic_score_codex":0.0016019014,"about_ca_topic_score_gemma":0.0037631248,"teacher_disagreement_score":0.008330451,"about_ca_system_score_codex":0.0005606584,"about_ca_system_score_gemma":0.0013839384,"threshold_uncertainty_score":0.017742395},"labels":[],"label_agreement":null},{"id":"W2056191114","doi":"10.14778/1920841.1921016","title":"MEET DB2","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"Computer science; Database; XML; Data migration; Source code; Executable; Process (computing); Software engineering; Operating system","score_opus":0.004885213329562108,"score_gpt":0.20289898918943194,"score_spread":0.19801377585986984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2056191114","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":[0.00828917,0.0011412674,0.10596487,0.0009637795,0.0005765121,0.00071860314,0.08019769,0.6121009,0.1900473],"genre_scores_gemma":[0.058008175,0.0011801057,0.14346565,0.0022475605,0.00037690377,0.0012439683,0.44545949,0.17776458,0.17025347],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99579394,0.00047764718,0.00037329856,0.00097652647,0.0019777857,0.00040080003],"domain_scores_gemma":[0.99544275,0.0006785772,0.0002704242,0.0014902327,0.0015591356,0.0005588511],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030837771,0.0023778984,0.0011659003,0.0028721606,0.0010581032,0.0064809243,0.004207209,0.0014429512,0.13965134],"category_scores_gemma":[0.006819768,0.0016975778,0.001320484,0.0026215157,0.0005178995,0.005910335,0.0040322538,0.0021981178,0.15163316],"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":[0.0012566254,0.00020043169,0.0026382143,0.0007474605,0.00012766986,0.00023582642,0.00041901195,0.001283275,0.0066377027,0.015109074,0.8138116,0.15753318],"study_design_scores_gemma":[0.00017664174,0.000110456866,0.0017708173,0.0000618389,0.000031215957,0.00034788015,0.00015685588,0.006681029,0.008200773,0.0047957725,0.97759104,0.00007569741],"about_ca_topic_score_codex":0.0055341306,"about_ca_topic_score_gemma":0.004671375,"teacher_disagreement_score":0.13965134,"about_ca_system_score_codex":0.0011364082,"about_ca_system_score_gemma":0.0022633797,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2058488627","doi":"10.14778/1687627.1687715","title":"Improved search for socially annotated data","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"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; Ranking (information retrieval); Scalability; Information retrieval; Annotation; Resource (disambiguation); Process (computing); Data mining; Similarity (geometry); Probabilistic logic; Cluster analysis; Machine learning; Database; Artificial intelligence","score_opus":0.045248350600693536,"score_gpt":0.2941844958611444,"score_spread":0.24893614526045088,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058488627","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":[0.045622483,0.0010801173,0.9433795,0.0012107451,0.000082817554,0.00019169287,0.0016816425,0.0032879026,0.0034631705],"genre_scores_gemma":[0.33114594,0.0005093559,0.6563451,0.00038456268,0.0001604278,0.00036002195,0.006554449,0.0004254505,0.004114712],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9914664,0.0037822218,0.000516582,0.0017141058,0.0021416498,0.00037900824],"domain_scores_gemma":[0.9887931,0.006161369,0.0008281162,0.0026660732,0.0013196849,0.00023176314],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0061486107,0.0015439522,0.002666839,0.005093778,0.0015177968,0.0033407134,0.0030192768,0.0026207878,0.0029110196],"category_scores_gemma":[0.028889991,0.0007921253,0.0015595455,0.006476096,0.0011052581,0.005898339,0.005734315,0.0014192795,0.0018609212],"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":[0.00076631334,0.00055547646,0.0068694837,0.0009426858,0.00038321165,0.00067179714,0.001652628,0.37940124,0.010985517,0.09670877,0.021757703,0.4793051],"study_design_scores_gemma":[0.000025880307,0.000055050063,0.00037748928,0.000018293556,0.000026866845,0.00007900206,0.00020129103,0.9592072,0.0015074342,0.035345767,0.0031347594,0.000020952193],"about_ca_topic_score_codex":0.008264619,"about_ca_topic_score_gemma":0.013073079,"teacher_disagreement_score":0.008264619,"about_ca_system_score_codex":0.0020246964,"about_ca_system_score_gemma":0.0032029096,"threshold_uncertainty_score":0.032517314},"labels":[],"label_agreement":null},{"id":"W2062482216","doi":"10.14778/1920841.1921029","title":"Efficient event processing through reconfigurable hardware for algorithmic trading","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Stock Market Forecasting Methods","field":"Decision Sciences","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Complex event processing; Computer science; Field-programmable gate array; Event (particle physics); Reconfigurable computing; Latency (audio); Predicate (mathematical logic); Computer architecture; Embedded system; Programming language","score_opus":0.08710940498047279,"score_gpt":0.38037972187073804,"score_spread":0.2932703168902653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2062482216","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":[0.086439736,0.0007971882,0.8865512,0.00034828967,0.0001861928,0.00009488147,0.00024172428,0.008402278,0.0169385],"genre_scores_gemma":[0.74431777,0.00036661557,0.25009856,0.00017335206,0.00006216848,0.00005847857,0.00032534648,0.00019244054,0.0044052117],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99971694,0.00004487379,0.000024397734,0.000057683414,0.000113187816,0.000042865162],"domain_scores_gemma":[0.99976796,0.00010979327,0.000020767326,0.00006459475,0.00002584358,0.000011112472],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024470995,0.00043539706,0.00022209297,0.00029617842,0.00017168054,0.00089956855,0.0008494254,0.00026431444,0.0047619976],"category_scores_gemma":[0.0006697675,0.0001978749,0.0002581201,0.00033426296,0.00025691773,0.0008336778,0.0003091873,0.0004309295,0.00072994246],"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":[0.0013053312,0.00020764649,0.0023803622,0.0003234319,0.00012299234,0.0007665934,0.00011147391,0.16504829,0.17496644,0.07248221,0.011732972,0.57055223],"study_design_scores_gemma":[0.00012690238,0.00022245418,0.00077520875,0.000025856612,0.000044605462,0.00026426514,0.000031994005,0.8597831,0.09234232,0.024804143,0.021545216,0.000034047178],"about_ca_topic_score_codex":0.00089962,"about_ca_topic_score_gemma":0.0015255518,"teacher_disagreement_score":0.0047619976,"about_ca_system_score_codex":0.00032743195,"about_ca_system_score_gemma":0.00036570837,"threshold_uncertainty_score":0.015930474},"labels":[],"label_agreement":null},{"id":"W2077780773","doi":"10.14778/1920841.1920948","title":"Computing closed skycubes","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Skyline; Linear subspace; Computer science; Representation (politics); Subspace topology; Computation; Theoretical computer science; Formal concept analysis; Closure (psychology); Space (punctuation); Algorithm; Data mining; Mathematics; Artificial intelligence","score_opus":0.006308341498846236,"score_gpt":0.20824132972052734,"score_spread":0.2019329882216811,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077780773","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":[0.057698105,0.00040574395,0.9366391,0.0002673534,0.000052891137,0.00012355713,0.0006141682,0.0009479034,0.0032511563],"genre_scores_gemma":[0.35312858,0.00053919014,0.6394295,0.0001568796,0.00007037857,0.00029064127,0.002866954,0.00039651062,0.0031213837],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99759346,0.0006650238,0.00020382211,0.0004144141,0.0008190798,0.00030415462],"domain_scores_gemma":[0.9953087,0.0024394866,0.00037990976,0.0009243291,0.0007603918,0.00018708759],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017591413,0.0006639026,0.0012198915,0.0016218502,0.0011957432,0.0026061707,0.001559206,0.0007579528,0.005189117],"category_scores_gemma":[0.009843668,0.0005085262,0.0012697097,0.002261128,0.001584271,0.0070981896,0.004009876,0.0011814805,0.00090464536],"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":[0.00052619696,0.00015243221,0.003971327,0.0007978328,0.00015719552,0.00039518852,0.0010277901,0.23999478,0.0143562835,0.5029123,0.009145083,0.22656353],"study_design_scores_gemma":[0.0000599346,0.00012651867,0.0004267893,0.000087361645,0.000027251923,0.00019106249,0.00051292503,0.45839983,0.010935871,0.51519156,0.014005478,0.000035376353],"about_ca_topic_score_codex":0.001800986,"about_ca_topic_score_gemma":0.002457627,"teacher_disagreement_score":0.005189117,"about_ca_system_score_codex":0.00089738553,"about_ca_system_score_gemma":0.0011239965,"threshold_uncertainty_score":0.017359316},"labels":[],"label_agreement":null},{"id":"W2077825087","doi":"10.14778/1920841.1921070","title":"Time for our field to grow up","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Wonder; Pride; Population; Field (mathematics); Haven; Set (abstract data type); Computer science; Data science; Political science; Sociology; Mathematics; Epistemology; Law","score_opus":0.06868736015691174,"score_gpt":0.3657685805251993,"score_spread":0.29708122036828755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2077825087","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035299966,0.019258149,0.0058263424,0.77671456,0.10545987,0.00025015828,0.0005085632,0.0014060317,0.087046355],"genre_scores_gemma":[0.05556832,0.027273828,0.022708984,0.4184822,0.05349887,0.0009679158,0.0018259027,0.0022185904,0.41745538],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98929954,0.0018066791,0.0004506395,0.0012734969,0.004276672,0.0028930458],"domain_scores_gemma":[0.94998235,0.0024888904,0.0012804252,0.0018784273,0.009610854,0.034758963],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0094442805,0.0023039335,0.0014539056,0.0024643946,0.012192611,0.026815087,0.0030059959,0.012013501,0.15636507],"category_scores_gemma":[0.029073026,0.00079613726,0.0017066875,0.00139104,0.0059222844,0.019084154,0.018531451,0.022151452,0.1279341],"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":[0.000055417542,0.00012011024,0.0003949999,0.00018283438,0.000016633461,0.0002215776,0.0011835082,0.00007408237,0.0005441473,0.018749712,0.92184174,0.05661529],"study_design_scores_gemma":[0.000016991462,0.000052448162,0.00024026229,0.00022016569,0.00000493609,0.00013716654,0.0020983797,0.00003120562,0.00007420058,0.0067180754,0.99038965,0.000016566322],"about_ca_topic_score_codex":0.0032834683,"about_ca_topic_score_gemma":0.003739461,"teacher_disagreement_score":0.9905557,"about_ca_system_score_codex":0.0042392453,"about_ca_system_score_gemma":0.018090487,"threshold_uncertainty_score":0.5230932},"labels":[],"label_agreement":null},{"id":"W2080132606","doi":"10.14778/1938545.1938547","title":"Automatic wrappers for large scale web extraction","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":135,"is_retracted":false,"has_abstract":true,"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; Noise (video); Scale (ratio); Noisy data; Extraction (chemistry); Data extraction; Data mining; Information extraction; Training set; Artificial intelligence; Machine learning; Information retrieval; Pattern recognition (psychology)","score_opus":0.024150020816393863,"score_gpt":0.24482076550129492,"score_spread":0.22067074468490105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2080132606","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":[0.0031549018,0.00019294542,0.9661502,0.000074524694,0.000031540334,0.000090067624,0.00072441215,0.029092854,0.000488524],"genre_scores_gemma":[0.047401983,0.00028773822,0.9415964,0.00015835953,0.000066819746,0.00019969858,0.005358897,0.003137903,0.0017921901],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99720925,0.0006067707,0.00040087255,0.00073269784,0.0008799857,0.0001704273],"domain_scores_gemma":[0.99242604,0.0019609737,0.000627239,0.0038188882,0.0010246101,0.00014217118],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026021963,0.0016685073,0.0016136818,0.004270602,0.0012979667,0.0021851072,0.0018846114,0.0014766214,0.0030215972],"category_scores_gemma":[0.011390459,0.001200313,0.0018287584,0.0040854034,0.0008858842,0.0043441863,0.0035201989,0.0020417792,0.0067998017],"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":[0.00022913846,0.00033475284,0.005182769,0.000811688,0.00033163125,0.00082561193,0.0005330854,0.025445588,0.049934812,0.017023679,0.04627993,0.85306734],"study_design_scores_gemma":[0.00006642007,0.00012907303,0.0032843056,0.00020809141,0.00021874384,0.0013361936,0.00019067098,0.61503506,0.20124885,0.09832586,0.079823345,0.00013338237],"about_ca_topic_score_codex":0.000926681,"about_ca_topic_score_gemma":0.001707474,"teacher_disagreement_score":0.004270602,"about_ca_system_score_codex":0.0004966936,"about_ca_system_score_gemma":0.0013461636,"threshold_uncertainty_score":0.013761878},"labels":[],"label_agreement":null},{"id":"W2098095723","doi":"10.14778/2732977.2732979","title":"Accordion","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":true,"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; Accordion; Partition (number theory); Heuristics; Bottleneck; Server; Distributed computing; Throughput; Distributed database; Database; Parallel computing; Operating system; Embedded system","score_opus":0.004902096397346088,"score_gpt":0.18554557464082128,"score_spread":0.18064347824347518,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098095723","genre_codex":"other","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":[0.011780718,0.0029304803,0.12486907,0.0044661313,0.0037415666,0.00078187545,0.014214568,0.16742036,0.6697952],"genre_scores_gemma":[0.10454515,0.0029443614,0.12089686,0.0036650754,0.0014132846,0.0009475531,0.058507696,0.033415332,0.67366457],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978116,0.00031562452,0.00012977971,0.0004723166,0.0009858033,0.00028479652],"domain_scores_gemma":[0.9967727,0.00045883094,0.00016202222,0.0009963633,0.0010846222,0.0005253346],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019852982,0.0014750375,0.00094474957,0.0015380438,0.0012019468,0.0045563406,0.0032078563,0.001792109,0.22887908],"category_scores_gemma":[0.005362922,0.000751793,0.0008381822,0.0013655969,0.00073716365,0.0042823097,0.004370139,0.002248552,0.14848438],"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":[0.0012093416,0.0002745814,0.0019764686,0.00053802744,0.000053163876,0.00041796814,0.00031062946,0.0027437971,0.010058479,0.033316135,0.64771104,0.30139038],"study_design_scores_gemma":[0.00014718976,0.00014645689,0.000887865,0.000102068894,0.000023002425,0.00036618082,0.000079120284,0.008552309,0.004479775,0.008482157,0.9766858,0.00004801772],"about_ca_topic_score_codex":0.0021331953,"about_ca_topic_score_gemma":0.0022986191,"teacher_disagreement_score":0.22887908,"about_ca_system_score_codex":0.00094503356,"about_ca_system_score_gemma":0.0018821482,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2100039551","doi":"10.14778/1687627.1687735","title":"Distribution based microdata anonymization","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Microdata (statistics); Computer science; Data anonymization; Closeness; Data mining; Aggregate (composite); Variety (cybernetics); Generalization; Information privacy; Artificial intelligence; Mathematics","score_opus":0.015626811778433624,"score_gpt":0.2368745563587915,"score_spread":0.2212477445803579,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100039551","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":[0.01955444,0.00033230087,0.97216445,0.0008339759,0.00009747488,0.0002448565,0.00077669555,0.0014584223,0.0045372723],"genre_scores_gemma":[0.6752003,0.0008037735,0.31184182,0.0005686707,0.00023432494,0.00058011187,0.0027382812,0.00040385255,0.0076288963],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9913694,0.0027989775,0.0006293337,0.0017762562,0.002813351,0.0006127029],"domain_scores_gemma":[0.9774825,0.004480132,0.0019326909,0.014153528,0.0016992885,0.00025195023],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050846688,0.0009273253,0.001448894,0.0020865807,0.0016585593,0.0030428316,0.002064298,0.0010003871,0.0024570136],"category_scores_gemma":[0.0158709,0.00053592125,0.0011487773,0.0039893156,0.0016841624,0.006620866,0.0050654,0.0020985564,0.0013062946],"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":[0.00081843924,0.0003070075,0.0096168835,0.000315792,0.00033688606,0.00039869928,0.0013121159,0.25832817,0.017609198,0.30196124,0.021201223,0.3877944],"study_design_scores_gemma":[0.00007780091,0.00014202374,0.0024037648,0.000059968082,0.00010753936,0.0010822277,0.0007987483,0.5937077,0.03944285,0.3107209,0.05137932,0.00007719288],"about_ca_topic_score_codex":0.0010151207,"about_ca_topic_score_gemma":0.0012314745,"teacher_disagreement_score":0.0050846688,"about_ca_system_score_codex":0.0017942445,"about_ca_system_score_gemma":0.0019568736,"threshold_uncertainty_score":0.026890576},"labels":[],"label_agreement":null},{"id":"W2102679851","doi":"10.14778/2350229.2350249","title":"Efficient indexing and querying over syntactically annotated trees","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Search engine indexing; Parsing; Coding (social sciences); Set (abstract data type); Natural language; Index (typography); Tree (set theory); Information retrieval; Artificial intelligence; Natural language processing; Mathematics","score_opus":0.010587261327778642,"score_gpt":0.2312921184450198,"score_spread":0.22070485711724114,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102679851","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":[0.11353021,0.0011881496,0.8592727,0.0007572629,0.0001095689,0.00032449767,0.0058009336,0.013534096,0.0054825633],"genre_scores_gemma":[0.3335313,0.0010165971,0.6451883,0.00026311463,0.00016155775,0.00036473142,0.015297513,0.0010240956,0.003152873],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99740285,0.00045977012,0.00038221118,0.0003520517,0.0011952457,0.00020774416],"domain_scores_gemma":[0.9907721,0.0045509753,0.00069332734,0.0022760888,0.001503553,0.0002039472],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016302887,0.000713146,0.0014848433,0.003568674,0.0010268644,0.0023199033,0.002282323,0.0010804661,0.002330263],"category_scores_gemma":[0.014498284,0.00053044,0.0008374597,0.007581635,0.0009992346,0.0072433134,0.0022357192,0.0011876498,0.0014507109],"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":[0.0009635679,0.00045700773,0.0074449885,0.0011357936,0.00013607995,0.0009163016,0.0018173632,0.062579766,0.12035946,0.08398446,0.046746112,0.6734591],"study_design_scores_gemma":[0.0001298993,0.00023238955,0.0029967981,0.00010295067,0.00010363994,0.0008591226,0.0008231554,0.76114273,0.076853864,0.13577974,0.020854203,0.00012151475],"about_ca_topic_score_codex":0.00316393,"about_ca_topic_score_gemma":0.00525225,"teacher_disagreement_score":0.003568674,"about_ca_system_score_codex":0.00094007957,"about_ca_system_score_gemma":0.002475032,"threshold_uncertainty_score":0.008621931},"labels":[],"label_agreement":null},{"id":"W2103437490","doi":"10.14778/2733085.2733095","title":"ADDICT","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Parallel Computing and Optimization Techniques","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"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; Cache; Locality; Latency (audio); CAS latency; Database transaction; Transaction processing; Memory footprint; Parallel computing; Software; Operating system; Multi-core processor; Distributed computing; Embedded system; Database; Memory controller","score_opus":0.007883670614523518,"score_gpt":0.21113710324098256,"score_spread":0.20325343262645904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103437490","genre_codex":"other","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":[0.10962546,0.008264289,0.33955103,0.0044396427,0.0023516144,0.0014525403,0.010505057,0.1294121,0.3943983],"genre_scores_gemma":[0.44546345,0.00558622,0.23482767,0.006870833,0.0007614798,0.0016460209,0.024447868,0.013403629,0.26699275],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988381,0.00016602833,0.0000742677,0.00030862764,0.0004349733,0.00017798605],"domain_scores_gemma":[0.99811953,0.0004426362,0.00020173605,0.0006017203,0.0004262427,0.00020823792],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010258778,0.000756789,0.00046849975,0.0008032986,0.0006592347,0.0018301064,0.0023957905,0.00082951214,0.0475352],"category_scores_gemma":[0.0035801579,0.00044433653,0.0005286756,0.0009211606,0.00063031254,0.0020352788,0.0022963916,0.001022931,0.028323507],"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":[0.0013439504,0.00071150623,0.0099260835,0.0014771657,0.00013297675,0.0008139842,0.00039612202,0.008585542,0.027131256,0.08202913,0.20014796,0.66730446],"study_design_scores_gemma":[0.00011598669,0.00063258526,0.0040178327,0.00012288876,0.00009083294,0.001880413,0.00011904836,0.042749114,0.031132026,0.027544072,0.89149684,0.00009838895],"about_ca_topic_score_codex":0.00064651377,"about_ca_topic_score_gemma":0.0011465165,"teacher_disagreement_score":0.0475352,"about_ca_system_score_codex":0.00059422385,"about_ca_system_score_gemma":0.00090780825,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2103615139","doi":"10.14778/1453856.1453955","title":"Keyword query cleaning","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; Ontario Tech University","funders":"","keywords":"Computer science; Query optimization; Query expansion; Web query classification; Web search query; Sargable; Query language; Information retrieval; Online aggregation; View; Set (abstract data type); Context (archaeology); Database; Query by Example; Data mining; Search engine","score_opus":0.016461420648945766,"score_gpt":0.19089173340955867,"score_spread":0.1744303127606129,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103615139","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":[0.019597169,0.002480983,0.9621047,0.0007570855,0.00028587316,0.0006551513,0.002136894,0.0074214623,0.004560646],"genre_scores_gemma":[0.18961003,0.001962342,0.78983134,0.001141436,0.0002727875,0.00060252304,0.007815471,0.002445092,0.0063191284],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9873661,0.0023895397,0.0019199775,0.0027757487,0.0045550456,0.0009936746],"domain_scores_gemma":[0.9793956,0.0059635495,0.0012226908,0.0074739805,0.005598207,0.0003460422],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004937681,0.0016732535,0.003820275,0.0035549812,0.0025916079,0.0047119088,0.0042400225,0.0024649843,0.005817516],"category_scores_gemma":[0.029346311,0.0010337067,0.0026037365,0.0066194898,0.0016084074,0.008339137,0.005772573,0.0025313713,0.0059205526],"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":[0.0015723181,0.00048803416,0.0073494622,0.0024524576,0.00044624702,0.0010919417,0.0017855704,0.026083695,0.07403389,0.039281342,0.06947888,0.7759361],"study_design_scores_gemma":[0.00031492603,0.0008275412,0.005487939,0.0004726626,0.0006274961,0.008258673,0.0044745654,0.39245567,0.20262761,0.13633709,0.24763837,0.00047742258],"about_ca_topic_score_codex":0.0032325184,"about_ca_topic_score_gemma":0.0024737972,"teacher_disagreement_score":0.005817516,"about_ca_system_score_codex":0.0013563127,"about_ca_system_score_gemma":0.0038431478,"threshold_uncertainty_score":0.026113272},"labels":[],"label_agreement":null},{"id":"W2104550444","doi":"10.14778/1453856.1453967","title":"Efficient skyline querying with variable user preferences on nominal attributes","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Chinese University of Hong Kong; National Natural Science Foundation of China","keywords":"Skyline; Computer science; Preference; Order (exchange); Tree (set theory); Variable (mathematics); Data mining; Mathematics; Statistics","score_opus":0.022724689187778764,"score_gpt":0.2061796146940327,"score_spread":0.18345492550625395,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2104550444","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":[0.32215056,0.0012506908,0.64757866,0.0008736514,0.00004452351,0.00039470292,0.0031423823,0.015319609,0.009245206],"genre_scores_gemma":[0.668922,0.0002594148,0.32281396,0.00014296554,0.00004595333,0.00016255368,0.0043900963,0.0004829592,0.0027800007],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976966,0.0007007855,0.00019715035,0.00030999366,0.0007801435,0.00031538925],"domain_scores_gemma":[0.99614173,0.0015902364,0.0003001579,0.0011980204,0.00060294155,0.00016689759],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018684062,0.00078478,0.0015068041,0.0010025406,0.0007355697,0.0019901576,0.0015022176,0.00083475653,0.003074718],"category_scores_gemma":[0.0054949936,0.00038032353,0.0005282953,0.0024907838,0.00047219344,0.00457098,0.0016039083,0.0004980673,0.0010221424],"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":[0.0061618695,0.0008045915,0.019354325,0.0011638787,0.0002541036,0.0008666403,0.002058187,0.11754274,0.11929967,0.032725986,0.053801347,0.6459667],"study_design_scores_gemma":[0.00027237448,0.00049444695,0.0032681404,0.0000363577,0.00005924936,0.0007944634,0.0010793254,0.92105126,0.038572315,0.022991937,0.011316477,0.00006363928],"about_ca_topic_score_codex":0.0020817271,"about_ca_topic_score_gemma":0.0038495169,"teacher_disagreement_score":0.003074718,"about_ca_system_score_codex":0.00057753717,"about_ca_system_score_gemma":0.00087930413,"threshold_uncertainty_score":0.010285914},"labels":[],"label_agreement":null},{"id":"W2106019582","doi":"10.14778/2168651.2168659","title":"ReStore","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":110,"is_retracted":false,"has_abstract":true,"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":"Computer science; Workflow; Dataflow; Reuse; Compiler; Implementation; Distributed computing; Operating system; Database; Parallel computing; Programming language","score_opus":0.013537665615544914,"score_gpt":0.21423723860527902,"score_spread":0.2006995729897341,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106019582","genre_codex":"software","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":[0.017962951,0.0013610618,0.1890476,0.0024704677,0.002142267,0.0010299688,0.032960434,0.5624081,0.19061717],"genre_scores_gemma":[0.17183954,0.0021408251,0.26017898,0.005616973,0.0009056763,0.0012996142,0.15494934,0.10180778,0.30126128],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99792504,0.00016736276,0.00012067109,0.0005408972,0.0009476177,0.00029840218],"domain_scores_gemma":[0.99659914,0.00035743447,0.00014049807,0.0018284629,0.00084111176,0.00023326601],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014827797,0.0014455781,0.00089165964,0.0019649793,0.0015096284,0.0036691623,0.0033397216,0.0012290734,0.08669438],"category_scores_gemma":[0.00577869,0.00088285195,0.0013551371,0.0016011808,0.001053896,0.0049632224,0.005616363,0.0022240027,0.07582542],"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":[0.0007449122,0.00023221143,0.0031323072,0.0007524079,0.00009696987,0.00029998508,0.0005116641,0.004112348,0.01021166,0.023807973,0.6692942,0.28680333],"study_design_scores_gemma":[0.00008543431,0.00008882695,0.0017366522,0.00007953403,0.00004052315,0.0003961402,0.00020481653,0.009924855,0.01706356,0.014120486,0.95618606,0.00007306657],"about_ca_topic_score_codex":0.0032383231,"about_ca_topic_score_gemma":0.003511785,"teacher_disagreement_score":0.08669438,"about_ca_system_score_codex":0.0010189654,"about_ca_system_score_gemma":0.0022938608,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2107047397","doi":"10.14778/1687553.1687600","title":"SMDM","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"Computer science; Ontology; Semantics (computer science); RDF; Business domain; IBM; Domain (mathematical analysis); Software engineering; Semantic Web; Business rule; Business process; Information retrieval; Programming language; Engineering","score_opus":0.009923090580914845,"score_gpt":0.21368471931580998,"score_spread":0.20376162873489514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107047397","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010533939,0.0014302095,0.68948144,0.0053657237,0.0014013563,0.00059804914,0.0072818464,0.017590042,0.26631743],"genre_scores_gemma":[0.18189697,0.0031023938,0.5896413,0.0038290678,0.0009148349,0.00092877855,0.04036116,0.0036141488,0.17571136],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975019,0.0005469349,0.00024658802,0.0005018988,0.00093566196,0.00026689234],"domain_scores_gemma":[0.99740976,0.0003083859,0.00011426156,0.0011548778,0.0008515713,0.00016111865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023883395,0.00050197175,0.0004676413,0.0015240213,0.0011375514,0.004383196,0.0022366543,0.0013174076,0.034668196],"category_scores_gemma":[0.004133022,0.0003221667,0.0008644941,0.0029973455,0.0006304209,0.0053105224,0.004464703,0.0013758201,0.018965682],"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":[0.00017916384,0.00014062745,0.0017731776,0.00047636783,0.00004917406,0.00045830963,0.00042143857,0.0044701947,0.0066422787,0.4128101,0.19090986,0.3816693],"study_design_scores_gemma":[0.000023307131,0.000029847144,0.00035247405,0.00006583104,0.000013205416,0.0003353199,0.00013718405,0.015902607,0.0040965,0.044370368,0.934658,0.000015421683],"about_ca_topic_score_codex":0.0028030295,"about_ca_topic_score_gemma":0.0033546044,"teacher_disagreement_score":0.034668196,"about_ca_system_score_codex":0.0014069089,"about_ca_system_score_gemma":0.0024481649,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2107080109","doi":"10.14778/1454159.1454220","title":"A revival of integrity constraints for data cleaning","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"Engineering and Physical Sciences Research Council","keywords":"Data integrity; Data quality; Computer science; Schema (genetic algorithms); Constraint (computer-aided design); Quality (philosophy); Risk analysis (engineering); Data science; Reliability engineering; Database; Information retrieval; Engineering; Operations management; Business; Mechanical engineering","score_opus":0.5397355688626213,"score_gpt":0.44987919658589165,"score_spread":0.08985637227672966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107080109","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":[0.0028282902,0.059265207,0.85972196,0.06505173,0.0032052407,0.00015165594,0.0007679772,0.0013773808,0.00763064],"genre_scores_gemma":[0.043935392,0.06594352,0.85882187,0.016313056,0.0062478227,0.0003991133,0.0016955896,0.0013918923,0.0052517345],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.93170804,0.025560662,0.005733217,0.006661665,0.028900038,0.001436318],"domain_scores_gemma":[0.83203167,0.092304304,0.0042348388,0.038077638,0.030883353,0.0024682283],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.065244034,0.00179881,0.003985106,0.0050329464,0.0039150766,0.013957338,0.0066970685,0.00608074,0.0045444933],"category_scores_gemma":[0.12633197,0.0021328705,0.003984455,0.010329834,0.010513643,0.036313187,0.012042548,0.023949435,0.0033579536],"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":[0.00018272456,0.000087703396,0.00088136556,0.0023553038,0.00026993558,0.00027440005,0.0014259241,0.0045965808,0.003450118,0.5161628,0.07784501,0.3924681],"study_design_scores_gemma":[0.00005291304,0.00009399231,0.00064633146,0.0012785398,0.00010344193,0.00087841623,0.0006140263,0.023695828,0.0070726722,0.43662885,0.52874213,0.00019278127],"about_ca_topic_score_codex":0.004303015,"about_ca_topic_score_gemma":0.00280382,"teacher_disagreement_score":0.065244034,"about_ca_system_score_codex":0.0058094407,"about_ca_system_score_gemma":0.0069112014,"threshold_uncertainty_score":0.34504753},"labels":[],"label_agreement":null},{"id":"W2110020044","doi":"10.14778/1453856.1453922","title":"A practical scalable distributed B-tree","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":135,"is_retracted":false,"has_abstract":true,"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; Distributed computing; Scalability; Tree (set theory); Fault tolerance; Distributed transaction; Concurrency; Transaction processing; Database transaction; Operating system; Database","score_opus":0.026289657278363878,"score_gpt":0.2516244494746975,"score_spread":0.2253347921963336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110020044","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":[0.0067523103,0.00054278306,0.97808313,0.00084338663,0.00013531931,0.00020846276,0.00029167987,0.00399447,0.009148495],"genre_scores_gemma":[0.08324079,0.0004819768,0.9094471,0.00031725626,0.00009595822,0.00029317575,0.0007589291,0.00024830847,0.005116498],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987914,0.00015787123,0.000094323885,0.00019485943,0.00064739725,0.00011425584],"domain_scores_gemma":[0.9985032,0.00028555762,0.00006812212,0.00037550158,0.0005924118,0.00017523253],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011580207,0.0003841064,0.00056871737,0.0006775389,0.0012667876,0.0013173211,0.0022663353,0.0012656286,0.010208172],"category_scores_gemma":[0.003908471,0.00039892035,0.000417063,0.0016308313,0.00052152644,0.0024236366,0.0022037127,0.0010356698,0.0040324647],"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":[0.00041750114,0.00025238894,0.0014501456,0.0006531661,0.000060803002,0.0006378749,0.0002455833,0.11399539,0.045147106,0.15404555,0.095703915,0.5873906],"study_design_scores_gemma":[0.0002989906,0.0003298957,0.0004128127,0.00007969943,0.000045665944,0.00097447314,0.00011551209,0.68607527,0.012553556,0.10897957,0.19007674,0.00005788406],"about_ca_topic_score_codex":0.002046647,"about_ca_topic_score_gemma":0.0022540842,"teacher_disagreement_score":0.010208172,"about_ca_system_score_codex":0.0006705831,"about_ca_system_score_gemma":0.0016769231,"threshold_uncertainty_score":0.034149706},"labels":[],"label_agreement":null},{"id":"W2111227779","doi":"10.14778/1454159.1454189","title":"P2P logging and timestamping for reconciliation","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Peer-to-Peer Network Technologies","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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; Distributed hash table; Peer-to-peer; Hash function; Distributed computing; Consistency (knowledge bases); Hash table; Logging; Computer network; Computer security","score_opus":0.025745073631701223,"score_gpt":0.22740153810719047,"score_spread":0.20165646447548924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111227779","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":[0.0063647022,0.0011634519,0.9714019,0.0011646859,0.0004841779,0.0003535433,0.00033178652,0.0070997085,0.011636089],"genre_scores_gemma":[0.2685668,0.0022559196,0.71377945,0.0009621641,0.0009281092,0.0007271236,0.0016429705,0.0014470859,0.009690425],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99113667,0.0032008057,0.0009193667,0.0011169731,0.0031889868,0.00043732097],"domain_scores_gemma":[0.9825752,0.0047075525,0.0012796527,0.008720637,0.002340297,0.00037661602],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006957316,0.0010998658,0.0010927395,0.0018686702,0.0022884815,0.004693866,0.0033352724,0.0023118444,0.005900814],"category_scores_gemma":[0.019060824,0.00075763866,0.0008286258,0.0032870767,0.0017155854,0.009796061,0.004819673,0.0031304772,0.003492233],"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":[0.00056214124,0.00031561803,0.0039645783,0.0011759178,0.00020839286,0.0008302907,0.0014098019,0.031328376,0.021868913,0.31340247,0.04215734,0.5827762],"study_design_scores_gemma":[0.00025280568,0.00043504487,0.0019828987,0.0003301704,0.00022252904,0.002593806,0.0006487731,0.2506431,0.06679781,0.18582761,0.48997596,0.00028954915],"about_ca_topic_score_codex":0.0012341975,"about_ca_topic_score_gemma":0.0009969083,"teacher_disagreement_score":0.006957316,"about_ca_system_score_codex":0.0008611345,"about_ca_system_score_gemma":0.0026200064,"threshold_uncertainty_score":0.036794305},"labels":[],"label_agreement":null},{"id":"W2111607365","doi":"10.14778/1687627.1687727","title":"Distance-join","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":207,"is_retracted":false,"has_abstract":true,"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; Join (topology); Graph; Theoretical computer science; Query optimization; Shortest path problem; Data mining; Mathematics; Combinatorics","score_opus":0.006708040847259433,"score_gpt":0.20173282838343293,"score_spread":0.1950247875361735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111607365","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":[0.014799422,0.0014533666,0.9535672,0.0006500403,0.0004368774,0.0006011936,0.0045195166,0.0087580625,0.0152142085],"genre_scores_gemma":[0.14778234,0.00087409467,0.8205541,0.00047533063,0.00032407657,0.0005161673,0.013797527,0.0014447197,0.014231552],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9952506,0.0006800601,0.0005134137,0.0012863864,0.001969492,0.0003001595],"domain_scores_gemma":[0.99529475,0.0011644065,0.0003038366,0.0022076815,0.0007572607,0.0002720557],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002386493,0.0012830684,0.002016978,0.0030127272,0.0019314492,0.0031744584,0.004250894,0.0015799775,0.022291165],"category_scores_gemma":[0.008770889,0.00067093596,0.0016539996,0.005367994,0.0006916338,0.005055226,0.004442924,0.0018402273,0.008290081],"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":[0.0011962441,0.0006818881,0.0036314996,0.00089062034,0.00035234677,0.0002803667,0.00047011056,0.04489188,0.010648757,0.086091645,0.08818434,0.76268035],"study_design_scores_gemma":[0.00038608225,0.0007611435,0.0019981316,0.00012444088,0.00020420126,0.0016282821,0.0007702838,0.49011257,0.032171942,0.2459951,0.22568123,0.00016653187],"about_ca_topic_score_codex":0.002195672,"about_ca_topic_score_gemma":0.0037372732,"teacher_disagreement_score":0.022291165,"about_ca_system_score_codex":0.00094003027,"about_ca_system_score_gemma":0.0014352334,"threshold_uncertainty_score":0.07457131},"labels":[],"label_agreement":null},{"id":"W2111811740","doi":"10.14778/2021017.2021018","title":"On pruning for top-k ranking in uncertain databases","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of Alberta","funders":"","keywords":"Tuple; Pruning; Ranking (information retrieval); Rank (graph theory); Computer science; Parameterized complexity; Key (lock); Range (aeronautics); Learning to rank; Semantics (computer science); Function (biology); Computation; Ranking SVM; Task (project management); Database; Information retrieval; Artificial intelligence; Data mining; Mathematics; Algorithm; Programming language; Combinatorics; Discrete mathematics","score_opus":0.06770205945613778,"score_gpt":0.2641757924783974,"score_spread":0.19647373302225962,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111811740","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":[0.022787083,0.0016371345,0.97204405,0.0005344452,0.000049739032,0.0000964724,0.00016355247,0.00044735344,0.0022401304],"genre_scores_gemma":[0.34979042,0.0016003407,0.64497125,0.00030245987,0.00019447062,0.00019696461,0.0006811132,0.00026535403,0.0019975703],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99177635,0.003349777,0.0005920557,0.0008798662,0.0026746246,0.0007273653],"domain_scores_gemma":[0.97834957,0.015825745,0.0008704213,0.0032153407,0.0013860471,0.00035297094],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0073932623,0.0012907118,0.0030591688,0.0026043581,0.00265284,0.004163521,0.0026200563,0.0023154246,0.0018962395],"category_scores_gemma":[0.031850133,0.0006858756,0.0017777779,0.004120819,0.0028467742,0.0078783985,0.0030600692,0.0029922414,0.0005819621],"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":[0.0005935527,0.00022381099,0.0028343578,0.00046121617,0.00019600593,0.000503913,0.00078962033,0.48991328,0.004731545,0.25029814,0.006229629,0.2432249],"study_design_scores_gemma":[0.00003638778,0.00010165837,0.00044018158,0.00007059588,0.00006126146,0.0002861362,0.00015811197,0.7340154,0.0027642213,0.25847912,0.0035417126,0.00004516723],"about_ca_topic_score_codex":0.0051340745,"about_ca_topic_score_gemma":0.006152779,"teacher_disagreement_score":0.0073932623,"about_ca_system_score_codex":0.0018631296,"about_ca_system_score_gemma":0.0023532698,"threshold_uncertainty_score":0.039099753},"labels":[],"label_agreement":null},{"id":"W2111814513","doi":"10.14778/2535570.2488330","title":"Partitioning and ranking tagged data sources","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":6,"is_retracted":false,"has_abstract":true,"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; Ranking (information retrieval); Information retrieval; Categorization; Partition (number theory); Set (abstract data type); Rank (graph theory); Learning to rank; Focus (optics); Data mining; Social media; Data science; World Wide Web; Artificial intelligence; Mathematics","score_opus":0.016600964907052655,"score_gpt":0.23227638965751488,"score_spread":0.21567542475046222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2111814513","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":[0.037838437,0.0007160653,0.9554579,0.0004765192,0.0001489785,0.0006673997,0.0021094063,0.0010158799,0.0015693643],"genre_scores_gemma":[0.15984592,0.000499286,0.82685983,0.00013864208,0.00017853442,0.00065788976,0.009847344,0.00027951875,0.0016929839],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.987938,0.0041287667,0.0010759485,0.002284331,0.0037796495,0.00079333875],"domain_scores_gemma":[0.97412777,0.0126583595,0.0016692796,0.0058234087,0.0049796626,0.00074151944],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0080415895,0.0022823287,0.002735862,0.011375735,0.002632932,0.0068335827,0.003752546,0.001962093,0.0015571082],"category_scores_gemma":[0.0383774,0.0010570921,0.0024195116,0.010564655,0.0011588146,0.0062522884,0.004679221,0.002024939,0.001431141],"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":[0.0012243037,0.000735226,0.01925947,0.0015266449,0.00063390954,0.00082546694,0.0024806615,0.19164406,0.021067789,0.05672936,0.015612173,0.688261],"study_design_scores_gemma":[0.00012655438,0.00038196912,0.005728985,0.0002520471,0.00030703572,0.0007199006,0.0022292368,0.81684583,0.019388784,0.12545653,0.028401813,0.00016131953],"about_ca_topic_score_codex":0.004630052,"about_ca_topic_score_gemma":0.008232278,"teacher_disagreement_score":0.011375735,"about_ca_system_score_codex":0.0021971893,"about_ca_system_score_gemma":0.0030563269,"threshold_uncertainty_score":0.04252851},"labels":[],"label_agreement":null},{"id":"W2112007194","doi":"10.14778/2536274.2536316","title":"IPS","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Flexibility (engineering); Computer science; Set (abstract data type); Task (project management); Point of interest; Human–computer interaction; Engineering; Artificial intelligence; Systems engineering","score_opus":0.0070832293950941465,"score_gpt":0.17799929477527485,"score_spread":0.1709160653801807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112007194","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014441143,0.00048454816,0.4698322,0.00071934104,0.00074203307,0.0005073824,0.023178658,0.26334646,0.22674814],"genre_scores_gemma":[0.24068321,0.0012833419,0.4738337,0.0009329145,0.00042212658,0.0013547218,0.07859316,0.024688534,0.17820829],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992974,0.00010779146,0.000047269677,0.0002131096,0.00024328388,0.000091171365],"domain_scores_gemma":[0.99865437,0.00026619376,0.000066002416,0.0006032939,0.00028983402,0.0001202261],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009284746,0.0011107341,0.000571706,0.0014526732,0.00091947964,0.0032006877,0.001912624,0.00090143684,0.1176783],"category_scores_gemma":[0.0034307956,0.0006135087,0.00069127465,0.001654454,0.00038386002,0.0034891248,0.0026151736,0.0011953014,0.06546022],"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":[0.0006793531,0.00013963995,0.0036149432,0.0004946201,0.00006446804,0.000461779,0.0007982898,0.006412411,0.0060801124,0.042743143,0.4596569,0.4788543],"study_design_scores_gemma":[0.00006804213,0.00012213664,0.001914713,0.00008479714,0.000049835406,0.00067764486,0.00033102775,0.03885493,0.012177488,0.02140662,0.92424506,0.00006777337],"about_ca_topic_score_codex":0.002648898,"about_ca_topic_score_gemma":0.0030581206,"teacher_disagreement_score":0.8823217,"about_ca_system_score_codex":0.000464173,"about_ca_system_score_gemma":0.0008989076,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2112485457","doi":"10.14778/1687553.1687567","title":"SQL/MapReduce","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":127,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"ASTER","funders":"","keywords":"Computer science; SQL; User-defined function; Database; Schema (genetic algorithms); Scalability; NoSQL; Programming language; Query by Example; Information retrieval","score_opus":0.007158007447369725,"score_gpt":0.213177163393006,"score_spread":0.20601915594563627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112485457","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":[0.0036482683,0.0005181087,0.59676063,0.0009794378,0.0005347223,0.0011045805,0.022950936,0.3411616,0.03234172],"genre_scores_gemma":[0.0767946,0.0017239376,0.6958285,0.0028941398,0.0005114646,0.0023064383,0.120700896,0.05727695,0.04196307],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954478,0.0005650685,0.0005322359,0.000902441,0.0021679734,0.0003844705],"domain_scores_gemma":[0.9966191,0.0005944523,0.00016143371,0.0013026114,0.00098631,0.0003359623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036377318,0.0023197366,0.0012255972,0.0015026703,0.0010127096,0.0036888674,0.0058805165,0.0009482381,0.023550596],"category_scores_gemma":[0.0058445833,0.0012690241,0.0021854567,0.0016646279,0.000776081,0.0033394804,0.004268332,0.003546485,0.032044742],"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":[0.000781192,0.00046706086,0.0017487651,0.0014320412,0.00030281898,0.00046088503,0.00063486357,0.011269496,0.017582018,0.06268515,0.6631722,0.23946348],"study_design_scores_gemma":[0.00024250096,0.00016750727,0.0012068474,0.00010120548,0.00005396668,0.0006207684,0.00027812287,0.075684644,0.032069746,0.049487393,0.8398957,0.00019160386],"about_ca_topic_score_codex":0.0042955275,"about_ca_topic_score_gemma":0.0032506026,"teacher_disagreement_score":0.023550596,"about_ca_system_score_codex":0.0010752216,"about_ca_system_score_gemma":0.003485363,"threshold_uncertainty_score":0.078784585},"labels":[],"label_agreement":null},{"id":"W2112840274","doi":"10.14778/1920841.1920870","title":"Sampling the repairs of functional dependency violations under hard constraints","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":127,"is_retracted":false,"has_abstract":true,"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); Class (philosophy); Functional dependency; Context (archaeology); Sampling (signal processing); Relation (database); Data integrity; Variety (cybernetics); Space (punctuation); Data mining; Metric (unit); Theoretical computer science; Relational database; Database; Artificial intelligence","score_opus":0.1812498346188758,"score_gpt":0.3696657815601582,"score_spread":0.18841594694128241,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112840274","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":[0.34021884,0.0006179192,0.6543396,0.00095517363,0.000053308137,0.00030856577,0.0006021249,0.0017206756,0.0011837415],"genre_scores_gemma":[0.71432114,0.00017814402,0.28265762,0.00021190505,0.000045558594,0.000273417,0.0012507311,0.0002609746,0.00080050295],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99403507,0.0025469132,0.0005231496,0.0011374186,0.0013864412,0.00037105868],"domain_scores_gemma":[0.94591707,0.040354434,0.0028566248,0.0075910506,0.00261215,0.0006686944],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0066418583,0.00079139916,0.0014659105,0.0014289627,0.0007763588,0.0010597713,0.00187476,0.0015941503,0.0010445641],"category_scores_gemma":[0.04814147,0.00062318635,0.0011557956,0.0013181594,0.001356201,0.0021323955,0.001753863,0.0017977966,0.00027082418],"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":[0.0013545019,0.0005490964,0.026768569,0.000587062,0.00025553184,0.00066285237,0.0012894645,0.6412677,0.016597504,0.016591145,0.006223891,0.28785264],"study_design_scores_gemma":[0.00008066854,0.0002045106,0.001781795,0.0000381374,0.00006413247,0.0003125748,0.00034691385,0.95949066,0.011576187,0.024320481,0.0017576977,0.000026260092],"about_ca_topic_score_codex":0.002166295,"about_ca_topic_score_gemma":0.0033350252,"teacher_disagreement_score":0.0066418583,"about_ca_system_score_codex":0.00094127696,"about_ca_system_score_gemma":0.0017450983,"threshold_uncertainty_score":0.03512597},"labels":[],"label_agreement":null},{"id":"W2113415503","doi":"10.14778/1687627.1687695","title":"Modeling and querying possible repairs in duplicate detection","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":57,"is_retracted":false,"has_abstract":true,"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; Data mining; Parameterized complexity; Cluster analysis; Identification (biology); Set (abstract data type); Database; Algorithm; Machine learning","score_opus":0.08693929288766766,"score_gpt":0.34729863073080464,"score_spread":0.260359337843137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113415503","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":[0.060350787,0.00064978836,0.93616796,0.00066856004,0.000019072038,0.00011827207,0.0007298392,0.00055565295,0.00074002997],"genre_scores_gemma":[0.58815515,0.00068780454,0.4079065,0.00019673625,0.00006462571,0.00032479194,0.0017453718,0.00012815002,0.0007908278],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9776638,0.008640594,0.0023521483,0.0043322453,0.0058632665,0.0011479724],"domain_scores_gemma":[0.9190702,0.05695133,0.0069022975,0.012796504,0.0035205225,0.00075923133],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01872432,0.0011442618,0.0027605547,0.004097712,0.0016488261,0.0063190577,0.006495992,0.0030682192,0.0012160786],"category_scores_gemma":[0.09447894,0.0018622846,0.0031444933,0.006874503,0.0031988504,0.011578902,0.004885925,0.002656908,0.0002922386],"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":[0.00029142576,0.00012449591,0.0074175294,0.00025690804,0.00016935091,0.00034648782,0.00087269256,0.88031226,0.0015327351,0.059227042,0.0011312884,0.048317846],"study_design_scores_gemma":[0.000024901005,0.00006799228,0.0004971216,0.000037983697,0.00004871634,0.00027024024,0.00028398447,0.90090454,0.0022202558,0.09434089,0.0012642163,0.00003914677],"about_ca_topic_score_codex":0.007569163,"about_ca_topic_score_gemma":0.0052659125,"teacher_disagreement_score":0.01872432,"about_ca_system_score_codex":0.0032104037,"about_ca_system_score_gemma":0.002540408,"threshold_uncertainty_score":0.09902489},"labels":[],"label_agreement":null},{"id":"W2114206928","doi":"10.14778/2536206.2536214","title":"RACE","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"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; Speedup; Cache; Cloud computing; Parallel computing; Sequence (biology); Representation (politics); Multi-core processor; Contrast (vision); Scaling; Artificial intelligence; Operating system; Mathematics","score_opus":0.006947487662152172,"score_gpt":0.19577461303583504,"score_spread":0.18882712537368287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114206928","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":[0.010011141,0.0025416687,0.40761036,0.0017539295,0.002598414,0.00091083214,0.05837656,0.35516146,0.16103561],"genre_scores_gemma":[0.085743695,0.0026418988,0.45338675,0.004722238,0.0010709119,0.002517344,0.18914764,0.0688385,0.19193101],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99665713,0.0004999171,0.00026303428,0.0011565299,0.0009979099,0.00042552646],"domain_scores_gemma":[0.9967409,0.0007601096,0.00025361584,0.0012725338,0.00078164414,0.00019128251],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021707201,0.001869461,0.0016425633,0.0016340245,0.0017866811,0.0036572686,0.003750796,0.0020778812,0.1181336],"category_scores_gemma":[0.0073691383,0.0012273215,0.0023449243,0.00216683,0.0007137096,0.0040007913,0.0034108334,0.0024266127,0.11139626],"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":[0.0014496258,0.00023957397,0.0034216589,0.0013508688,0.000209281,0.00031911788,0.00033458037,0.0056217695,0.020816408,0.05013237,0.668868,0.24723671],"study_design_scores_gemma":[0.00021166571,0.0001989789,0.0012962068,0.00013631245,0.00011101533,0.00052484986,0.00007459738,0.024553634,0.01734051,0.024581965,0.9308477,0.00012254584],"about_ca_topic_score_codex":0.0023260743,"about_ca_topic_score_gemma":0.002943098,"teacher_disagreement_score":0.1181336,"about_ca_system_score_codex":0.0009541815,"about_ca_system_score_gemma":0.0025164522,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2115215982","doi":"10.14778/1453856.1453883","title":"Hashed samples","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; York University","funders":"","keywords":"Similarity (geometry); Estimator; A priori and a posteriori; Computer science; Overhead (engineering); Set (abstract data type); Sampling (signal processing); Cosine similarity; Algorithm; Data mining; Pattern recognition (psychology); Mathematics; Artificial intelligence; Statistics","score_opus":0.2910039101624485,"score_gpt":0.3717421828373335,"score_spread":0.08073827267488504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2115215982","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":[0.025844723,0.00028987697,0.97073495,0.00020050793,0.00005993759,0.00018417463,0.00034637266,0.0011961401,0.0011432981],"genre_scores_gemma":[0.443255,0.0003471438,0.55104184,0.00029902658,0.0002459524,0.0004468014,0.001448263,0.0002319907,0.0026840384],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9945505,0.001136465,0.00041310195,0.0011233565,0.0024313538,0.00034533706],"domain_scores_gemma":[0.9784055,0.011736033,0.0016165841,0.0057623526,0.0020851041,0.00039445606],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045654657,0.0006726794,0.0013622333,0.0022301343,0.000921086,0.0020514354,0.0022303266,0.0011232707,0.004500926],"category_scores_gemma":[0.037763372,0.0005905165,0.000822903,0.0025668168,0.0011854644,0.005785041,0.0026672094,0.0013258606,0.0016924266],"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":[0.0018529679,0.00037009097,0.016496953,0.0005989676,0.00021480629,0.00036969344,0.0008046987,0.11804186,0.025769064,0.1614239,0.009754835,0.6643022],"study_design_scores_gemma":[0.00013937874,0.0004997972,0.0017619636,0.000048504044,0.00005759515,0.0008079214,0.00027215428,0.85748637,0.029777082,0.099556014,0.009540191,0.00005301665],"about_ca_topic_score_codex":0.00088472944,"about_ca_topic_score_gemma":0.00090282987,"teacher_disagreement_score":0.0045654657,"about_ca_system_score_codex":0.0010667172,"about_ca_system_score_gemma":0.0010600479,"threshold_uncertainty_score":0.024144769},"labels":[],"label_agreement":null},{"id":"W2117538598","doi":"10.14778/1687627.1687659","title":"A scalable, predictable join operator for highly concurrent data warehouses","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":87,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"ASTER","funders":"","keywords":"Computer science; Query plan; Scalability; Query optimization; Online aggregation; Data warehouse; Tuple; Throughput; Computation; Query language; Pipeline (software); Sargable; Database; Data mining; Distributed computing; Web search query; Search engine; Information retrieval; Algorithm; Programming language","score_opus":0.035230274181042985,"score_gpt":0.27187334763998267,"score_spread":0.23664307345893967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117538598","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":[0.085927665,0.0002727742,0.89041823,0.0002680905,0.00008081797,0.00028006025,0.00018183955,0.020662552,0.0019080575],"genre_scores_gemma":[0.5524328,0.00018168066,0.4438276,0.00017512027,0.00010251691,0.00019874256,0.00044907877,0.0005956781,0.0020368616],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986507,0.00013202179,0.00009276383,0.00025768395,0.0007286532,0.00013809715],"domain_scores_gemma":[0.9984194,0.00045569698,0.00013219302,0.00052932755,0.0002719664,0.00019132764],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019360201,0.00050874706,0.0005529565,0.0004267008,0.000803351,0.0013282615,0.0027246808,0.0005036845,0.0012959958],"category_scores_gemma":[0.002925025,0.0006188284,0.0004519189,0.00058044674,0.0006813659,0.0019151407,0.0019171132,0.0012459446,0.00041278356],"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":[0.0031635913,0.0011259685,0.010841678,0.00033601394,0.00015406875,0.0006994168,0.0012291957,0.06636148,0.36759812,0.02747726,0.025750143,0.49526298],"study_design_scores_gemma":[0.00037691186,0.0006037064,0.0017137072,0.00001416711,0.00008142966,0.00034490426,0.00015747616,0.9210724,0.05316964,0.008804369,0.013592032,0.0000691705],"about_ca_topic_score_codex":0.004068205,"about_ca_topic_score_gemma":0.0039870827,"teacher_disagreement_score":0.004068205,"about_ca_system_score_codex":0.0005839977,"about_ca_system_score_gemma":0.0019721508,"threshold_uncertainty_score":0.010238767},"labels":[],"label_agreement":null},{"id":"W2119323564","doi":"10.14778/2536206.2536208","title":"A data-adaptive and dynamic segmentation index for whole matching on time series","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":99,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Search engine indexing; Series (stratigraphy); Segmentation; Computer science; Matching (statistics); Index (typography); Time series; Similarity (geometry); Tree (set theory); Nearest neighbor search; Data mining; Algorithm; Pattern recognition (psychology); Mathematics; Artificial intelligence; Machine learning; Statistics; Image (mathematics)","score_opus":0.0158004004847677,"score_gpt":0.22996060156035852,"score_spread":0.21416020107559083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2119323564","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":[0.026001418,0.0012134344,0.9669055,0.00022894955,0.00014127196,0.00013552887,0.0013499362,0.0017663147,0.002257567],"genre_scores_gemma":[0.211367,0.0010867534,0.7805803,0.00017271844,0.00021120855,0.00029059226,0.004262335,0.00027970527,0.0017494541],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980932,0.00024880838,0.00030880712,0.00047012485,0.00078240683,0.000096702424],"domain_scores_gemma":[0.99599767,0.0013021439,0.00041047478,0.0011381534,0.0009362596,0.00021537911],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019573616,0.00060205994,0.0014266832,0.0049589924,0.00089651806,0.0018450683,0.0016734528,0.0008962239,0.0019796744],"category_scores_gemma":[0.013275112,0.0003535975,0.0006598708,0.008064051,0.0007904985,0.006622171,0.002334304,0.0011561876,0.0012425245],"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":[0.00047872862,0.00022883799,0.0073311417,0.000380214,0.00011050221,0.00018067271,0.00050223444,0.07269931,0.03145682,0.07492201,0.019587068,0.79212254],"study_design_scores_gemma":[0.000054369033,0.00034547053,0.0031870946,0.000059422702,0.000070488786,0.00067590905,0.0002248961,0.8702545,0.01595977,0.076714024,0.032362677,0.000091344795],"about_ca_topic_score_codex":0.0023055135,"about_ca_topic_score_gemma":0.0028298616,"teacher_disagreement_score":0.0049589924,"about_ca_system_score_codex":0.0012082412,"about_ca_system_score_gemma":0.0018118721,"threshold_uncertainty_score":0.010351598},"labels":[],"label_agreement":null},{"id":"W2122028373","doi":"10.14778/1454159.1454213","title":"Capri/MR","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Protein Structure and Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; National Research Council Canada","funders":"","keywords":"Computer science; Protein structure database; Nearest neighbor search; Protein Data Bank; Data mining; Protein structure; Database; Biology; Gene; Sequence database","score_opus":0.0066496077312751325,"score_gpt":0.20217461489252136,"score_spread":0.19552500716124624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122028373","genre_codex":"software","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":[0.010731445,0.008791405,0.18488812,0.0027522075,0.002716981,0.0012857682,0.048844315,0.61932856,0.12066123],"genre_scores_gemma":[0.075868234,0.0068876795,0.47220114,0.006141888,0.0020968148,0.0030401058,0.26539415,0.08267158,0.08569837],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9952194,0.0005737965,0.0003977454,0.0013178913,0.0019549301,0.0005360946],"domain_scores_gemma":[0.9942768,0.0008198497,0.00047541212,0.0016789808,0.0022383584,0.0005105824],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039304607,0.004168985,0.00298578,0.0041932226,0.002088825,0.0066896407,0.00888312,0.0037362578,0.07231608],"category_scores_gemma":[0.010266,0.0022749428,0.0021787665,0.0033520751,0.0010149925,0.0060753888,0.0047592614,0.004899362,0.11630553],"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":[0.0014444218,0.0001458916,0.0013052874,0.001228426,0.00020926923,0.00047572237,0.00021876706,0.0021658395,0.01622788,0.008962733,0.8494847,0.11813109],"study_design_scores_gemma":[0.00041234455,0.00029153377,0.0012364546,0.00025434353,0.00018148794,0.00174476,0.000111730784,0.037597165,0.052762806,0.009224069,0.8958459,0.00033741188],"about_ca_topic_score_codex":0.0039898064,"about_ca_topic_score_gemma":0.0032252974,"teacher_disagreement_score":0.07231608,"about_ca_system_score_codex":0.0017692795,"about_ca_system_score_gemma":0.0030843606,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2126547925","doi":"10.14778/1920841.1920906","title":"MRShare","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":226,"is_retracted":false,"has_abstract":true,"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; Cloud computing; Batch processing; Context (archaeology); Distributed computing; Work (physics); Core (optical fiber); Database; Operating system","score_opus":0.00697231335455234,"score_gpt":0.1997996891760867,"score_spread":0.19282737582153436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126547925","genre_codex":"software","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":[0.010547194,0.0014961391,0.2700234,0.003366326,0.0020233265,0.0012819387,0.03351859,0.40283814,0.27490485],"genre_scores_gemma":[0.15354604,0.0019054043,0.25340623,0.003094911,0.0009320015,0.0014276032,0.12373036,0.052640416,0.40931708],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977192,0.00026220098,0.00014549213,0.00046171693,0.0010679822,0.0003434217],"domain_scores_gemma":[0.9971349,0.00039547138,0.00012001856,0.0012153758,0.0007528387,0.00038135308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015382738,0.0014220122,0.001249679,0.0014725626,0.0010776678,0.003805078,0.0042289966,0.0016123427,0.18429643],"category_scores_gemma":[0.005464657,0.0009656087,0.0014484719,0.0014803889,0.0006136687,0.0051108818,0.0042972025,0.0022442695,0.12915456],"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":[0.0009334768,0.00024163447,0.0009044626,0.0005906096,0.00010071084,0.00035463035,0.00017598286,0.0040161437,0.008220611,0.023171913,0.76240903,0.19888091],"study_design_scores_gemma":[0.00021610962,0.00014783186,0.00063617143,0.00005456369,0.00003120265,0.0004955349,0.000083545565,0.01951059,0.009782199,0.016474718,0.95246947,0.0000981141],"about_ca_topic_score_codex":0.0015544033,"about_ca_topic_score_gemma":0.0019418714,"teacher_disagreement_score":0.18429643,"about_ca_system_score_codex":0.00072884565,"about_ca_system_score_gemma":0.0018071024,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2128248866","doi":"10.14778/1687627.1687734","title":"k-automorphism","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":407,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Popularity; Computer science; Extension (predicate logic); Personally identifiable information; Automorphism; Computer security; Data mining; Theoretical computer science; Mathematics; Discrete mathematics; Programming language","score_opus":0.018531082140186377,"score_gpt":0.24089356765210648,"score_spread":0.22236248551192012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128248866","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":[0.08529949,0.000643977,0.87670004,0.0021389618,0.00039181026,0.00039438237,0.00083122985,0.0031713925,0.030428642],"genre_scores_gemma":[0.77568644,0.0008624443,0.20043504,0.0012573921,0.0003982674,0.00037621567,0.0013022496,0.000841552,0.018840466],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9948801,0.0011794628,0.0005402142,0.0016454083,0.0010068477,0.00074797746],"domain_scores_gemma":[0.9796474,0.0074291117,0.0015983885,0.008787775,0.0018167385,0.0007204728],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003147309,0.0011706242,0.0018983823,0.0014115826,0.00272334,0.00360982,0.001976487,0.0028091471,0.0069976267],"category_scores_gemma":[0.020575583,0.00081429526,0.0033869252,0.0012593793,0.0050404184,0.008743367,0.0063416944,0.0042863595,0.005495713],"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":[0.00063085597,0.00019412588,0.003641982,0.0005633452,0.00023423287,0.000547422,0.0016844787,0.03075872,0.012449758,0.8233978,0.018126925,0.107770294],"study_design_scores_gemma":[0.00006793401,0.0001734329,0.0005031736,0.000058421036,0.000060089045,0.0008628143,0.00023091935,0.045159094,0.01015387,0.9264411,0.016222512,0.00006680098],"about_ca_topic_score_codex":0.0006060633,"about_ca_topic_score_gemma":0.00041227997,"teacher_disagreement_score":0.0069976267,"about_ca_system_score_codex":0.001488464,"about_ca_system_score_gemma":0.0022187477,"threshold_uncertainty_score":0.023409367},"labels":[],"label_agreement":null},{"id":"W2128418848","doi":"10.14778/1453856.1453926","title":"Dynamic partitioning of the cache hierarchy in shared data centers","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"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; Cache; Database; Cache algorithms; Cache invalidation; Quality of service; Computer network; Operating system; Distributed computing; CPU cache","score_opus":0.028961392859703554,"score_gpt":0.23278076208879522,"score_spread":0.20381936922909166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128418848","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":[0.6775112,0.00047978904,0.31948486,0.00012943437,0.00002133079,0.00007204788,0.00004357278,0.0011314544,0.0011262647],"genre_scores_gemma":[0.97157943,0.000034643483,0.028105779,0.000023714123,0.000005264313,0.000025803327,0.000030114657,0.000024461853,0.0001707829],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985971,0.00036643498,0.00007045209,0.00037053294,0.00032792296,0.00026748548],"domain_scores_gemma":[0.99739456,0.00089478574,0.0004558159,0.00064155506,0.00041031258,0.00020287688],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014357051,0.000387216,0.0006328727,0.0005454123,0.0008480456,0.0012207702,0.001643278,0.00043171755,0.00028667718],"category_scores_gemma":[0.004828686,0.00044101404,0.00024199646,0.0007729432,0.0008939587,0.00161949,0.0010303295,0.0005565855,0.000105409876],"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":[0.0008660439,0.00044590834,0.023685228,0.00011506359,0.00015408652,0.0003028333,0.0007223626,0.7318207,0.07771687,0.009898936,0.0015239082,0.15274806],"study_design_scores_gemma":[0.00005321985,0.00018850468,0.0035256576,0.000010497895,0.000047765923,0.000120913544,0.00022200332,0.96701807,0.023406612,0.0044096117,0.00096979685,0.000027409762],"about_ca_topic_score_codex":0.0052567045,"about_ca_topic_score_gemma":0.0075507704,"teacher_disagreement_score":0.0052567045,"about_ca_system_score_codex":0.0015222505,"about_ca_system_score_gemma":0.0016718347,"threshold_uncertainty_score":0.0110448},"labels":[],"label_agreement":null},{"id":"W2128841495","doi":"10.14778/2021017.2021019","title":"PLP","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Parallel Computing and Optimization Techniques","field":"Computer Science","cited_by":74,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"European Social Fund; National Science Foundation","keywords":"Computer science; Heap (data structure); Thread (computing); Parallel computing; Multi-core processor; Distributed computing; Operating system; Programming language","score_opus":0.026744184673494705,"score_gpt":0.21851641650143064,"score_spread":0.19177223182793593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128841495","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010614209,0.0016665054,0.34231687,0.004367877,0.002526009,0.00065866706,0.010688441,0.0713843,0.5557772],"genre_scores_gemma":[0.12925187,0.0023935433,0.1898898,0.0033120099,0.0012309853,0.00086437014,0.03812084,0.012606848,0.6223297],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99894744,0.00013043969,0.000069979265,0.00023070077,0.0004801948,0.00014124085],"domain_scores_gemma":[0.99799097,0.00017802208,0.00008224497,0.00084204547,0.0007303069,0.00017640708],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00091910205,0.00082080846,0.00039023577,0.00096843025,0.0008930005,0.0026956724,0.0017857375,0.0010792608,0.20260103],"category_scores_gemma":[0.0028578872,0.0004865106,0.00048618353,0.0011683094,0.00040077209,0.0029662424,0.0025489174,0.0012302205,0.15252024],"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":[0.00036161768,0.000120674544,0.0012258536,0.00042486232,0.000028996572,0.0003302641,0.00016101704,0.002035525,0.016061528,0.044732045,0.42662603,0.50789154],"study_design_scores_gemma":[0.00004687424,0.00007239483,0.0005482269,0.000043057153,0.000010810675,0.0003499876,0.000050306324,0.005352002,0.008206426,0.012061096,0.9732404,0.000018436358],"about_ca_topic_score_codex":0.001226094,"about_ca_topic_score_gemma":0.0013232927,"teacher_disagreement_score":0.20260103,"about_ca_system_score_codex":0.000753364,"about_ca_system_score_gemma":0.001323071,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2129889089","doi":"10.14778/1920841.1921055","title":"Just-in-time data integration in action","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Data integration; Computer science; Flexibility (engineering); Scalability; Process (computing); Information integration; Set (abstract data type); Ontology-based data integration; Enterprise information integration; System integration; Data virtualization; Database; Data science; Architecture","score_opus":0.054697867894580074,"score_gpt":0.29652098718718384,"score_spread":0.24182311929260375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129889089","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":[0.016178641,0.00048277542,0.94606584,0.0022019194,0.0003825162,0.00029547093,0.0005682845,0.020725807,0.013098645],"genre_scores_gemma":[0.19827916,0.0005705639,0.78016347,0.0013330376,0.00015274769,0.00027667172,0.0023674264,0.0028314611,0.014025516],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9931191,0.0019172145,0.0006193036,0.001634433,0.0022694345,0.00044054323],"domain_scores_gemma":[0.9909017,0.0024449106,0.00042303206,0.004822728,0.0009402416,0.0004673633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00916329,0.0011827957,0.0012247511,0.001267993,0.002215777,0.007005932,0.003983993,0.0024130794,0.0072715087],"category_scores_gemma":[0.01310714,0.001147383,0.0014457,0.0016278668,0.0032416913,0.0140192,0.009578864,0.0027439543,0.003156202],"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":[0.0017767793,0.00096550334,0.016176634,0.0012564328,0.0006312812,0.0026128548,0.010668867,0.024680108,0.047398217,0.31273448,0.060082152,0.5210167],"study_design_scores_gemma":[0.00019740194,0.00026758225,0.0032588148,0.00037068114,0.00037125484,0.0013792277,0.0020290362,0.15551104,0.04346015,0.32230693,0.47054884,0.00029898604],"about_ca_topic_score_codex":0.0050746775,"about_ca_topic_score_gemma":0.0060596038,"teacher_disagreement_score":0.00916329,"about_ca_system_score_codex":0.0009518751,"about_ca_system_score_gemma":0.0022866216,"threshold_uncertainty_score":0.048460662},"labels":[],"label_agreement":null},{"id":"W2130846554","doi":"10.14778/3402707.3402714","title":"RemusDB","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Waterloo","funders":"","keywords":"Failover; Computer science; Downtime; Overhead (engineering); Database; Virtualization; High availability; Operating system; Virtual machine; Cloud computing","score_opus":0.02204543285509568,"score_gpt":0.19394199897665343,"score_spread":0.17189656612155774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2130846554","genre_codex":"software","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":[0.021381028,0.0096751135,0.27266982,0.004298624,0.0020132898,0.0010404884,0.03901657,0.37903917,0.27086598],"genre_scores_gemma":[0.27112252,0.007046517,0.32612482,0.00680197,0.001246215,0.001201236,0.1798624,0.035438064,0.17115638],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99715674,0.0004228656,0.00030481836,0.0005493606,0.0012151984,0.00035103862],"domain_scores_gemma":[0.9977545,0.00024582993,0.000118402895,0.0013206449,0.0003947501,0.00016595247],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019728437,0.0010667004,0.00087974075,0.0022672843,0.00095534866,0.003980661,0.0042330585,0.0011294155,0.045530647],"category_scores_gemma":[0.004826207,0.00078194524,0.00074086065,0.001912713,0.000554128,0.0058011403,0.004874358,0.001531622,0.03501654],"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":[0.0009568906,0.00016356737,0.0022876791,0.0009409062,0.00014879831,0.00047794433,0.00053364097,0.0029803035,0.009953244,0.07246224,0.58189857,0.32719612],"study_design_scores_gemma":[0.000069927526,0.00005811211,0.00073454465,0.000086677835,0.000030764957,0.000451395,0.00013400955,0.009521268,0.010135618,0.011531904,0.96719927,0.00004661777],"about_ca_topic_score_codex":0.0034731405,"about_ca_topic_score_gemma":0.0025904262,"teacher_disagreement_score":0.045530647,"about_ca_system_score_codex":0.0011215487,"about_ca_system_score_gemma":0.001343968,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2131027594","doi":"10.14778/2733004.2733021","title":"DGFIndex for smart grid","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"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; Search engine indexing; Grid file; Database; Grid; Big data; Data mining; Range query (database); Distributed computing; Grid computing; Information retrieval; Web search query; Sargable; Search engine","score_opus":0.009341681633792011,"score_gpt":0.2041474509423922,"score_spread":0.1948057693086002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131027594","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":[0.024558403,0.0035338206,0.5444648,0.0019603497,0.0017266318,0.0008470657,0.030721221,0.3105702,0.08161743],"genre_scores_gemma":[0.2566587,0.0027254377,0.57334083,0.0015233081,0.00044041488,0.0011541216,0.099981,0.017203828,0.046972327],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989303,0.000111479094,0.00011440323,0.00020256973,0.0005306318,0.00011057738],"domain_scores_gemma":[0.9979522,0.00031073767,0.00013690782,0.0010095293,0.00042771114,0.00016283143],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009769055,0.0009983197,0.0006903793,0.001569561,0.00087204017,0.0028836865,0.0022016347,0.00078907114,0.021810837],"category_scores_gemma":[0.005193259,0.00051855575,0.0005963922,0.0031883733,0.00068308006,0.004472331,0.0037445545,0.0014558607,0.013354413],"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":[0.0007040108,0.0001813456,0.00430953,0.0007987483,0.00008903949,0.00048562957,0.0005078199,0.013241562,0.009196919,0.047231402,0.4568122,0.46644175],"study_design_scores_gemma":[0.00016462083,0.00012838178,0.0019606124,0.00016087473,0.00003419,0.00066339417,0.00029050306,0.09018856,0.021335974,0.04771346,0.8372356,0.00012375237],"about_ca_topic_score_codex":0.003420584,"about_ca_topic_score_gemma":0.0024519654,"teacher_disagreement_score":0.021810837,"about_ca_system_score_codex":0.0010159086,"about_ca_system_score_gemma":0.0013012821,"threshold_uncertainty_score":0.07296449},"labels":[],"label_agreement":null},{"id":"W2131164945","doi":"10.14778/2367502.2367512","title":"Solving big data challenges for enterprise application performance management","year":2012,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":232,"is_retracted":false,"has_abstract":true,"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; Big data; Data science; Context (archaeology); Analytics; Instrumentation (computer programming); Data management; Enterprise system; System monitoring; Database; Data mining; Operating system","score_opus":0.050458643549431095,"score_gpt":0.2445988120777852,"score_spread":0.19414016852835408,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131164945","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":[0.19135073,0.036050107,0.48143178,0.2073299,0.0050883275,0.0011970443,0.008345593,0.025337698,0.04386874],"genre_scores_gemma":[0.6577726,0.014313484,0.29573414,0.008292442,0.0044462946,0.00063223863,0.011243711,0.0030529764,0.0045121894],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9864094,0.0034188824,0.0010357962,0.0016978465,0.00647645,0.00096161437],"domain_scores_gemma":[0.94986635,0.018626668,0.0027086071,0.012626779,0.011925154,0.004246465],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018550817,0.0017294053,0.0019178542,0.0027905046,0.0030080816,0.015150659,0.005218593,0.0027205546,0.0022572791],"category_scores_gemma":[0.048912603,0.0011978209,0.0011055203,0.0062833517,0.0024758584,0.022921745,0.007411846,0.008974658,0.0019703852],"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":[0.0010755396,0.0009508397,0.049434606,0.0021240425,0.00067443534,0.0010650433,0.0064672735,0.062278345,0.015052082,0.09677759,0.17727056,0.5868296],"study_design_scores_gemma":[0.00017894764,0.00039364144,0.024932062,0.0010978014,0.00022202998,0.00091787265,0.014291714,0.32185078,0.0137893045,0.36551288,0.2563973,0.0004157155],"about_ca_topic_score_codex":0.0051650647,"about_ca_topic_score_gemma":0.005510787,"teacher_disagreement_score":0.018550817,"about_ca_system_score_codex":0.0022826577,"about_ca_system_score_gemma":0.005188588,"threshold_uncertainty_score":0.09810728},"labels":[],"label_agreement":null},{"id":"W2133246278","doi":"10.14778/1453856.1453895","title":"Efficient search for the top-k probable nearest neighbors in uncertain databases","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":149,"is_retracted":false,"has_abstract":true,"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; Data mining; Query optimization; Online aggregation; k-nearest neighbors algorithm; Query language; Semantics (computer science); Information retrieval; Sargable; Object (grammar); Point (geometry); Database; Feature (linguistics); Web query classification; Web search query; Search engine; Artificial intelligence","score_opus":0.05346318793657699,"score_gpt":0.2715280562480217,"score_spread":0.21806486831144467,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133246278","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":[0.17573752,0.0031427033,0.81504804,0.0012193901,0.00006339722,0.00019182921,0.0012278561,0.001691636,0.0016777643],"genre_scores_gemma":[0.48442835,0.00061501854,0.51118755,0.00018718434,0.000103656675,0.00013810962,0.0022349204,0.00018329892,0.0009218142],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9953053,0.0012680939,0.0006558885,0.0010161038,0.0014244618,0.0003301212],"domain_scores_gemma":[0.98712057,0.009335232,0.00079833774,0.001407169,0.0010495082,0.00028920427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039843055,0.0009378948,0.0033646068,0.0044468977,0.0016462222,0.0033508583,0.0038666788,0.0023076038,0.0016353374],"category_scores_gemma":[0.025117988,0.0010143458,0.0009179089,0.0063813482,0.00092770316,0.00732142,0.0024272026,0.0012199462,0.00058644],"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":[0.0017395514,0.0006127961,0.015211785,0.0006535663,0.00033741727,0.0005832411,0.0011358742,0.45957366,0.0069668246,0.023601962,0.015731843,0.47385144],"study_design_scores_gemma":[0.00005500209,0.000070828595,0.00076031557,0.000018278224,0.000045162742,0.00024574014,0.00034344784,0.9706444,0.0021153642,0.024754303,0.0009205624,0.000026660076],"about_ca_topic_score_codex":0.0071888925,"about_ca_topic_score_gemma":0.01182296,"teacher_disagreement_score":0.0071888925,"about_ca_system_score_codex":0.001354977,"about_ca_system_score_gemma":0.0018274914,"threshold_uncertainty_score":0.021071255},"labels":[],"label_agreement":null},{"id":"W2133343243","doi":"10.14778/1920841.1921042","title":"QUICK","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Numbering; Computer science; Schema (genetic algorithms); Information retrieval; Semantic Web; World Wide Web; Linked data; Programming language","score_opus":0.008614527310647397,"score_gpt":0.21608670940417116,"score_spread":0.20747218209352375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133343243","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015905624,0.002301557,0.01807471,0.007538096,0.008480662,0.00058124255,0.030079339,0.014705605,0.9166482],"genre_scores_gemma":[0.010152808,0.0026389174,0.0129494425,0.004845202,0.0014242409,0.00026707712,0.03040227,0.0047605406,0.93255955],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988827,0.00013973172,0.00006617742,0.0002484686,0.0005460961,0.00011683176],"domain_scores_gemma":[0.995849,0.00079624116,0.00016852368,0.0008755121,0.0017589867,0.00055167585],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0012820599,0.0009758529,0.0007887866,0.003383865,0.0019615402,0.0066671437,0.002735685,0.002383601,0.8492091],"category_scores_gemma":[0.008039958,0.00056961057,0.0008522447,0.0038669216,0.00072811247,0.006710754,0.004163003,0.0019184522,0.66304827],"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":[0.0000823902,0.0000430466,0.00036330428,0.0003380879,0.000009502495,0.00010281151,0.00011767461,0.00014001859,0.0004705667,0.011705386,0.8329434,0.1536837],"study_design_scores_gemma":[0.00001020335,0.000010295296,0.00024176009,0.000059625963,0.0000036690778,0.00007514905,0.00007043595,0.000097881515,0.00021057168,0.0026766974,0.99653614,0.000007620035],"about_ca_topic_score_codex":0.003143039,"about_ca_topic_score_gemma":0.0053227297,"teacher_disagreement_score":0.15079093,"about_ca_system_score_codex":0.0013084313,"about_ca_system_score_gemma":0.0022690357,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2133623058","doi":"10.14778/1687627.1687729","title":"Creating competitive products","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Skyline; Dominance (genetics); Set (abstract data type); Computer science; Competitive advantage; Data mining; Business; Marketing","score_opus":0.011451827037985965,"score_gpt":0.21806801338273515,"score_spread":0.20661618634474918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133623058","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":[0.15363796,0.001433546,0.7893676,0.0013285042,0.0002757921,0.0023531753,0.0025393083,0.0021378992,0.046926208],"genre_scores_gemma":[0.18490776,0.0005147007,0.8044261,0.00027076187,0.000077477096,0.0007952509,0.002989127,0.00036429625,0.005654551],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99585956,0.0011134993,0.00023797559,0.00086353556,0.001622912,0.00030246505],"domain_scores_gemma":[0.9918058,0.0036253682,0.00071900274,0.0017924708,0.0015934462,0.00046397324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003877573,0.0019201135,0.0016765745,0.004045139,0.0020191695,0.003294736,0.002877159,0.0019947658,0.011833327],"category_scores_gemma":[0.01363522,0.0010809607,0.0025300486,0.0036770606,0.0012123758,0.0055595366,0.0036614777,0.0017477892,0.0028796992],"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":[0.00080767856,0.0014877585,0.013894099,0.0017969043,0.00044622968,0.001410111,0.0015092157,0.11107848,0.021305118,0.13114223,0.038356483,0.6767656],"study_design_scores_gemma":[0.00043823812,0.0014122688,0.0038437585,0.00034563668,0.0005014093,0.0031706563,0.0018901689,0.59085816,0.03267386,0.2233596,0.1412832,0.0002230631],"about_ca_topic_score_codex":0.0012228948,"about_ca_topic_score_gemma":0.0022197128,"teacher_disagreement_score":0.011833327,"about_ca_system_score_codex":0.0010500887,"about_ca_system_score_gemma":0.0019397951,"threshold_uncertainty_score":0.039586484},"labels":[],"label_agreement":null},{"id":"W2134489155","doi":"10.14778/2535570.2488332","title":"DAX","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"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; Eventual consistency; Scalability; Cloud computing; Consistency (knowledge bases); Exploit; Database; High availability; Weak consistency; Strong consistency; Distributed computing; Data consistency; Operating system; Consistency model; Computer security","score_opus":0.007384205337364365,"score_gpt":0.18660272779575862,"score_spread":0.17921852245839426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134489155","genre_codex":"other","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":[0.014191234,0.0032274816,0.15192212,0.0035356113,0.003613438,0.0010606401,0.020104423,0.15692061,0.6454244],"genre_scores_gemma":[0.11068611,0.0033675747,0.11308682,0.003926658,0.001176571,0.0012334198,0.08361211,0.02320612,0.6597046],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99854267,0.0001743498,0.00009822143,0.00034070402,0.00062034716,0.0002237859],"domain_scores_gemma":[0.99794334,0.00017267183,0.00009053956,0.00082443847,0.0006708585,0.00029813853],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014635895,0.0010443656,0.00064345624,0.0012296074,0.001272261,0.0040970235,0.0030737796,0.0011618563,0.171357],"category_scores_gemma":[0.0030106183,0.0006343267,0.0005729822,0.0014161275,0.0004892579,0.0043220567,0.0040941313,0.001984432,0.13386506],"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":[0.0007243347,0.0002147817,0.0017137802,0.00052711,0.000052289954,0.00027645545,0.00030786384,0.0011629584,0.0116949715,0.059423424,0.6687797,0.25512236],"study_design_scores_gemma":[0.00004335323,0.00005299221,0.00050764036,0.000033206423,0.00000894845,0.00015769983,0.00005453709,0.002264338,0.0040503684,0.0037046657,0.9891049,0.000017299411],"about_ca_topic_score_codex":0.0019430519,"about_ca_topic_score_gemma":0.0016706683,"teacher_disagreement_score":0.171357,"about_ca_system_score_codex":0.000952995,"about_ca_system_score_gemma":0.0013328462,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2140944908","doi":"10.14778/1687627.1687722","title":"Improving the performance of list intersection","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":45,"is_retracted":false,"has_abstract":true,"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; Intersection (aeronautics); Identifier; Overhead (engineering); Hash function; Sorting; Cache; Parallel computing; Hash table; Data structure; Algorithm; Theoretical computer science; Operating system; Programming language","score_opus":0.0073209096403227154,"score_gpt":0.2025471456109298,"score_spread":0.19522623597060706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140944908","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":[0.20298858,0.0029505051,0.7704503,0.00062391435,0.00017193369,0.00015516112,0.00031774267,0.015892424,0.006449464],"genre_scores_gemma":[0.5777852,0.0007589205,0.41635042,0.00016238619,0.00014023884,0.00015626149,0.0010718214,0.00069686206,0.0028778876],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9965609,0.00060615246,0.00029305075,0.000468936,0.0015713314,0.0004996203],"domain_scores_gemma":[0.9885107,0.0057624183,0.00058908935,0.0024516566,0.002463617,0.00022236843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022639127,0.0010212585,0.0011762041,0.0026875495,0.001187651,0.0025911606,0.0025167824,0.0008814541,0.0032800275],"category_scores_gemma":[0.016847821,0.0004556359,0.00049439445,0.0049963547,0.000808825,0.005817433,0.0029808895,0.0011214195,0.0018600479],"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":[0.0016934623,0.0002827146,0.01163524,0.00026682572,0.000115038536,0.0001392042,0.0005880866,0.07680458,0.040203374,0.01560503,0.010648329,0.8420182],"study_design_scores_gemma":[0.000095728814,0.00042753233,0.0019229227,0.00002684196,0.0000611903,0.0003179142,0.00023422997,0.8979215,0.078125015,0.012138858,0.008673761,0.000054519467],"about_ca_topic_score_codex":0.0025779747,"about_ca_topic_score_gemma":0.002447515,"teacher_disagreement_score":0.0032800275,"about_ca_system_score_codex":0.001406499,"about_ca_system_score_gemma":0.0024809663,"threshold_uncertainty_score":0.011972904},"labels":[],"label_agreement":null},{"id":"W2146496747","doi":"10.14778/2732951.2732957","title":"Workload matters","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"SPARQL; RDF; Workload; Computer science; RDF Schema; Simple Knowledge Organization System; Cwm; Set (abstract data type); Linked data; RDF query language; Semantic Web; Information retrieval; Database; Programming language; Operating system; Search engine; Web search query","score_opus":0.007596812784350849,"score_gpt":0.19535866088341539,"score_spread":0.18776184809906454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146496747","genre_codex":"other","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":[0.028466364,0.021012854,0.02697408,0.42146066,0.035269614,0.00032039307,0.002676558,0.0016135086,0.46220595],"genre_scores_gemma":[0.42552805,0.022544814,0.0120499935,0.19402082,0.033584565,0.00060890213,0.0033530125,0.0036895082,0.30462036],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9942814,0.0013476708,0.00038417368,0.0012094645,0.0018796922,0.0008976078],"domain_scores_gemma":[0.98355156,0.004693776,0.0012749622,0.0022159861,0.004826421,0.0034373016],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005470784,0.00089414266,0.00076858036,0.000963227,0.0026114886,0.00959912,0.0020814815,0.002942165,0.08762405],"category_scores_gemma":[0.03574364,0.00044097332,0.0006141417,0.0013875721,0.003310221,0.0116492435,0.0044250744,0.00373241,0.03440248],"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":[0.00033709183,0.00018572676,0.008352087,0.0006866767,0.00007838289,0.0003737962,0.0023767832,0.0004594502,0.0021430282,0.15676807,0.58430046,0.24393837],"study_design_scores_gemma":[0.000036892456,0.00008302489,0.0037093319,0.00041961562,0.000035601483,0.0007384454,0.0026887073,0.00047094122,0.0008248834,0.10268072,0.8882731,0.00003870635],"about_ca_topic_score_codex":0.0029691125,"about_ca_topic_score_gemma":0.00356396,"teacher_disagreement_score":0.08762405,"about_ca_system_score_codex":0.0034366176,"about_ca_system_score_gemma":0.0040288847,"threshold_uncertainty_score":0.29313165},"labels":[],"label_agreement":null},{"id":"W2147033904","doi":"10.14778/1687627.1687702","title":"Power-law based estimation of set similarity join size","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Hash function; Exploit; Computer science; Set (abstract data type); Similarity (geometry); Algorithm; Data mining; Signature (topology); Representation (politics); Nearest neighbor search; Mathematics; Theoretical computer science; Pattern recognition (psychology); Artificial intelligence; Law","score_opus":0.013244692484878601,"score_gpt":0.23772138781357566,"score_spread":0.22447669532869705,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2147033904","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":[0.042851247,0.00020362556,0.95501137,0.00009615765,0.000028975757,0.000076802906,0.00011132921,0.00076766073,0.0008528099],"genre_scores_gemma":[0.51651496,0.00022743143,0.4816266,0.00007090865,0.00014983742,0.00022804532,0.00040104505,0.00016607049,0.00061513484],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9939237,0.0010473501,0.0003471848,0.0008231912,0.003647274,0.00021119727],"domain_scores_gemma":[0.9658494,0.022524199,0.0036521247,0.004069834,0.0033381023,0.00056641106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032359632,0.0006586469,0.0011066635,0.0034832868,0.0005188052,0.001414684,0.0027054443,0.0008905126,0.0013998036],"category_scores_gemma":[0.035606533,0.00056318234,0.0005404153,0.0022022862,0.0009131025,0.004504072,0.0017678239,0.001293785,0.00083362736],"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":[0.00090078486,0.0005571353,0.03837159,0.00039175298,0.00020774003,0.00043574977,0.0006937365,0.22919632,0.09769297,0.03956952,0.0033168183,0.58866584],"study_design_scores_gemma":[0.00001718581,0.000098927885,0.0018269913,0.0000111318395,0.000010933701,0.00028524225,0.000040342908,0.9718726,0.016384285,0.00865763,0.0007648937,0.000029751109],"about_ca_topic_score_codex":0.0007166724,"about_ca_topic_score_gemma":0.0007312026,"teacher_disagreement_score":0.0034832868,"about_ca_system_score_codex":0.0007429418,"about_ca_system_score_gemma":0.00071698497,"threshold_uncertainty_score":0.017113626},"labels":[],"label_agreement":null},{"id":"W2148524305","doi":"10.14778/1687627.1687771","title":"Framework for evaluating clustering algorithms in duplicate detection","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":232,"is_retracted":false,"has_abstract":true,"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; Cluster analysis; Data mining; Data deduplication; Process (computing); Algorithm; Machine learning; Database","score_opus":0.2418184006816228,"score_gpt":0.4629068859449667,"score_spread":0.2210884852633439,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148524305","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":[0.028105475,0.0013989379,0.95894456,0.0006685856,0.00012316741,0.00202075,0.0008784202,0.0018990178,0.005961181],"genre_scores_gemma":[0.111879796,0.000395148,0.8841136,0.00017435553,0.00007874138,0.0016521391,0.00086940237,0.00016274957,0.00067410903],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9419968,0.032028615,0.004135181,0.003337024,0.017257826,0.0012445531],"domain_scores_gemma":[0.9364167,0.036468968,0.0055688643,0.008679273,0.011738107,0.001128086],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.058792125,0.0026015257,0.0029192106,0.011800542,0.002354739,0.0056906724,0.0053763054,0.0042554396,0.0017944205],"category_scores_gemma":[0.10664215,0.00083471637,0.0024088775,0.009012589,0.0026959646,0.00483065,0.0059125344,0.0026109472,0.0008506694],"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":[0.0009888812,0.001387541,0.0132535575,0.0012733405,0.0012114419,0.00025337542,0.0007350098,0.5713453,0.009797089,0.15096971,0.008617209,0.24016754],"study_design_scores_gemma":[0.0001858201,0.0011664745,0.0027726737,0.00016800388,0.00019230189,0.00022330823,0.00028831643,0.93214977,0.0057406253,0.04993129,0.007071983,0.00010932784],"about_ca_topic_score_codex":0.007920526,"about_ca_topic_score_gemma":0.0051574674,"teacher_disagreement_score":0.058792125,"about_ca_system_score_codex":0.004568572,"about_ca_system_score_gemma":0.004954954,"threshold_uncertainty_score":0.3109262},"labels":[],"label_agreement":null},{"id":"W2151131744","doi":"10.14778/1687553.1687573","title":"Efficient index compression in DB2 LUW","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; IBM (Canada)","funders":"","keywords":"Computer science; Index (typography); Workload; Unix; Database; Response time; Data compression; Memory footprint; Real-time computing; Operating system","score_opus":0.007764069645988815,"score_gpt":0.22779144631109907,"score_spread":0.22002737666511027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2151131744","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":[0.08668643,0.0019903425,0.8805865,0.00073109043,0.0001892374,0.0005210422,0.0006429768,0.019139577,0.009512804],"genre_scores_gemma":[0.29564628,0.00079229876,0.6928241,0.00070648233,0.00018268074,0.000530274,0.0018761419,0.00096589315,0.006475881],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99846834,0.00020505571,0.00018610519,0.00019707531,0.00078481017,0.00015847487],"domain_scores_gemma":[0.99794155,0.00045760773,0.00019304112,0.0006840203,0.0006279294,0.00009586716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012099163,0.0007719572,0.00082796125,0.0016202852,0.00088386575,0.0021744706,0.0021263873,0.00083856453,0.0018031596],"category_scores_gemma":[0.004095724,0.0006388933,0.0003394597,0.0033251403,0.0006162828,0.0032875235,0.001974151,0.0009133366,0.0014613194],"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":[0.0015461638,0.00065235514,0.0046393746,0.00048993586,0.00009200629,0.0007513876,0.0006880305,0.028414445,0.16153815,0.02714741,0.037157945,0.73688275],"study_design_scores_gemma":[0.00025860907,0.00071119436,0.0022569562,0.000071029215,0.00006453464,0.0016790567,0.00032885978,0.6606423,0.26119286,0.017079836,0.055594873,0.000119949],"about_ca_topic_score_codex":0.0017306685,"about_ca_topic_score_gemma":0.0019873278,"teacher_disagreement_score":0.0021744706,"about_ca_system_score_codex":0.0007059591,"about_ca_system_score_gemma":0.001017756,"threshold_uncertainty_score":0.0063987374},"labels":[],"label_agreement":null},{"id":"W2152102944","doi":"10.14778/2536354.2536355","title":"Hybrid storage management for database systems","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"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; Workload; Storage management; Data striping; Computer data storage; Flash (photography); Operating system; Distributed computing","score_opus":0.01574298037299049,"score_gpt":0.22913046462758513,"score_spread":0.21338748425459464,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2152102944","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":[0.10305401,0.007024835,0.8545198,0.0018521005,0.00039941876,0.0004744256,0.001198761,0.016059656,0.015416996],"genre_scores_gemma":[0.6339836,0.0012358148,0.35470188,0.00047631134,0.00015976206,0.0003178763,0.001510284,0.00035418945,0.007260194],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99837923,0.0003928739,0.00019729578,0.00026031918,0.0006348641,0.00013542481],"domain_scores_gemma":[0.9970067,0.0006998903,0.00017624571,0.0012366251,0.00072309753,0.0001574275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017753883,0.0005782719,0.00074288936,0.0008560927,0.0010123961,0.0035167984,0.0032392717,0.00090628373,0.004120574],"category_scores_gemma":[0.0037632713,0.0005507153,0.00042616663,0.001735345,0.00044663838,0.004850277,0.0024430067,0.0008859695,0.0009998926],"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":[0.0012026477,0.00043663607,0.008077495,0.0010922229,0.00041709427,0.0005254565,0.00069258665,0.11321903,0.045769755,0.11177934,0.05335271,0.66343504],"study_design_scores_gemma":[0.00021225627,0.00031355472,0.0022942966,0.00008858973,0.00014802169,0.0008535373,0.00035105046,0.8135246,0.026586352,0.089440905,0.066099964,0.0000868254],"about_ca_topic_score_codex":0.0018928078,"about_ca_topic_score_gemma":0.0022831226,"teacher_disagreement_score":0.004120574,"about_ca_system_score_codex":0.001314309,"about_ca_system_score_gemma":0.0009165713,"threshold_uncertainty_score":0.013784766},"labels":[],"label_agreement":null},{"id":"W2153257312","doi":"10.14778/1687553.1687599","title":"Linkage Query Writer","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":25,"is_retracted":false,"has_abstract":true,"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; SQL; Linkage (software); Interface (matter); Query language; Relational database; Linked data; Information retrieval; Process (computing); Programming language; Semantic Web","score_opus":0.1069606726619127,"score_gpt":0.37673219067138963,"score_spread":0.26977151800947696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153257312","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025054528,0.0005531538,0.73061556,0.0013022188,0.00037891505,0.0009559208,0.010260396,0.2336056,0.01982274],"genre_scores_gemma":[0.055718172,0.0010721192,0.81254447,0.00393288,0.0005002337,0.0017905219,0.044807673,0.044889163,0.034744777],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9838792,0.0037951348,0.0021213323,0.003115896,0.0061525963,0.0009358173],"domain_scores_gemma":[0.966402,0.011148562,0.001579564,0.010220328,0.009620065,0.0010294256],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019570697,0.001707782,0.00256768,0.004568843,0.0020735916,0.009725362,0.0065158373,0.003025261,0.057845797],"category_scores_gemma":[0.04496669,0.0019669675,0.0025509973,0.0041403547,0.0014763887,0.011793788,0.01100076,0.0040966147,0.038239606],"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":[0.0013436674,0.0004344051,0.006096487,0.0016285533,0.00030664826,0.0008737358,0.0019344353,0.004429451,0.011327384,0.08474396,0.47886577,0.40801543],"study_design_scores_gemma":[0.00041257526,0.00016938854,0.000970328,0.00023174097,0.00013681182,0.0009681618,0.0004797951,0.045655664,0.030283116,0.038893323,0.88153625,0.0002627447],"about_ca_topic_score_codex":0.0028881829,"about_ca_topic_score_gemma":0.002081652,"teacher_disagreement_score":0.057845797,"about_ca_system_score_codex":0.0015179254,"about_ca_system_score_gemma":0.0045759683,"threshold_uncertainty_score":0.19351351},"labels":[],"label_agreement":null},{"id":"W2153974857","doi":"10.14778/3402707.3402754","title":"Efficient rank join with aggregation constraints","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Join (topology); Computer science; Rank (graph theory); Probabilistic logic; Semantics (computer science); Theoretical computer science; Data mining; Artificial intelligence; Mathematics; Programming language","score_opus":0.01904745898815237,"score_gpt":0.1970963168557973,"score_spread":0.1780488578676449,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153974857","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":[0.011833892,0.00019941317,0.98393285,0.00030173617,0.000039122016,0.00008900212,0.00025777647,0.0016200576,0.0017261156],"genre_scores_gemma":[0.3058473,0.00026434625,0.6888143,0.00031377107,0.00016975305,0.00014220347,0.00094788254,0.00047616073,0.0030242528],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9906008,0.0028003452,0.0007139688,0.0011418572,0.0040956032,0.00064747146],"domain_scores_gemma":[0.9857119,0.007363549,0.0009690133,0.0042447387,0.0014202298,0.00029074893],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007126879,0.0009387786,0.0014105942,0.0010698283,0.0012705995,0.003643682,0.0021372156,0.0015230044,0.0029140094],"category_scores_gemma":[0.021054652,0.0006025264,0.0009192331,0.0024795413,0.0012349995,0.0053627975,0.003525723,0.0022840742,0.0011940161],"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":[0.00090662297,0.00037514739,0.004607774,0.0005301669,0.00016602957,0.0004337649,0.00072022766,0.275109,0.025691364,0.27213162,0.023639163,0.39568922],"study_design_scores_gemma":[0.00007174754,0.00012618532,0.00042228613,0.000019789517,0.00003602051,0.00036080935,0.00013228056,0.84498894,0.014061573,0.12790069,0.0118231075,0.000056540604],"about_ca_topic_score_codex":0.0020556904,"about_ca_topic_score_gemma":0.004163596,"teacher_disagreement_score":0.007126879,"about_ca_system_score_codex":0.00075603,"about_ca_system_score_gemma":0.0024805735,"threshold_uncertainty_score":0.037690997},"labels":[],"label_agreement":null},{"id":"W2154764667","doi":"10.14778/1453856.1453953","title":"On efficiently searching trajectories and archival data for historical similarities","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; IBM (Canada)","funders":"","keywords":"Computer science; False positive paradox; Series (stratigraphy); Computation; Similarity (geometry); Set (abstract data type); Data mining; Class (philosophy); Algorithm; Artificial intelligence","score_opus":0.05566045184704826,"score_gpt":0.2447184203810245,"score_spread":0.18905796853397625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154764667","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":[0.20366038,0.0027010203,0.7866699,0.0008522773,0.00007106203,0.0002870135,0.0017443671,0.0026755573,0.0013384013],"genre_scores_gemma":[0.5012844,0.0011778076,0.4910282,0.00012896798,0.00017796071,0.00022668825,0.004232688,0.00020016427,0.0015431892],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979784,0.00042601745,0.00027081175,0.0005119359,0.0006746315,0.00013826462],"domain_scores_gemma":[0.9904585,0.005757612,0.0013199286,0.001130638,0.0010261993,0.00030714134],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025073504,0.0010989441,0.0020821316,0.006775939,0.0009487027,0.0023618718,0.0018917045,0.0016835965,0.0012175519],"category_scores_gemma":[0.024162594,0.00060964737,0.0008059757,0.0076133166,0.00077848684,0.0069058756,0.0015760605,0.00089056016,0.00069700874],"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":[0.0009976014,0.00039261204,0.021957302,0.0008804666,0.00025468497,0.0007201771,0.0019870605,0.22110574,0.014199549,0.02520352,0.009279162,0.7030222],"study_design_scores_gemma":[0.00004374326,0.00025378747,0.0039440934,0.000048226142,0.00004538258,0.0005367252,0.0008558574,0.94630194,0.005460141,0.037916254,0.004546142,0.00004771582],"about_ca_topic_score_codex":0.0068331324,"about_ca_topic_score_gemma":0.006477459,"teacher_disagreement_score":0.0068331324,"about_ca_system_score_codex":0.0011512997,"about_ca_system_score_gemma":0.0012009614,"threshold_uncertainty_score":0.0135867},"labels":[],"label_agreement":null},{"id":"W2154768509","doi":"10.14778/1453856.1453934","title":"Efficient network aware search in collaborative tagging sites","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":112,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Popularity; Cluster analysis; Seekers; Context (archaeology); Upper and lower bounds; Heuristic; Space (punctuation); Information retrieval; Data mining; Machine learning; Artificial intelligence; Mathematics; Geography","score_opus":0.018816654655215464,"score_gpt":0.23477065336095104,"score_spread":0.2159539987057356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154768509","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":[0.34673777,0.0015544937,0.64321226,0.0007131455,0.000045400913,0.00024278507,0.0008934593,0.0018732917,0.004727425],"genre_scores_gemma":[0.7774791,0.00035911234,0.21722426,0.000103844584,0.000065490334,0.00012337025,0.0012889899,0.00018156455,0.0031741671],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99717194,0.0009333146,0.00021334026,0.00067240925,0.0006092726,0.00039973424],"domain_scores_gemma":[0.989407,0.0063480297,0.001103732,0.001869678,0.0008283364,0.00044322922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033433486,0.00081422663,0.002247597,0.0031864238,0.0017521456,0.002921827,0.0027721478,0.0018065227,0.001515398],"category_scores_gemma":[0.017424155,0.00075781415,0.000850577,0.005506529,0.00119131,0.0054787574,0.0026987444,0.00090648496,0.00095634995],"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":[0.0017021932,0.0006145177,0.029224414,0.0007888292,0.0003021344,0.0005879742,0.0022187154,0.587754,0.023107907,0.045677185,0.011010867,0.2970113],"study_design_scores_gemma":[0.00005298417,0.000107094274,0.0016530032,0.000017741264,0.00005585298,0.00023097666,0.00033383653,0.9607445,0.0040275226,0.031034673,0.0017157829,0.00002599858],"about_ca_topic_score_codex":0.005331591,"about_ca_topic_score_gemma":0.009775597,"teacher_disagreement_score":0.005331591,"about_ca_system_score_codex":0.0015558944,"about_ca_system_score_gemma":0.0016143281,"threshold_uncertainty_score":0.01768148},"labels":[],"label_agreement":null},{"id":"W2156972533","doi":"10.14778/1920841.1920874","title":"SECRET","year":2010,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":109,"is_retracted":false,"has_abstract":true,"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); Key (lock); Variation (astronomy); Range (aeronautics); Window (computing); World Wide Web; Programming language; Computer security; Engineering","score_opus":0.004724064018066928,"score_gpt":0.20121431826377942,"score_spread":0.19649025424571248,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156972533","genre_codex":"other","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":[0.021342011,0.0017680075,0.3278565,0.007424854,0.0017454526,0.001013293,0.032850992,0.046866663,0.5591322],"genre_scores_gemma":[0.304287,0.004110622,0.13730028,0.005588649,0.0009455418,0.0011567565,0.06285513,0.011300148,0.47245592],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99821436,0.00023428735,0.00014758248,0.00037613156,0.00082245935,0.00020524026],"domain_scores_gemma":[0.9962876,0.00059564353,0.00027185687,0.0016713516,0.00092561584,0.00024785567],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017777085,0.0006213028,0.00055393635,0.0011111324,0.00082511286,0.004968903,0.0021587599,0.001377278,0.07540943],"category_scores_gemma":[0.0073209866,0.00053195615,0.0006505111,0.0012185815,0.0008053074,0.0077379504,0.002654738,0.0011040505,0.047999013],"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":[0.00076775945,0.00012391174,0.004996776,0.00051589333,0.00004944202,0.00041492394,0.0005794838,0.008812725,0.0048578735,0.49594897,0.29093382,0.19199847],"study_design_scores_gemma":[0.000047390662,0.000054395976,0.0005317967,0.0000848311,0.000018192346,0.00047231393,0.00009469346,0.016498547,0.003647977,0.056417685,0.9220969,0.000035369212],"about_ca_topic_score_codex":0.0025196727,"about_ca_topic_score_gemma":0.0015727683,"teacher_disagreement_score":0.07540943,"about_ca_system_score_codex":0.001227042,"about_ca_system_score_gemma":0.0020634406,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2159157420","doi":"10.14778/1454159.1454222","title":"Scheduling continuous queries in data stream management systems","year":2008,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"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; Salient; Scheduling (production processes); Distributed computing; Operations research; Mathematical optimization; Artificial intelligence","score_opus":0.03159481624336894,"score_gpt":0.24339173839996645,"score_spread":0.2117969221565975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159157420","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":[0.12519142,0.005833382,0.8604769,0.001910274,0.00037126377,0.0003123994,0.00038907962,0.0021875224,0.003327765],"genre_scores_gemma":[0.82634693,0.0021144773,0.16849515,0.00032083123,0.00038831634,0.00019573669,0.0003881988,0.00014175285,0.0016085199],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974962,0.0009128967,0.00026095606,0.00035194986,0.0007478809,0.00023007055],"domain_scores_gemma":[0.9962631,0.002236809,0.00022262547,0.00035841615,0.00047132059,0.00044765332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005133814,0.00048652594,0.0011424022,0.00062298117,0.0008626275,0.0027020203,0.0016323054,0.0009451196,0.0010663657],"category_scores_gemma":[0.008772746,0.0004761955,0.00038142514,0.0016864857,0.000987894,0.0021437448,0.0012604654,0.0012843033,0.00023308875],"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":[0.0024677676,0.000629263,0.0086117815,0.0009739257,0.00020583163,0.0007326429,0.001780196,0.54257566,0.038157485,0.15146603,0.021778686,0.2306207],"study_design_scores_gemma":[0.00011304617,0.00011573372,0.000594024,0.00001880353,0.00002283124,0.000059560065,0.00016318892,0.96768534,0.0031926592,0.021736855,0.0062764245,0.00002140495],"about_ca_topic_score_codex":0.0044741533,"about_ca_topic_score_gemma":0.003050505,"teacher_disagreement_score":0.005133814,"about_ca_system_score_codex":0.0014070489,"about_ca_system_score_gemma":0.0017017404,"threshold_uncertainty_score":0.027150512},"labels":[],"label_agreement":null},{"id":"W2160152607","doi":"10.14778/1687627.1687754","title":"Efficient method for maximizing bichromatic reverse nearest neighbor","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":129,"is_retracted":false,"has_abstract":true,"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":"Computer science; k-nearest neighbors algorithm; Point (geometry); Best bin first; Exponential function; Algorithm; Exponential growth; Theoretical computer science; Mathematics; Artificial intelligence","score_opus":0.014581193934234049,"score_gpt":0.25726412710710916,"score_spread":0.24268293317287512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160152607","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":[0.013429034,0.0005591736,0.98252904,0.00013040447,0.00004309442,0.00011958795,0.00013182592,0.0010281708,0.0020296804],"genre_scores_gemma":[0.09750411,0.00022960654,0.8994473,0.00008888402,0.000038437687,0.00020120134,0.00040639602,0.00019085256,0.0018932427],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99764204,0.0005515056,0.0001347962,0.0004520314,0.0010290174,0.00019069249],"domain_scores_gemma":[0.9980293,0.0007052377,0.0001990311,0.00038534516,0.00060479256,0.00007631326],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015720602,0.0011313003,0.0018772769,0.0023617866,0.00086056115,0.0010226021,0.0026746388,0.0011679356,0.004400157],"category_scores_gemma":[0.0062883524,0.00063561334,0.0008446203,0.0028124324,0.0006000947,0.0022648457,0.002824784,0.00091342797,0.0018032754],"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":[0.00044815734,0.00023186857,0.0018362644,0.00037556066,0.00008333039,0.00015672174,0.00027931132,0.15288286,0.015917938,0.019222856,0.009857926,0.79870725],"study_design_scores_gemma":[0.00006724329,0.00009956697,0.00057981187,0.000025534579,0.00002845482,0.0003891127,0.000107584994,0.96903056,0.010167378,0.013292927,0.006174479,0.000037310598],"about_ca_topic_score_codex":0.0038927475,"about_ca_topic_score_gemma":0.0061986945,"teacher_disagreement_score":0.004400157,"about_ca_system_score_codex":0.00092426845,"about_ca_system_score_gemma":0.0016593373,"threshold_uncertainty_score":0.014720023},"labels":[],"label_agreement":null},{"id":"W2162783807","doi":"10.14778/2021017.2021025","title":"Keyword search in graphs","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":132,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Substructure; Clique; Computer science; Graph; Combinatorics; Theoretical computer science; Time complexity; Mathematics; Algorithm","score_opus":0.036510813204070874,"score_gpt":0.2195111061961708,"score_spread":0.18300029299209994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162783807","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":[0.04531558,0.005098478,0.9254759,0.0026069314,0.00017137,0.0004927976,0.0040235603,0.0029681614,0.013847203],"genre_scores_gemma":[0.28606388,0.004916594,0.69141895,0.0008215114,0.00024157985,0.00037560557,0.006168046,0.00051714573,0.009476676],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99727684,0.000910157,0.00024048227,0.0008716729,0.00045007383,0.00025078538],"domain_scores_gemma":[0.99399275,0.0041166656,0.0005052529,0.0007695441,0.0004229021,0.0001928289],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012357202,0.0009957463,0.0016016557,0.0031398495,0.0015225998,0.0036798355,0.0019880033,0.0020240827,0.0069523132],"category_scores_gemma":[0.011150416,0.0009229629,0.001308278,0.0077954847,0.0013177645,0.008909615,0.0024320106,0.0012962526,0.0029225647],"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":[0.00062776945,0.00028094646,0.0031632774,0.0028884711,0.00029710287,0.00085721834,0.00122683,0.21969709,0.011904722,0.31667033,0.055028748,0.38735747],"study_design_scores_gemma":[0.000092065544,0.000105362175,0.0006494442,0.00012692486,0.00007200029,0.0009228826,0.00043345295,0.28435084,0.004845701,0.6707673,0.037582252,0.000051719744],"about_ca_topic_score_codex":0.0047476958,"about_ca_topic_score_gemma":0.0049530873,"teacher_disagreement_score":0.0069523132,"about_ca_system_score_codex":0.0020369252,"about_ca_system_score_gemma":0.0016293918,"threshold_uncertainty_score":0.023257792},"labels":[],"label_agreement":null},{"id":"W2163438246","doi":"10.14778/1687553.1687556","title":"StatAdvisor","year":2009,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"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; IBM; Matching (statistics); SQL; Data mining; Oracle; Key (lock); Plan (archaeology); Construct (python library); Workload; Query plan; Information retrieval; Database; Statistics; Software engineering; Web search query; Mathematics; Sargable; Search engine; Programming language","score_opus":0.00712529554645085,"score_gpt":0.21475078658725172,"score_spread":0.20762549104080086,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163438246","genre_codex":"software","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":[0.011992129,0.0013447163,0.40501705,0.001105998,0.0003892513,0.0011853796,0.038392372,0.46080488,0.0797682],"genre_scores_gemma":[0.13906278,0.0020577647,0.5529056,0.0020900583,0.00038465066,0.0027642364,0.12350705,0.070695676,0.106532186],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99764025,0.00033000633,0.00021576582,0.00070030463,0.0009472849,0.00016640445],"domain_scores_gemma":[0.9940042,0.002268154,0.0003453553,0.0020756542,0.0009519937,0.00035467595],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00264646,0.0010755545,0.001113621,0.0017615183,0.00057710876,0.0043605617,0.004647346,0.000922018,0.057755012],"category_scores_gemma":[0.014144393,0.0015193694,0.000836488,0.0021419963,0.0006690609,0.003950494,0.003846236,0.0022051907,0.02985708],"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":[0.0016705556,0.00016456154,0.00471972,0.0010444418,0.00011200043,0.0002573548,0.00043994703,0.0068552108,0.006782078,0.032517113,0.4739409,0.47149613],"study_design_scores_gemma":[0.00065123534,0.0003245173,0.0032766836,0.00026178855,0.00007430226,0.00063406955,0.0001676511,0.09421501,0.013240962,0.032924537,0.8541039,0.00012533578],"about_ca_topic_score_codex":0.0021584455,"about_ca_topic_score_gemma":0.002656563,"teacher_disagreement_score":0.057755012,"about_ca_system_score_codex":0.0010354833,"about_ca_system_score_gemma":0.0026822905,"threshold_uncertainty_score":0.19320977},"labels":[],"label_agreement":null},{"id":"W2168284908","doi":"10.14778/2757807.2757808","title":"ALID","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China; National Science Foundation","keywords":"Speedup; Scalability; Computer science; Parameterized complexity; Time complexity; Graph; Theoretical computer science; Algorithm; Parallel computing","score_opus":0.021096972285128485,"score_gpt":0.22558753531170597,"score_spread":0.20449056302657748,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168284908","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006679036,0.0015945083,0.46421322,0.002248158,0.0021200331,0.0009590458,0.03392717,0.31579623,0.17246261],"genre_scores_gemma":[0.09629298,0.0019290522,0.5020933,0.0037869238,0.00076342805,0.0017422823,0.14833607,0.045132793,0.19992317],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99777836,0.00023658424,0.0001423052,0.0006297792,0.00091783813,0.00029517413],"domain_scores_gemma":[0.99772424,0.0003682309,0.00013225601,0.00088151306,0.00068551156,0.00020817271],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0015403434,0.0017774208,0.0012906117,0.002492078,0.0013039309,0.004255086,0.004696301,0.0020991426,0.13753492],"category_scores_gemma":[0.005522452,0.0010828169,0.0017411188,0.0020652597,0.00063494314,0.0041675977,0.0047681755,0.0023287304,0.11488284],"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":[0.0004969533,0.00017210368,0.0023347724,0.00059894554,0.00012792356,0.00031935112,0.00021062285,0.0060957572,0.006568289,0.026535766,0.674663,0.28187647],"study_design_scores_gemma":[0.00016926008,0.00008771782,0.0013583787,0.00009421407,0.000049071066,0.00051195925,0.00014820567,0.06458357,0.009873903,0.031344242,0.89169145,0.000088040135],"about_ca_topic_score_codex":0.0045157694,"about_ca_topic_score_gemma":0.005472243,"teacher_disagreement_score":0.8624651,"about_ca_system_score_codex":0.0013792096,"about_ca_system_score_gemma":0.002232915,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2168773517","doi":"10.14778/2556549.2556568","title":"Expressiveness and complexity of order dependencies","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada); York University; University of Toronto","funders":"","keywords":"Lexicographical order; Inference; Functional dependency; Tuple; Completeness (order theory); Computer science; Time complexity; SQL; Theoretical computer science; Dependency (UML); Class (philosophy); Rule of inference; Mathematics; Algorithm; Relational database; Discrete mathematics; Data mining; Artificial intelligence; Combinatorics; Programming language","score_opus":0.02515564258788441,"score_gpt":0.22424121724989018,"score_spread":0.19908557466200577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168773517","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":[0.34040356,0.0030785098,0.6165432,0.0075197727,0.00009593046,0.00025838066,0.0062181195,0.001289032,0.024593446],"genre_scores_gemma":[0.87127835,0.0019448391,0.11753345,0.00063154474,0.00021745353,0.00024963595,0.0035248627,0.0002624399,0.0043574446],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9919974,0.001620542,0.00092482165,0.0018535731,0.0028939277,0.00070961914],"domain_scores_gemma":[0.95797676,0.034866598,0.0018849559,0.0031033617,0.0015304589,0.0006378205],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003846278,0.00067100214,0.0010629285,0.0017225172,0.0013121769,0.006080746,0.0025887745,0.0015058904,0.0035912741],"category_scores_gemma":[0.02635304,0.001246442,0.0020106053,0.0034736143,0.0025661027,0.0141418455,0.0036125274,0.0040697935,0.00037268118],"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":[0.00035202413,0.00026724735,0.011607146,0.0009302885,0.00015054298,0.0006768813,0.0017179773,0.10555193,0.006507279,0.779049,0.006495326,0.08669441],"study_design_scores_gemma":[0.000034581833,0.000025691415,0.0010882108,0.000056740417,0.00006181056,0.0004949966,0.00031002017,0.12890756,0.003606773,0.85959476,0.005785014,0.00003384169],"about_ca_topic_score_codex":0.0035412044,"about_ca_topic_score_gemma":0.002841934,"teacher_disagreement_score":0.006080746,"about_ca_system_score_codex":0.0027901265,"about_ca_system_score_gemma":0.0018015492,"threshold_uncertainty_score":0.020341277},"labels":[],"label_agreement":null},{"id":"W2170712852","doi":"10.14778/2536258.2536262","title":"Discovering denial constraints","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":252,"is_retracted":false,"has_abstract":true,"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; Rotation formalisms in three dimensions; Scalability; Inference; Data integrity; Set (abstract data type); Functional dependency; Constraint (computer-aided design); Theoretical computer science; Rank (graph theory); Semantics (computer science); Function (biology); Data mining; Artificial intelligence; Programming language; Database; Relational database","score_opus":0.08220481571088507,"score_gpt":0.3393964871060694,"score_spread":0.25719167139518434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170712852","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":[0.063680425,0.0010135486,0.90881777,0.004080155,0.00025331488,0.0008495224,0.0064142724,0.0037341616,0.011156847],"genre_scores_gemma":[0.35496354,0.000549057,0.62687516,0.001354953,0.00018242103,0.00047059183,0.011079638,0.0007399461,0.0037846258],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9747038,0.007525744,0.002251216,0.00443259,0.009265797,0.0018208324],"domain_scores_gemma":[0.94536006,0.033583045,0.0032669096,0.008111537,0.008476475,0.0012019246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010028486,0.001878893,0.0021406065,0.005793695,0.0028175898,0.0054228473,0.0048436513,0.0024929447,0.0060765715],"category_scores_gemma":[0.06524177,0.001217365,0.002575205,0.00521947,0.001869019,0.01194757,0.005861333,0.0043458138,0.0014685246],"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":[0.000666251,0.0005313721,0.03876958,0.0017199554,0.00052659627,0.001871795,0.0012233825,0.111981764,0.011172602,0.23649277,0.056818943,0.53822505],"study_design_scores_gemma":[0.00007268655,0.000079262194,0.0029103525,0.0002359403,0.00014200316,0.0011650067,0.0011452487,0.6139483,0.0162853,0.31896272,0.04493577,0.00011747353],"about_ca_topic_score_codex":0.008869141,"about_ca_topic_score_gemma":0.011006945,"teacher_disagreement_score":0.010028486,"about_ca_system_score_codex":0.0027524345,"about_ca_system_score_gemma":0.0067490707,"threshold_uncertainty_score":0.053036332},"labels":[],"label_agreement":null},{"id":"W2182859693","doi":"10.14778/3402707.3402711","title":"A Framework for supporting DBMS-like indexes in the cloud","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Caching and Content Delivery","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":true,"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; Cloud computing; Search engine indexing; Distributed computing; Database; Distributed database; Overhead (engineering); Node (physics); Hash table; High availability; Data mining; Hash function; Operating system; Information retrieval","score_opus":0.04723732750376692,"score_gpt":0.26008486779994716,"score_spread":0.21284754029618025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2182859693","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":[0.013793745,0.00080146093,0.9512024,0.00069556077,0.00016925184,0.0007886552,0.0004935205,0.027562588,0.0044929516],"genre_scores_gemma":[0.12081136,0.0005784528,0.87290776,0.00030730758,0.00013085538,0.0003128682,0.001254365,0.0010310689,0.0026660047],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9973053,0.00032716844,0.0004539729,0.00037990336,0.0012135482,0.00032023608],"domain_scores_gemma":[0.9965752,0.00042718512,0.00023764679,0.0013199635,0.00068024214,0.0007597745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004139274,0.00062762876,0.0010503383,0.0014662364,0.0021062645,0.0049823383,0.005092019,0.0013386002,0.001907466],"category_scores_gemma":[0.0055492567,0.0010260048,0.0011523548,0.002244089,0.0015182276,0.0061895982,0.0043523386,0.0023010033,0.0012681796],"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":[0.0010650286,0.001172224,0.010343891,0.0012035461,0.000338554,0.0015033269,0.001925154,0.05099959,0.0799332,0.4141306,0.062458616,0.37492633],"study_design_scores_gemma":[0.0003517469,0.00048817016,0.0027694465,0.00021966232,0.00021279218,0.0014149266,0.0004075618,0.6482491,0.03905315,0.06311833,0.24339989,0.0003151707],"about_ca_topic_score_codex":0.013283753,"about_ca_topic_score_gemma":0.009353796,"teacher_disagreement_score":0.013283753,"about_ca_system_score_codex":0.0018710642,"about_ca_system_score_gemma":0.004535589,"threshold_uncertainty_score":0.026412904},"labels":[],"label_agreement":null},{"id":"W2189052568","doi":"10.14778/2850583.2850586","title":"The iBench integration metadata generator","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Metadata; Computer science; Schema evolution; Data integration; Generality; Schema (genetic algorithms); Generator (circuit theory); Data mapping; Data element; Data science; Information retrieval; Data mining; Database; World Wide Web; Database schema","score_opus":0.2684436286050799,"score_gpt":0.39426327919983467,"score_spread":0.12581965059475475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2189052568","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.045728046,0.0006375628,0.49358281,0.0011393466,0.0006122246,0.0026828924,0.015424799,0.41694412,0.023248132],"genre_scores_gemma":[0.22860622,0.00060116267,0.65722114,0.00093843567,0.00015086783,0.0036947888,0.057286076,0.03996333,0.011537927],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9949244,0.0012729967,0.00068463635,0.0006499757,0.0021388673,0.00032910105],"domain_scores_gemma":[0.98130196,0.006048956,0.0007543368,0.0073435716,0.0037135568,0.00083769136],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010562896,0.0015749226,0.0009364938,0.004237203,0.0012177501,0.0038574976,0.004111745,0.0014121709,0.0104395775],"category_scores_gemma":[0.039223056,0.0015017636,0.0013452267,0.0031702002,0.0012173296,0.0052701156,0.0071255974,0.0030413985,0.0054846425],"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":[0.0035891202,0.0020415368,0.030168256,0.0025393092,0.00045428987,0.0019982865,0.0036230357,0.07137456,0.030587593,0.088142894,0.2579482,0.50753295],"study_design_scores_gemma":[0.0010942215,0.00084449723,0.0059450227,0.0004959655,0.00019337445,0.0009596358,0.0009220625,0.5072443,0.07644674,0.066217706,0.33925876,0.00037776394],"about_ca_topic_score_codex":0.0032056158,"about_ca_topic_score_gemma":0.0022504753,"teacher_disagreement_score":0.010562896,"about_ca_system_score_codex":0.0015392151,"about_ca_system_score_gemma":0.003282546,"threshold_uncertainty_score":0.055862606},"labels":[],"label_agreement":null},{"id":"W2190899134","doi":"10.14778/2850578.2850579","title":"Messing up with BART","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":80,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Scalability; Tuple; Computer science; Benchmarking; Completeness (order theory); Greedy algorithm; Process (computing); Benchmark (surveying); Property (philosophy); Scale (ratio); Control (management); Mathematical optimization; Algorithm; Database; Mathematics; Artificial intelligence; Programming language","score_opus":0.2772987549438379,"score_gpt":0.3926154589624858,"score_spread":0.11531670401864791,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2190899134","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":[0.10082119,0.0019633283,0.8649973,0.0076547395,0.00065462536,0.00023554444,0.0015204378,0.0035747963,0.018578015],"genre_scores_gemma":[0.6928823,0.0015139712,0.28527236,0.0019407201,0.00047627612,0.0002945711,0.002625486,0.001064332,0.01393005],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99037814,0.0027576508,0.0006147687,0.0022158464,0.003154902,0.0008787841],"domain_scores_gemma":[0.93841195,0.027278237,0.0044221687,0.022964537,0.0054696742,0.0014535175],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012157496,0.00074120634,0.0014042972,0.0022441617,0.001964065,0.0052579483,0.0031294744,0.0018330895,0.009954876],"category_scores_gemma":[0.07621719,0.000671392,0.0011584188,0.0033112736,0.0037763442,0.0110402005,0.0058099995,0.002596701,0.0027232803],"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":[0.0008437378,0.00032042578,0.020713445,0.00052668154,0.00024665133,0.0010999211,0.0017863599,0.14541118,0.004094273,0.45317465,0.038467854,0.33331487],"study_design_scores_gemma":[0.00008081607,0.00023095553,0.0029496944,0.00017745547,0.00009290198,0.00079369964,0.00083880575,0.3059918,0.0064091543,0.6264041,0.055941198,0.000089498],"about_ca_topic_score_codex":0.0023797376,"about_ca_topic_score_gemma":0.0014902019,"teacher_disagreement_score":0.012157496,"about_ca_system_score_codex":0.0010142699,"about_ca_system_score_gemma":0.0020297968,"threshold_uncertainty_score":0.06429577},"labels":[],"label_agreement":null},{"id":"W2207847180","doi":"10.14778/2824032.2824049","title":"Fuzzy joins in MapReduce","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Joins; Skyline; Computer science; Point (geometry); Hamming distance; Fuzzy logic; Binary number; Algorithm; Data mining; Theoretical computer science; Artificial intelligence; Mathematics; Programming language; Arithmetic","score_opus":0.03116395072858845,"score_gpt":0.24520602154516505,"score_spread":0.2140420708165766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2207847180","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":[0.5174333,0.0039302367,0.35098675,0.0028312334,0.0015852825,0.0016076011,0.0057292744,0.03777688,0.07811945],"genre_scores_gemma":[0.6643422,0.0007647894,0.3223282,0.0006564548,0.0002267289,0.000395999,0.005470452,0.0006303902,0.0051846844],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.994206,0.0010151591,0.00033481471,0.00097580545,0.0028367958,0.00063154515],"domain_scores_gemma":[0.99545693,0.0016666582,0.00016143073,0.0014803646,0.00083933055,0.0003953603],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036824725,0.0011063463,0.0012332445,0.0011782051,0.0026637893,0.0025517656,0.0032082836,0.0012386262,0.0052903723],"category_scores_gemma":[0.0066602402,0.00062511844,0.0010964867,0.0026587457,0.0010340337,0.005090884,0.0027533195,0.001856405,0.0018188988],"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":[0.007912346,0.004774502,0.0115281725,0.0026981984,0.000791632,0.00080273073,0.0020608762,0.23203468,0.044756543,0.06991165,0.09677851,0.5259502],"study_design_scores_gemma":[0.0011894081,0.0019510981,0.0049622925,0.00011237338,0.00025928026,0.00091596955,0.002116526,0.71854943,0.07701304,0.1252849,0.067470714,0.00017497213],"about_ca_topic_score_codex":0.00771948,"about_ca_topic_score_gemma":0.0073085944,"teacher_disagreement_score":0.00771948,"about_ca_system_score_codex":0.0014110061,"about_ca_system_score_gemma":0.00234579,"threshold_uncertainty_score":0.019475043},"labels":[],"label_agreement":null},{"id":"W2212315060","doi":"10.14778/2856318.2856323","title":"Approximate closest community search in networks","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Caching and Content Delivery","field":"Computer Science","cited_by":217,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Truss; Computer science; Approximation algorithm; Greedy algorithm; Set (abstract data type); Graph; Efficient algorithm; Theoretical computer science; Mathematical optimization; Mathematics; Combinatorics; Algorithm","score_opus":0.05422769107867213,"score_gpt":0.2469106196643252,"score_spread":0.19268292858565306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2212315060","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":[0.052396487,0.0025556313,0.93577194,0.0012243877,0.00010130435,0.00022492914,0.00060126284,0.00075318443,0.006370844],"genre_scores_gemma":[0.49016175,0.0015618056,0.49685472,0.00042135228,0.00016558394,0.00048164703,0.0020074092,0.00030896306,0.008036776],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99782664,0.0007739355,0.0000871337,0.00063404447,0.00044574667,0.00023242013],"domain_scores_gemma":[0.9927394,0.0050454033,0.0007005463,0.0005540361,0.00063207414,0.00032848289],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002277179,0.0016327337,0.0026317853,0.0026515936,0.0015480991,0.0020385552,0.003694601,0.0030069863,0.004041819],"category_scores_gemma":[0.015153441,0.0010093561,0.001387598,0.0035574601,0.0015288709,0.005133215,0.0030981896,0.002020596,0.0008805576],"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":[0.0002074515,0.000102970545,0.0012438898,0.00026618535,0.000104428036,0.00015276251,0.00020559444,0.90994173,0.00097491243,0.03853371,0.005671347,0.042595],"study_design_scores_gemma":[0.000022473321,0.0000145579625,0.00008391026,0.000014892227,0.000008394672,0.00003578314,0.000037301696,0.9634996,0.00015659137,0.035263166,0.0008574291,0.0000059321133],"about_ca_topic_score_codex":0.009859422,"about_ca_topic_score_gemma":0.012961104,"teacher_disagreement_score":0.009859422,"about_ca_system_score_codex":0.002900251,"about_ca_system_score_gemma":0.0016376948,"threshold_uncertainty_score":0.021042943},"labels":[],"label_agreement":null},{"id":"W2243935923","doi":"10.14778/2757807.2757810","title":"Compaction management in distributed key-value datastores","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":61,"is_retracted":false,"has_abstract":true,"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; Workload; Scalability; Server; Compaction; Distributed computing; Operating system","score_opus":0.02208657302277573,"score_gpt":0.2408768714961972,"score_spread":0.21879029847342146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2243935923","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":[0.46514687,0.0031408155,0.49577224,0.001242882,0.00031751892,0.0005202924,0.00062194164,0.027230881,0.0060065663],"genre_scores_gemma":[0.9210742,0.00037813044,0.07373207,0.00019497408,0.00008632644,0.00016503502,0.00069746096,0.0004869808,0.0031847593],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.996899,0.00048212052,0.00039248486,0.0007370308,0.0010561497,0.00043318304],"domain_scores_gemma":[0.9897363,0.0020476913,0.00090471405,0.005235344,0.0014804906,0.000595492],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030596256,0.000688298,0.00094725844,0.0012363801,0.0013925146,0.0025871915,0.0036742787,0.0008828351,0.0014271826],"category_scores_gemma":[0.009219704,0.00074322533,0.000360506,0.002264908,0.0011411378,0.0065248357,0.0032413309,0.00092343305,0.0006045567],"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":[0.0030744164,0.0010072478,0.039662622,0.00080037204,0.00033237183,0.0022042529,0.004029906,0.16942935,0.118895225,0.059092987,0.03801062,0.56346065],"study_design_scores_gemma":[0.00034001702,0.00044793347,0.0075491127,0.000084522224,0.00016361327,0.001016507,0.0012248034,0.78245705,0.1391082,0.033562448,0.033892464,0.00015332858],"about_ca_topic_score_codex":0.0024876334,"about_ca_topic_score_gemma":0.0024024325,"teacher_disagreement_score":0.0036742787,"about_ca_system_score_codex":0.0015733952,"about_ca_system_score_gemma":0.0015526674,"threshold_uncertainty_score":0.016181052},"labels":[],"label_agreement":null},{"id":"W2244188037","doi":"10.14778/2824032.2824046","title":"A scalable distributed graph partitioner","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Interconnection Networks and Systems","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Ministry of Economy, Trade and Industry","keywords":"Graph partition; Scalability; Partition (number theory); Computer science; Bounded function; Graph; Parallel computing; Space partitioning; Theoretical computer science; Algorithm; Combinatorics; Mathematics; Database","score_opus":0.024293433423822605,"score_gpt":0.2216525087113379,"score_spread":0.19735907528751528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2244188037","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":[0.021062061,0.0005443796,0.942062,0.00026957225,0.00011001547,0.00012826454,0.0005144452,0.028438983,0.00687032],"genre_scores_gemma":[0.15373966,0.00034310357,0.83229715,0.00020004228,0.00005629154,0.00024104175,0.0023897626,0.0023728495,0.008360156],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995303,0.000064492924,0.000024712986,0.0001384984,0.00019110208,0.000050887633],"domain_scores_gemma":[0.99935097,0.00014989819,0.000035989837,0.0002646371,0.00013036079,0.000068176],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005216477,0.00074922544,0.0007121341,0.00072664855,0.0006690453,0.000948431,0.0020131078,0.00074136176,0.010054822],"category_scores_gemma":[0.0016442382,0.0004745163,0.00052883587,0.00094250997,0.0005141895,0.0023926562,0.0022916833,0.0010353304,0.0032297538],"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":[0.00080144114,0.00026046715,0.0014373468,0.00049398525,0.00012639625,0.00031297668,0.00042024098,0.15661241,0.07121822,0.035201803,0.078862295,0.65425235],"study_design_scores_gemma":[0.0002543894,0.00015621413,0.0006796675,0.000029152638,0.000034444412,0.00036035816,0.00013315374,0.8902818,0.024454825,0.029762331,0.053814743,0.000038922615],"about_ca_topic_score_codex":0.0019576102,"about_ca_topic_score_gemma":0.0051076533,"teacher_disagreement_score":0.010054822,"about_ca_system_score_codex":0.00067630416,"about_ca_system_score_gemma":0.00096140517,"threshold_uncertainty_score":0.03363675},"labels":[],"label_agreement":null},{"id":"W2244331736","doi":"10.14778/2536222.2536255","title":"Toward scalable transaction processing","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Scalability; Transaction processing system; Computer science; Transaction processing; Multi-core processor; Online transaction processing; Database transaction; Metadata; Distributed transaction; Distributed computing; Computer architecture; Embedded system; Parallel computing; Database; Operating system","score_opus":0.01483689700489075,"score_gpt":0.2104329737351254,"score_spread":0.19559607673023466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2244331736","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":[0.01074348,0.0036297583,0.9588871,0.00521125,0.0005586085,0.0002977423,0.00021501975,0.0030005511,0.017456502],"genre_scores_gemma":[0.12737776,0.006498398,0.8501528,0.0016642363,0.0009137875,0.00067965814,0.0008535467,0.0006350478,0.011224745],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99436486,0.0012025201,0.00046591825,0.00066755206,0.0029231647,0.00037599722],"domain_scores_gemma":[0.9927188,0.0018639703,0.00035052473,0.002506778,0.0021598826,0.00040007746],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049786353,0.0009217291,0.00094027305,0.0009550798,0.0011199004,0.0038397487,0.0030370827,0.0016072791,0.0062481333],"category_scores_gemma":[0.012218521,0.0011501263,0.00093935785,0.0020351675,0.0019548296,0.008660661,0.0052340277,0.005538691,0.0040599005],"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":[0.00031328402,0.00026079928,0.0013828956,0.0012893551,0.00013486043,0.00041824876,0.0010695497,0.047050286,0.044034313,0.58186036,0.047488455,0.27469754],"study_design_scores_gemma":[0.00013505053,0.00023585443,0.00037981407,0.00027781364,0.00006550059,0.00041153003,0.00048488355,0.34961578,0.01956812,0.47206983,0.15669373,0.00006217122],"about_ca_topic_score_codex":0.0011070521,"about_ca_topic_score_gemma":0.0011121691,"teacher_disagreement_score":0.0062481333,"about_ca_system_score_codex":0.0014403234,"about_ca_system_score_gemma":0.0023309106,"threshold_uncertainty_score":0.026329875},"labels":[],"label_agreement":null},{"id":"W2256219868","doi":"10.14778/2752939.2752949","title":"Understanding the causes of consistency anomalies in Apache Cassandra","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Garbage collection; Computer science; Consistency (knowledge bases); Garbage; Workload; Java; Server; Throughput; Operating system; Database; Real-time computing; Programming language; Artificial intelligence","score_opus":0.13004382623379485,"score_gpt":0.2601769927376143,"score_spread":0.13013316650381945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2256219868","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":[0.97701955,0.0005981899,0.018246422,0.00059262244,0.000051718365,0.00003643891,0.00020649856,0.0021802587,0.0010682081],"genre_scores_gemma":[0.99691963,0.00009563264,0.0025220704,0.00004835674,0.000017034781,0.000019386009,0.00013700247,0.00009926598,0.00014168891],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99557245,0.00079883606,0.00031505534,0.0008238666,0.001768785,0.0007209003],"domain_scores_gemma":[0.98217106,0.008768632,0.0030117938,0.002994998,0.002416972,0.0006366516],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003120679,0.00057832483,0.0008518861,0.0014447641,0.0011076487,0.0019388425,0.0015576605,0.0012909109,0.00035887584],"category_scores_gemma":[0.021482365,0.0007282215,0.00038332195,0.0024441422,0.0015850918,0.002230857,0.0011724065,0.0025761237,0.00015975708],"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":[0.0028431236,0.0009391641,0.544797,0.00075652276,0.0002968873,0.004759627,0.0061222557,0.18767391,0.12668999,0.022420535,0.0071520545,0.09554883],"study_design_scores_gemma":[0.000117844866,0.0005036405,0.15753701,0.000103253675,0.00017409633,0.0018447526,0.002043379,0.69874877,0.10643466,0.026910415,0.005394138,0.00018810124],"about_ca_topic_score_codex":0.004329236,"about_ca_topic_score_gemma":0.003456367,"teacher_disagreement_score":0.004329236,"about_ca_system_score_codex":0.0014100557,"about_ca_system_score_gemma":0.0022747945,"threshold_uncertainty_score":0.01650393},"labels":[],"label_agreement":null},{"id":"W2260484439","doi":"10.14778/2824032.2824036","title":"SEMA-JOIN","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":true,"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; Joins; Join (topology); Hash join; Table (database); Theoretical computer science; Data mining; Database; Programming language; Mathematics","score_opus":0.2932342444414948,"score_gpt":0.4016846612147633,"score_spread":0.1084504167732685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2260484439","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":[0.005088016,0.00035264046,0.93887156,0.00037085518,0.00023529642,0.0004191787,0.005992004,0.039714493,0.008955968],"genre_scores_gemma":[0.062338136,0.00028284345,0.9093468,0.00050810835,0.00020219403,0.00059679704,0.016667908,0.0038467862,0.006210474],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9920025,0.0014772835,0.0008342109,0.0017985193,0.0035681548,0.00031939056],"domain_scores_gemma":[0.99186194,0.0029959015,0.00057850505,0.0031380313,0.001070841,0.00035469697],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0067278533,0.0011610825,0.0013388679,0.0025429972,0.0017648822,0.0044462197,0.0038195197,0.0012198699,0.01859626],"category_scores_gemma":[0.013819616,0.0009098294,0.0024249605,0.0034201012,0.0013175688,0.005640713,0.0058515277,0.0021240392,0.008025618],"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":[0.001367992,0.00055523694,0.010361987,0.001324896,0.0005683992,0.00053224474,0.0012116589,0.025526438,0.016205773,0.20491359,0.19062479,0.546807],"study_design_scores_gemma":[0.0002060314,0.0003583563,0.0025434918,0.0001563545,0.00011322205,0.001222432,0.00051337853,0.31320184,0.027596142,0.23550934,0.41843736,0.00014200983],"about_ca_topic_score_codex":0.0016015635,"about_ca_topic_score_gemma":0.0033064082,"teacher_disagreement_score":0.01859626,"about_ca_system_score_codex":0.0007420597,"about_ca_system_score_gemma":0.0021019091,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2262592273","doi":"10.14778/2824032.2824109","title":"KATARA","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":45,"is_retracted":false,"has_abstract":true,"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; Table (database); Crowdsourcing; Ambiguity; Tuple; Information retrieval; Annotation; Task (project management); Semantics (computer science); Reliability (semiconductor); Data mining; World Wide Web; Programming language; Artificial intelligence","score_opus":0.34404765762554373,"score_gpt":0.4117242000185209,"score_spread":0.06767654239297716,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2262592273","genre_codex":"other","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":[0.022866318,0.007839766,0.3081153,0.01293324,0.008711325,0.0012320622,0.05623797,0.11794175,0.4641222],"genre_scores_gemma":[0.14118034,0.0056861546,0.26111183,0.006753407,0.0019579544,0.0014884114,0.116329536,0.020051843,0.4454404],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99565244,0.00066803535,0.00044139792,0.0014613223,0.0013850472,0.00039174632],"domain_scores_gemma":[0.99126774,0.0012561235,0.00053823733,0.00333904,0.0027181683,0.0008806323],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035849188,0.0013855334,0.001360952,0.0033498728,0.0021070254,0.0063824006,0.0028213796,0.002075268,0.19347334],"category_scores_gemma":[0.011869022,0.0010703319,0.001377244,0.0027884343,0.0009913269,0.0063070743,0.007136745,0.0024399213,0.20779921],"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":[0.0011540338,0.00019144836,0.007184934,0.0018745416,0.00019847415,0.0007565383,0.0011127007,0.0020059922,0.0126015395,0.043597545,0.4071482,0.52217406],"study_design_scores_gemma":[0.000052801945,0.00007480574,0.0017358921,0.00025788165,0.000058333106,0.0006711484,0.00028415068,0.0032081008,0.0066335304,0.0146901095,0.972272,0.00006125144],"about_ca_topic_score_codex":0.002117492,"about_ca_topic_score_gemma":0.002176022,"teacher_disagreement_score":0.19347334,"about_ca_system_score_codex":0.0013823997,"about_ca_system_score_gemma":0.0033568384,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2262876226","doi":"10.14778/2824032.2824069","title":"Towards scalable real-time analytics","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Systems, Applications & Products in Data Processing (Canada)","funders":"","keywords":"Computer science; Analytics; Scalability; Timestamp; Online transaction processing; Database; Asynchronous communication; Online analytical processing; Distributed computing; Distributed transaction; Distributed database; Snapshot (computer storage); Transaction processing; Real-time computing; Database transaction; Data warehouse; Computer network","score_opus":0.021671129085309535,"score_gpt":0.2320368607474169,"score_spread":0.21036573166210737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2262876226","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":[0.014354384,0.0039095394,0.96037287,0.0043652616,0.00041825615,0.00012569045,0.00031175275,0.0058251135,0.010317193],"genre_scores_gemma":[0.358663,0.005757573,0.62394667,0.0014074465,0.0010343542,0.00026661024,0.0016843994,0.0007086209,0.0065312944],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970925,0.00041897362,0.0001971167,0.0005353575,0.0014564118,0.00029962746],"domain_scores_gemma":[0.9946844,0.001486656,0.00039759793,0.0017147645,0.0013038096,0.00041271804],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036292765,0.0011938802,0.00092173344,0.000945251,0.0005740843,0.0051565813,0.0028880786,0.0012585638,0.0032048712],"category_scores_gemma":[0.008388493,0.0007714961,0.0005387551,0.0017566109,0.0014397167,0.008499421,0.004461066,0.0038497674,0.0016765775],"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":[0.0007206349,0.0003453292,0.0028864443,0.0011858088,0.00021571708,0.0007390406,0.00096956897,0.14193012,0.07391977,0.353767,0.05051563,0.37280485],"study_design_scores_gemma":[0.00008101907,0.00013088196,0.0004845024,0.00010395558,0.00005353553,0.00019608272,0.0002638678,0.71644616,0.014282072,0.19544797,0.07246138,0.000048654463],"about_ca_topic_score_codex":0.0015486194,"about_ca_topic_score_gemma":0.0013678219,"teacher_disagreement_score":0.0051565813,"about_ca_system_score_codex":0.0011488999,"about_ca_system_score_gemma":0.0018067067,"threshold_uncertainty_score":0.019193709},"labels":[],"label_agreement":null},{"id":"W2263157912","doi":"10.14778/2777598.2777604","title":"Giraph unchained","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":145,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Asynchronous communication; Computer science; Scalability; Computation; Synchronization (alternating current); Distributed computing; Bulk synchronous parallel; Graph; Parallel computing; Model of computation; Theoretical computer science; Algorithm; Computer network; Operating system","score_opus":0.01784287209176177,"score_gpt":0.2045423580511298,"score_spread":0.18669948595936803,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2263157912","genre_codex":"software","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":[0.04708662,0.00091789087,0.3444376,0.0008246782,0.0008671346,0.0004965434,0.0068780524,0.44795376,0.15053771],"genre_scores_gemma":[0.43965068,0.00085816375,0.37078044,0.0015073278,0.00025396937,0.00095232116,0.02697431,0.03543712,0.12358559],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99887997,0.00019830267,0.000052036034,0.0003854765,0.00033268283,0.00015152845],"domain_scores_gemma":[0.9975011,0.00032536217,0.00006185065,0.0015622792,0.00039549643,0.00015389129],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008898036,0.00094986893,0.00051145995,0.00088988536,0.00096483115,0.0016218023,0.0025272884,0.0008274679,0.03614961],"category_scores_gemma":[0.0037110504,0.0007601601,0.00069899904,0.0008320847,0.000949964,0.0028464578,0.003092166,0.0014490631,0.017511938],"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":[0.0028026656,0.0004121504,0.004937518,0.0011774072,0.00020764841,0.00044040655,0.0008129368,0.03496903,0.041927263,0.11804061,0.42658266,0.36768973],"study_design_scores_gemma":[0.0005006552,0.0003870334,0.0028843062,0.00013463398,0.00012406388,0.00066246575,0.00022278022,0.21425843,0.061768368,0.1023443,0.6165552,0.00015773065],"about_ca_topic_score_codex":0.003738397,"about_ca_topic_score_gemma":0.0042413417,"teacher_disagreement_score":0.03614961,"about_ca_system_score_codex":0.00079700974,"about_ca_system_score_gemma":0.0014063503,"threshold_uncertainty_score":0.12093252},"labels":[],"label_agreement":null},{"id":"W2264475115","doi":"10.14778/2809974.2809977","title":"Worker skill estimation in team-based tasks","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Mobile Crowdsensing and Crowdsourcing","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Nautical Research Society","funders":"Army Research Office; National Science Foundation","keywords":"Task (project management); Computer science; Scalability; Outcome (game theory); Estimation; Machine learning; Artificial intelligence; Knowledge management; Mathematics; Engineering","score_opus":0.015130512596530404,"score_gpt":0.23288734720550677,"score_spread":0.21775683460897638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2264475115","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":[0.076206185,0.00044829195,0.91976804,0.00044537787,0.00004656696,0.00013889285,0.00031000923,0.0005856152,0.0020511555],"genre_scores_gemma":[0.8068059,0.00027741754,0.18942317,0.00013661874,0.000093011855,0.00023374702,0.00061468873,0.00013299049,0.0022823703],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977551,0.00080249924,0.00013512555,0.0007224935,0.00034099873,0.00024383586],"domain_scores_gemma":[0.99402255,0.003569525,0.0007216864,0.0006939709,0.00063653477,0.00035575827],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035532704,0.0014200074,0.001682784,0.0017063664,0.0009606063,0.0014170315,0.0023432996,0.0019509409,0.0017064706],"category_scores_gemma":[0.015688267,0.00083530316,0.00078637974,0.001400205,0.0011100036,0.0018676327,0.002417013,0.0014258499,0.00076827826],"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":[0.00049971836,0.0002800168,0.015229289,0.000300164,0.00014647309,0.00017851932,0.0008651408,0.791594,0.003218078,0.010764481,0.0036383169,0.1732858],"study_design_scores_gemma":[0.000028058657,0.0000450839,0.002493762,0.000028881006,0.000015147515,0.00004021207,0.00017279953,0.9681814,0.0011211872,0.026830953,0.0010214437,0.000021139731],"about_ca_topic_score_codex":0.012485125,"about_ca_topic_score_gemma":0.009050555,"teacher_disagreement_score":0.012485125,"about_ca_system_score_codex":0.0014086297,"about_ca_system_score_gemma":0.0017572657,"threshold_uncertainty_score":0.024824917},"labels":[],"label_agreement":null},{"id":"W2266577329","doi":"10.14778/2904483.2904488","title":"A general-purpose query-centric framework for querying big graphs","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":45,"is_retracted":false,"has_abstract":true,"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; Graph database; Workload; Theoretical computer science; Graph; Analytics; Big data; Computation; Vertex (graph theory); Database; Data mining; Programming language; Operating system","score_opus":0.016580142920986043,"score_gpt":0.2329751982869722,"score_spread":0.21639505536598613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2266577329","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":[0.006401502,0.00045871534,0.88896793,0.0005113642,0.00008475239,0.00035116094,0.0016962251,0.09827921,0.003249144],"genre_scores_gemma":[0.1407674,0.0007874059,0.8309201,0.00091189495,0.00015322535,0.00072676054,0.010150494,0.010533466,0.0050491896],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99656445,0.0005669067,0.00031912993,0.00097543566,0.00123533,0.00033887968],"domain_scores_gemma":[0.9947983,0.0013616997,0.00021190384,0.002596169,0.0006905089,0.00034134378],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004428691,0.0022860952,0.0017558051,0.002367698,0.0019909893,0.0045996215,0.0075231385,0.0018363659,0.005400068],"category_scores_gemma":[0.007963038,0.0016132048,0.002415364,0.0037465577,0.0030065374,0.010862462,0.0066053923,0.004172469,0.0029069092],"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":[0.0028872208,0.00083619385,0.009019652,0.0023472647,0.00061973464,0.00072132243,0.002417513,0.1019322,0.07424613,0.30107218,0.21728759,0.28661296],"study_design_scores_gemma":[0.0003976579,0.00026148933,0.0014925881,0.000093430725,0.00013276435,0.0005719675,0.00045156173,0.7000275,0.035902083,0.16118033,0.09926668,0.0002219355],"about_ca_topic_score_codex":0.015236175,"about_ca_topic_score_gemma":0.022148624,"teacher_disagreement_score":0.015236175,"about_ca_system_score_codex":0.0022713318,"about_ca_system_score_gemma":0.0036024766,"threshold_uncertainty_score":0.030294955},"labels":[],"label_agreement":null},{"id":"W2268506948","doi":"10.14778/2824032.2824128","title":"S+EPPs","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"SPARQL; Computer science; Component (thermodynamics); Graph; Theoretical computer science; Information retrieval; RDF; Semantic Web","score_opus":0.0193053903847535,"score_gpt":0.20903259207919872,"score_spread":0.18972720169444524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2268506948","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":[0.020296723,0.00023203192,0.73804647,0.0003733845,0.00015499287,0.00022471722,0.0046644607,0.22388585,0.012121356],"genre_scores_gemma":[0.27544,0.00049708626,0.65708643,0.0010235597,0.00012916105,0.00045837078,0.024941208,0.021531602,0.01889249],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989613,0.00015456697,0.00009776049,0.00030602308,0.0003927338,0.0000875641],"domain_scores_gemma":[0.9983456,0.00040466592,0.00009560827,0.0007441249,0.00034461168,0.0000653709],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00091419194,0.0010619411,0.0005365159,0.00067635864,0.000402387,0.0016874631,0.0016690579,0.000469603,0.014513643],"category_scores_gemma":[0.0038153096,0.00054689514,0.0008056494,0.00083944754,0.00048382,0.003011367,0.0022284247,0.0010186844,0.0059220484],"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":[0.0014136215,0.0002811062,0.0051182457,0.0011404502,0.00019000853,0.000667872,0.00074661605,0.04412558,0.055131763,0.07219938,0.18488823,0.6340971],"study_design_scores_gemma":[0.0002304385,0.00039251093,0.0019899588,0.00007918023,0.000115226554,0.0007429475,0.00025014198,0.52776664,0.1277236,0.06760506,0.27299163,0.00011272513],"about_ca_topic_score_codex":0.0033342803,"about_ca_topic_score_gemma":0.0033930615,"teacher_disagreement_score":0.014513643,"about_ca_system_score_codex":0.0005704024,"about_ca_system_score_gemma":0.0010022006,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2281494333","doi":"10.14778/2732977.2732980","title":"An experimental comparison of pregel-like graph processing systems","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":194,"is_retracted":false,"has_abstract":true,"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; PageRank; Graph; Global Positioning System; Theoretical computer science; Operating system","score_opus":0.014468431683686635,"score_gpt":0.2619565896032245,"score_spread":0.24748815791953785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2281494333","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":[0.9606584,0.0009395862,0.0132800685,0.00062268315,0.0004130526,0.000385699,0.0020666786,0.008253878,0.013379931],"genre_scores_gemma":[0.94050515,0.00048303354,0.044412587,0.00030968175,0.000100564066,0.00033246473,0.0084000835,0.00085933786,0.004597082],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996111,0.000976745,0.00041852507,0.0008782812,0.0011624239,0.000453091],"domain_scores_gemma":[0.98111606,0.009237604,0.00067862537,0.0045189173,0.0035377599,0.0009110452],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029423316,0.0009901201,0.0007837994,0.0013705387,0.001074618,0.0013048565,0.0022577024,0.0010411057,0.005131413],"category_scores_gemma":[0.015143482,0.00045617323,0.00043291735,0.0022552568,0.0011654389,0.003750167,0.0014431394,0.0013961105,0.0015987145],"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":[0.018455027,0.012623264,0.025455661,0.005010725,0.0011914842,0.0008346339,0.002395611,0.27142504,0.14001103,0.024309976,0.12194855,0.376339],"study_design_scores_gemma":[0.002178128,0.016217288,0.042625155,0.00019299217,0.00038993446,0.00079007033,0.0030433636,0.6974812,0.17314315,0.018859893,0.044813756,0.00026504503],"about_ca_topic_score_codex":0.0037232826,"about_ca_topic_score_gemma":0.004334174,"teacher_disagreement_score":0.005131413,"about_ca_system_score_codex":0.0011492502,"about_ca_system_score_gemma":0.001108922,"threshold_uncertainty_score":0.017166317},"labels":[],"label_agreement":null},{"id":"W2291620117","doi":"10.14778/2850469.2850471","title":"K-core decomposition of large networks on a single PC","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":214,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Computer science; Decomposition; Implementation; Core (optical fiber); Metric (unit); Graph; Multi-core processor; Vertex (graph theory); Theoretical computer science; Parallel computing; Algorithm","score_opus":0.03872705365176489,"score_gpt":0.26743815820515754,"score_spread":0.22871110455339266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2291620117","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":[0.1806637,0.00033581257,0.8000292,0.0005712329,0.000084208696,0.0001789682,0.00044028234,0.006468044,0.011228528],"genre_scores_gemma":[0.5085057,0.00020456203,0.48438832,0.00017682901,0.000026034219,0.00021351388,0.0010680892,0.00064524816,0.0047717323],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99927574,0.00013944187,0.000044792516,0.00019474598,0.00020803584,0.00013733293],"domain_scores_gemma":[0.9969958,0.00091551,0.00019718893,0.0013015564,0.00042691306,0.00016308797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010449333,0.0008093466,0.00068808097,0.0009295138,0.0010042824,0.001489421,0.0016227012,0.0007513224,0.004907934],"category_scores_gemma":[0.006103801,0.00049839925,0.000692374,0.0014942645,0.000873805,0.0035087497,0.0017585318,0.0012219249,0.001373386],"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":[0.00090095046,0.0003024004,0.0057541626,0.0003796038,0.00014566106,0.0003656886,0.0007113432,0.61212045,0.018326862,0.065403365,0.017558277,0.27803117],"study_design_scores_gemma":[0.00003876512,0.000055714714,0.00058392005,0.000014010639,0.000016103553,0.0000623538,0.00015949622,0.9538427,0.004528257,0.037234314,0.0034547716,0.000009485914],"about_ca_topic_score_codex":0.006659556,"about_ca_topic_score_gemma":0.01219268,"teacher_disagreement_score":0.006659556,"about_ca_system_score_codex":0.0015576977,"about_ca_system_score_gemma":0.0017229051,"threshold_uncertainty_score":0.016418695},"labels":[],"label_agreement":null},{"id":"W2292011317","doi":"10.14778/2735461.2735463","title":"Top-k nearest neighbor search in uncertain data series","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Series (stratigraphy); Computer science; Nearest neighbor search; k-nearest neighbors algorithm; Metric (unit); Data mining; Independence (probability theory); Time series; Uncertain data; Variety (cybernetics); Synthetic data; Algorithm; Artificial intelligence; Machine learning; Mathematics; Statistics","score_opus":0.03384988583387695,"score_gpt":0.25089819564159177,"score_spread":0.21704830980771483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2292011317","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":[0.059446324,0.0015667938,0.93710816,0.00027630688,0.000074712334,0.000049863927,0.00027532605,0.00033043235,0.0008720491],"genre_scores_gemma":[0.6791968,0.0007694945,0.3178506,0.000114531664,0.00012694922,0.000074040494,0.0008385943,0.00008494298,0.00094407616],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980774,0.0005935715,0.00020582517,0.00058650546,0.00041421675,0.00012247697],"domain_scores_gemma":[0.9925638,0.005327981,0.0007242724,0.00060625526,0.0006383634,0.00013939256],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034012415,0.0008382721,0.0023743808,0.001970493,0.0010540847,0.0016219938,0.0022548318,0.0015539718,0.00076543086],"category_scores_gemma":[0.014561188,0.0005779832,0.001000996,0.0031856573,0.00094424386,0.0031769217,0.0011608509,0.0013829304,0.0002973064],"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":[0.00016401956,0.00005806212,0.0032744135,0.00018281343,0.000116401265,0.00018469182,0.00013647781,0.9117641,0.0009636269,0.007626716,0.0012948469,0.07423391],"study_design_scores_gemma":[0.0000042683055,0.000014638569,0.00030507141,0.000008222487,0.000008062273,0.00003716332,0.000044117096,0.9878808,0.0004174547,0.010968127,0.00030305,0.000008960709],"about_ca_topic_score_codex":0.007063354,"about_ca_topic_score_gemma":0.00723175,"teacher_disagreement_score":0.007063354,"about_ca_system_score_codex":0.0010406233,"about_ca_system_score_gemma":0.0008605332,"threshold_uncertainty_score":0.017987669},"labels":[],"label_agreement":null},{"id":"W2293393493","doi":"10.14778/2732967.2732970","title":"ConfluxDB","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"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; Replication (statistics); Distributed computing; Snapshot (computer storage); Database transaction; Database; Transaction processing; Operating system; Parallel computing","score_opus":0.005146339465848849,"score_gpt":0.18617502323385682,"score_spread":0.18102868376800796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2293393493","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":[0.013962521,0.0034624503,0.118593656,0.002958352,0.0009177927,0.00094767386,0.11049612,0.5535797,0.19508186],"genre_scores_gemma":[0.11617663,0.0028219211,0.14264481,0.0041252775,0.0005038434,0.0012181652,0.58043796,0.04665439,0.10541704],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969325,0.0003270889,0.0003190512,0.0006126323,0.0014580959,0.00035054155],"domain_scores_gemma":[0.9955432,0.00052925077,0.00018137734,0.0021096629,0.001328623,0.00030792534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026861338,0.0015210327,0.0012942151,0.0027978015,0.0015945293,0.0065422733,0.0072048507,0.0016636519,0.071781754],"category_scores_gemma":[0.0070629125,0.0010717262,0.00096057437,0.0037711724,0.0008520295,0.0072410135,0.0052771773,0.002050121,0.060867082],"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":[0.0006948901,0.00017525803,0.0011563308,0.0006359159,0.00009148612,0.00021776983,0.00033852432,0.0015709557,0.0048637893,0.0197599,0.86768055,0.102814645],"study_design_scores_gemma":[0.00029419735,0.00009797487,0.0009553619,0.000079881145,0.000043412034,0.00033117965,0.00021340842,0.013001007,0.008180508,0.012422625,0.9643027,0.00007777068],"about_ca_topic_score_codex":0.01643923,"about_ca_topic_score_gemma":0.009976102,"teacher_disagreement_score":0.071781754,"about_ca_system_score_codex":0.0023324501,"about_ca_system_score_gemma":0.0027378805,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2293703278","doi":"10.14778/3402707.3402744","title":"Publishing set-valued data via differential privacy","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":221,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Differential privacy; Computer science; Data publishing; Data mining; Scalability; Context (archaeology); Data anonymization; Set (abstract data type); Information privacy; Information retrieval; Theoretical computer science; Publishing; Database; Computer security","score_opus":0.09995646936477759,"score_gpt":0.2744083175834562,"score_spread":0.1744518482186786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2293703278","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":[0.021435797,0.0003557164,0.97402936,0.0015716893,0.000056086345,0.00012860914,0.00031797265,0.00024769382,0.0018571714],"genre_scores_gemma":[0.70133454,0.0010713299,0.29113957,0.001005172,0.00030620833,0.0005074717,0.00082765747,0.00013652825,0.0036713965],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97657055,0.009680839,0.0018877108,0.0034954406,0.0074441787,0.00092139374],"domain_scores_gemma":[0.9385475,0.029546097,0.0034741259,0.025321562,0.0022627772,0.0008479986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016359415,0.0007787725,0.0024532392,0.0014073276,0.0018446575,0.0061037345,0.0038861115,0.0026190272,0.0015361346],"category_scores_gemma":[0.05496402,0.0009338431,0.0018740678,0.0049478593,0.004266058,0.015480244,0.0070584463,0.004732266,0.0008515019],"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":[0.00097029994,0.00029144177,0.0054256297,0.0004696366,0.00027310345,0.00094605755,0.0016671778,0.092008196,0.01315459,0.673807,0.0047446117,0.2062422],"study_design_scores_gemma":[0.000105850966,0.00026025967,0.000715318,0.000057507597,0.00008462782,0.0013930651,0.00030657544,0.30332613,0.016330589,0.66778487,0.009573798,0.00006153259],"about_ca_topic_score_codex":0.0003655682,"about_ca_topic_score_gemma":0.00025034233,"teacher_disagreement_score":0.016359415,"about_ca_system_score_codex":0.0018400258,"about_ca_system_score_gemma":0.0021646565,"threshold_uncertainty_score":0.08651793},"labels":[],"label_agreement":null},{"id":"W2294111665","doi":"10.14778/2732951.2732960","title":"Scalable logging through emerging non-volatile memory","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Memory and Neural Computing","field":"Engineering","cited_by":206,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Commit; Scalability; Computer science; Logging; Dram; Bottleneck; Embedded system; Overhead (engineering); Cache; Operating system; Computer hardware; Database; Forestry","score_opus":0.009201764651845435,"score_gpt":0.2140710501752543,"score_spread":0.20486928552340886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2294111665","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":[0.5198535,0.0050480673,0.44044986,0.0019972033,0.00057197915,0.00022516436,0.0004276515,0.016028987,0.015397585],"genre_scores_gemma":[0.94155383,0.0006249355,0.05297644,0.00021797718,0.000053597993,0.00010566846,0.00030315708,0.00017119964,0.003993199],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993467,0.000107697095,0.000048371254,0.000103693244,0.00028760367,0.00010597929],"domain_scores_gemma":[0.99767643,0.000545478,0.00017781941,0.0009095019,0.0005612258,0.00012949573],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00077840156,0.00032273514,0.00032835346,0.0005071205,0.00061163114,0.0015290406,0.0022466134,0.0003952077,0.0019940047],"category_scores_gemma":[0.0030227252,0.0002968167,0.00016747795,0.0006430881,0.00067105744,0.0036199703,0.002077036,0.0010127357,0.00047412812],"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":[0.0017291191,0.00064325583,0.018763455,0.00086447014,0.00012634716,0.0013288298,0.0015002874,0.11142679,0.13391756,0.07285539,0.034689397,0.6221551],"study_design_scores_gemma":[0.0002057588,0.0005803013,0.0025546066,0.00010762996,0.00007736764,0.0006991707,0.0009901886,0.778019,0.12484772,0.06097034,0.030863544,0.00008431304],"about_ca_topic_score_codex":0.0014023307,"about_ca_topic_score_gemma":0.0027368704,"teacher_disagreement_score":0.0022466134,"about_ca_system_score_codex":0.0005312896,"about_ca_system_score_gemma":0.001079503,"threshold_uncertainty_score":0.006670654},"labels":[],"label_agreement":null},{"id":"W2295333720","doi":"10.14778/2856318.2856331","title":"CLAMShell","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Mobile Crowdsensing and Crowdsourcing","field":"Computer Science","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Crowds; Computer science; Latency (audio); Speedup; Data science; Computer security; Operating system; Telecommunications","score_opus":0.020412257649798923,"score_gpt":0.20830406149783764,"score_spread":0.18789180384803872,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2295333720","genre_codex":"software","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020482076,0.0013559955,0.42277756,0.0026835823,0.001334631,0.0015311349,0.016611908,0.46677303,0.066450074],"genre_scores_gemma":[0.21390882,0.00094752474,0.59628856,0.0033891092,0.0005609173,0.0019661968,0.045072697,0.04183712,0.09602903],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99670726,0.0004724255,0.00018781345,0.0012301131,0.00115412,0.00024826673],"domain_scores_gemma":[0.9921828,0.0019338009,0.00029728274,0.0040969807,0.0009854393,0.0005036591],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030062746,0.0015223454,0.0012251262,0.0019900335,0.0020568087,0.0026487114,0.0033303213,0.0017795966,0.049664002],"category_scores_gemma":[0.013342664,0.0013111333,0.0011937489,0.0015453544,0.0011909402,0.004849966,0.0075019877,0.0021170308,0.03711115],"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":[0.0017346231,0.00028203978,0.0035047336,0.00094135274,0.00021407839,0.00043392554,0.0013848514,0.0065313615,0.020952918,0.020503843,0.5526616,0.39085466],"study_design_scores_gemma":[0.00027251686,0.00023449992,0.003126406,0.00018092447,0.00006201925,0.0005436103,0.00040058928,0.11720335,0.026049461,0.048102558,0.8036551,0.00016902175],"about_ca_topic_score_codex":0.0054781055,"about_ca_topic_score_gemma":0.010059356,"teacher_disagreement_score":0.049664002,"about_ca_system_score_codex":0.0011518094,"about_ca_system_score_gemma":0.0022683558,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2295468252","doi":"10.14778/2856318.2856325","title":"Combining quantitative and logical data cleaning","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":104,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Tech University; McMaster University; University of Toronto","funders":"","keywords":"Computer science; Metric (unit); Inference; Functional dependency; Set (abstract data type); Distortion (music); Statistical inference; Data mining; Algorithm; Theoretical computer science; Dependency (UML); Quality (philosophy); Data quality; Artificial intelligence; Relational database; Mathematics","score_opus":0.6149975953066957,"score_gpt":0.46831911800896336,"score_spread":0.14667847729773237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2295468252","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":[0.0035928295,0.00014942081,0.99268883,0.0006299151,0.000032367505,0.00010547845,0.00017439549,0.00170846,0.00091824675],"genre_scores_gemma":[0.083998635,0.00016240145,0.9131966,0.00047913415,0.000047913436,0.0001651064,0.00075363653,0.00044064515,0.0007557734],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9740335,0.007640894,0.0025319667,0.0034688057,0.011448135,0.00087671354],"domain_scores_gemma":[0.9354174,0.030127212,0.0032918307,0.023640499,0.006916509,0.0006065988],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019632932,0.0017632205,0.0020027785,0.0053363573,0.0016248866,0.0050921766,0.0063893017,0.0020154293,0.0025666994],"category_scores_gemma":[0.053283576,0.001416168,0.003621856,0.0052259257,0.0046578194,0.009263154,0.011523994,0.0046287426,0.00091144926],"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":[0.00043006218,0.00048352548,0.010628262,0.0014547125,0.0005501051,0.00070050027,0.0012547749,0.16969982,0.026191847,0.16582385,0.011863963,0.6109186],"study_design_scores_gemma":[0.00008481284,0.00024936403,0.002097468,0.00024168649,0.00021772363,0.0009744816,0.00082373753,0.6208883,0.040977735,0.30364805,0.02963696,0.00015966262],"about_ca_topic_score_codex":0.0030136353,"about_ca_topic_score_gemma":0.0043534073,"teacher_disagreement_score":0.019632932,"about_ca_system_score_codex":0.0024096156,"about_ca_system_score_gemma":0.004903195,"threshold_uncertainty_score":0.1038301},"labels":[],"label_agreement":null},{"id":"W2295513305","doi":"10.14778/3402755.3402776","title":"Debugging data exchange with vagabond","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Debugging; Computer science; Process (computing); Programming language; Background debug mode interface; Algorithmic program debugging; Data exchange; World Wide Web","score_opus":0.08127179066559559,"score_gpt":0.22769392641453015,"score_spread":0.14642213574893456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2295513305","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052663323,0.0005894571,0.7012402,0.00080664526,0.00026380835,0.00031460277,0.0012959149,0.23345369,0.009372401],"genre_scores_gemma":[0.42073876,0.00031213724,0.55383885,0.000594809,0.000045134166,0.00022920167,0.0033000982,0.013401671,0.007539393],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9959859,0.0013657535,0.00030833328,0.001127039,0.001041141,0.00017175774],"domain_scores_gemma":[0.98575413,0.007301373,0.0007717444,0.004802326,0.0010039484,0.00036661004],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006309295,0.001085374,0.0008508928,0.0015904057,0.0006784429,0.002281486,0.003240804,0.0012618942,0.00682128],"category_scores_gemma":[0.021789854,0.0011531019,0.00070158846,0.0007084909,0.0012926968,0.006503921,0.0052565443,0.0021928144,0.0016421566],"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":[0.00608523,0.0008498976,0.03100359,0.0016018363,0.0005714931,0.0026796856,0.010639827,0.029308217,0.052378446,0.046355147,0.09179835,0.7267284],"study_design_scores_gemma":[0.00065046,0.0007141651,0.007912339,0.0006311645,0.00029696454,0.0028299945,0.0012397718,0.4441478,0.16429472,0.04321604,0.33357942,0.0004871126],"about_ca_topic_score_codex":0.0012096893,"about_ca_topic_score_gemma":0.0015452508,"teacher_disagreement_score":0.00682128,"about_ca_system_score_codex":0.00074866123,"about_ca_system_score_gemma":0.0012505649,"threshold_uncertainty_score":0.033367157},"labels":[],"label_agreement":null},{"id":"W2296703446","doi":"10.14778/2733004.2733009","title":"TPC-DI","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":69,"is_retracted":false,"has_abstract":true,"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; Master data; Data warehouse; Variety (cybernetics); Data integration; Enterprise data management; Data management; Context (archaeology); Data science; Analytics; Business intelligence; Database; Enterprise information system","score_opus":0.006928866262876887,"score_gpt":0.2013198379173866,"score_spread":0.19439097165450972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2296703446","genre_codex":"other","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":[0.013979112,0.0015715489,0.09517548,0.0055435025,0.0035152866,0.0028844238,0.06193112,0.10828101,0.7071185],"genre_scores_gemma":[0.10627417,0.0013807261,0.14459471,0.004662905,0.0012735728,0.0029160436,0.3662708,0.024935499,0.34769166],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99076676,0.0012007505,0.0004622287,0.0014036495,0.004890568,0.0012761155],"domain_scores_gemma":[0.9750421,0.0015451125,0.00049336295,0.005772541,0.012720103,0.0044267476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069417832,0.0016049867,0.0012673294,0.003791661,0.0022642575,0.009134666,0.004887708,0.0026565706,0.12481578],"category_scores_gemma":[0.016582849,0.0008613786,0.0006813039,0.005635035,0.0008406202,0.0059934,0.005385824,0.0041100583,0.11615886],"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":[0.0007983202,0.00052179385,0.0019534102,0.00024935522,0.000030260375,0.0001364402,0.0001593814,0.0014175405,0.0048169605,0.019726016,0.81573945,0.15445094],"study_design_scores_gemma":[0.00019770494,0.00029182778,0.0025947841,0.00008864598,0.000017293629,0.00031134425,0.00011980312,0.011242148,0.007570408,0.0043197167,0.9731886,0.000057741276],"about_ca_topic_score_codex":0.01548752,"about_ca_topic_score_gemma":0.008852303,"teacher_disagreement_score":0.12481578,"about_ca_system_score_codex":0.005099148,"about_ca_system_score_gemma":0.009570589,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2301743601","doi":"10.14778/2904483.2904486","title":"Leopard","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"VLSI and FPGA Design Techniques","field":"Engineering","cited_by":93,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Graph partition; Computer science; Graph; Space partitioning; Vertex (graph theory); Algorithm; Theoretical computer science","score_opus":0.0057559873206896065,"score_gpt":0.165789516692164,"score_spread":0.1600335293714744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2301743601","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":[0.017252395,0.0015630064,0.66955554,0.0010181352,0.00091597886,0.0004892748,0.0028863624,0.051552355,0.254767],"genre_scores_gemma":[0.155492,0.0012640739,0.54878825,0.00087121665,0.0001595604,0.0004044976,0.010018055,0.006759912,0.2762425],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99949086,0.000049824892,0.00002531981,0.00013316827,0.00023931827,0.000061428844],"domain_scores_gemma":[0.9995421,0.00006643276,0.00002120512,0.0001831965,0.00013674253,0.00005031278],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004165445,0.0007336564,0.0005107598,0.0010209973,0.0009231182,0.0017417422,0.0016602095,0.0008015404,0.05232141],"category_scores_gemma":[0.0010906226,0.0004097563,0.00054192304,0.00080649,0.00040085346,0.0016354681,0.0017246068,0.0010220172,0.024482267],"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":[0.00059697934,0.00013100794,0.0013405731,0.00036329063,0.000053074542,0.0004232337,0.00024565982,0.015933475,0.029905856,0.057839938,0.13914907,0.75401783],"study_design_scores_gemma":[0.00008980252,0.0001861433,0.0011117611,0.00007617205,0.000027939675,0.0007978191,0.00012440918,0.08816254,0.024279835,0.017291823,0.8677877,0.00006403474],"about_ca_topic_score_codex":0.0030208689,"about_ca_topic_score_gemma":0.005101326,"teacher_disagreement_score":0.05232141,"about_ca_system_score_codex":0.0007756401,"about_ca_system_score_gemma":0.00082139566,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2396123608","doi":"10.14778/3402707.3402730","title":"Business policy modeling and enforcement in databases","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Service-Oriented Architecture and Web Services","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"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; Business rule; Business process; Workflow; Business process modeling; Enforcement; Database design; Business logic; Business; Work in process","score_opus":0.03058924432200034,"score_gpt":0.23099414169690724,"score_spread":0.2004048973749069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2396123608","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":[0.005342597,0.0012221509,0.9813037,0.0020819167,0.0000966298,0.00018140634,0.0002960715,0.00079980295,0.008675717],"genre_scores_gemma":[0.1891767,0.0037650908,0.7948704,0.00091515225,0.00030951083,0.0006970748,0.0012690119,0.00023534501,0.00876174],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99037665,0.004245174,0.001399227,0.0011605461,0.002365425,0.00045301914],"domain_scores_gemma":[0.992959,0.0036294211,0.0006187938,0.0017139488,0.00083684403,0.00024187475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010448995,0.00081609737,0.001141955,0.0023492621,0.0018951223,0.010343259,0.003501613,0.002513385,0.0021633222],"category_scores_gemma":[0.011437295,0.0011914405,0.0020986812,0.0032644535,0.004394689,0.008983365,0.0032570641,0.0034334199,0.00085049233],"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":[0.000031075244,0.000048138358,0.0004141278,0.00013452156,0.00003055003,0.0001517917,0.0005719519,0.04326381,0.00057364313,0.93294907,0.0017752453,0.020056032],"study_design_scores_gemma":[0.000036288562,0.000025812373,0.00014770341,0.00017385003,0.000054798173,0.00012408623,0.0003178797,0.29366672,0.0020883454,0.6496848,0.053642992,0.000036727422],"about_ca_topic_score_codex":0.015422307,"about_ca_topic_score_gemma":0.00950925,"teacher_disagreement_score":0.015422307,"about_ca_system_score_codex":0.0037141985,"about_ca_system_score_gemma":0.005309762,"threshold_uncertainty_score":0.05526024},"labels":[],"label_agreement":null},{"id":"W2401646429","doi":"10.14778/2831360.2831363","title":"Finding Pareto optimal groups","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":80,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Air Force Office of Scientific Research; Science and Technology Planning Project of Guangdong Province; Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Skyline; Computer science; Pruning; Pareto optimal; Scalability; Computation; Point (geometry); Heuristic; Set (abstract data type); Pareto principle; Group (periodic table); Data mining; Algorithm; Theoretical computer science; Mathematics; Mathematical optimization; Artificial intelligence; Database","score_opus":0.036021360278773625,"score_gpt":0.24106280418211742,"score_spread":0.2050414439033438,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2401646429","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":[0.09495758,0.00048161918,0.89613044,0.00039717578,0.000034665933,0.0003649867,0.00096036046,0.0007541738,0.0059190053],"genre_scores_gemma":[0.28492212,0.00028651126,0.70703065,0.00017303682,0.00005238554,0.00045319696,0.0032251752,0.0002840534,0.0035729802],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99804187,0.0004991798,0.00012268116,0.0004940514,0.00056739314,0.00027482907],"domain_scores_gemma":[0.99639636,0.001706921,0.00040118326,0.00050385774,0.00074627943,0.0002453032],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019757552,0.0015188602,0.0020292704,0.0051804055,0.0018730816,0.00253376,0.0016628735,0.0015408525,0.005317073],"category_scores_gemma":[0.008365124,0.00072647823,0.0017385399,0.003721265,0.0011155343,0.0032760315,0.0028124724,0.0011111991,0.0012977837],"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":[0.0006513778,0.0004891765,0.0146885,0.00065420987,0.00038552398,0.0004330963,0.0013917387,0.4471321,0.008547134,0.1074646,0.020964572,0.39719802],"study_design_scores_gemma":[0.00010252413,0.00015936243,0.0013945577,0.00007324917,0.00005887819,0.00016304317,0.00065809494,0.8566238,0.0039278064,0.1293202,0.0074857264,0.000032739288],"about_ca_topic_score_codex":0.0034141755,"about_ca_topic_score_gemma":0.0042788787,"teacher_disagreement_score":0.005317073,"about_ca_system_score_codex":0.0014292622,"about_ca_system_score_gemma":0.002158912,"threshold_uncertainty_score":0.017787337},"labels":[],"label_agreement":null},{"id":"W2402668406","doi":"10.14778/2735461.2735467","title":"Interpretable and informative explanations of outcomes","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Automatic summarization; Heuristics; Computer science; Construct (python library); Set (abstract data type); Dimension (graph theory); Data mining; Binary classification; Binary number; Machine learning; Sample (material); Artificial intelligence; Mathematics","score_opus":0.0047771768864128625,"score_gpt":0.19223746017089116,"score_spread":0.1874602832844783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2402668406","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":[0.038751516,0.0020284418,0.94562936,0.0023668048,0.000107833584,0.00052153494,0.007343734,0.0019449496,0.0013057507],"genre_scores_gemma":[0.20659043,0.0010983448,0.77749294,0.00035896344,0.00025662684,0.00044483325,0.012553986,0.0002711251,0.0009326946],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9942781,0.0025657823,0.00058652414,0.001340446,0.0010356868,0.00019348192],"domain_scores_gemma":[0.9509837,0.038450096,0.003395645,0.0046619284,0.0020701275,0.00043848177],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0076865274,0.0019863388,0.0018784577,0.005205008,0.00082376326,0.0037718534,0.0021459982,0.0017262446,0.004442412],"category_scores_gemma":[0.052522507,0.0010619775,0.0018010784,0.004460458,0.0009831814,0.0070347744,0.0020260506,0.0027149364,0.000906672],"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":[0.0016766438,0.00043083768,0.021773947,0.0037702874,0.0010201288,0.0011548192,0.0035885083,0.13866667,0.010337412,0.09333403,0.03208365,0.6921631],"study_design_scores_gemma":[0.00034508426,0.00041105942,0.008354545,0.0007952171,0.0009521783,0.0008443969,0.0021145882,0.46866038,0.016456407,0.47062203,0.03022397,0.00022012122],"about_ca_topic_score_codex":0.0016684263,"about_ca_topic_score_gemma":0.0029703984,"teacher_disagreement_score":0.0076865274,"about_ca_system_score_codex":0.0012864701,"about_ca_system_score_gemma":0.00261763,"threshold_uncertainty_score":0.040650725},"labels":[],"label_agreement":null},{"id":"W2405215503","doi":"10.14778/3402707.3402722","title":"Data coordination","year":2011,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Base (topology); Data source; Distributed computing; Data mining; Mathematics","score_opus":0.09979098660822669,"score_gpt":0.25600075673108735,"score_spread":0.15620977012286066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2405215503","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":[0.0047523347,0.00053737656,0.9401848,0.0024153283,0.0005529193,0.0010141674,0.0022264444,0.012422456,0.03589427],"genre_scores_gemma":[0.19255792,0.00116366,0.7441426,0.00261106,0.00059165,0.0015430104,0.014624661,0.006729611,0.03603579],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9759986,0.0063211475,0.0034780656,0.006593495,0.006276252,0.0013324212],"domain_scores_gemma":[0.9530836,0.0090587055,0.0018148771,0.0287537,0.005503864,0.001785145],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.021722004,0.0016684438,0.002089309,0.0032453542,0.004274661,0.012048024,0.007774148,0.0023498198,0.018483037],"category_scores_gemma":[0.04253867,0.0015556908,0.0020945917,0.005086179,0.0031037736,0.014568682,0.020766543,0.0039250003,0.007551232],"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":[0.00087764155,0.0002853485,0.0067686974,0.0009649795,0.00035239497,0.0010954096,0.005688902,0.015679069,0.008261882,0.46455634,0.124449916,0.37101936],"study_design_scores_gemma":[0.0001386708,0.000115629875,0.00095158775,0.00031689854,0.00018909365,0.0007276246,0.0015213859,0.03409495,0.015484744,0.15813752,0.7881651,0.00015680182],"about_ca_topic_score_codex":0.007512508,"about_ca_topic_score_gemma":0.0039996817,"teacher_disagreement_score":0.021722004,"about_ca_system_score_codex":0.0026622897,"about_ca_system_score_gemma":0.007238782,"threshold_uncertainty_score":0.1148783},"labels":[],"label_agreement":null},{"id":"W2406955896","doi":"10.14778/2732219.2732227","title":"Multi-core, main-memory joins","year":2013,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":247,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Computer science; Hash join; Joins; Merge sort; Parallel computing; Join (topology); sort; Merge (version control); Hash function; Merge algorithm; SIMD; Theoretical computer science; Sorting algorithm; Database; Programming language; Mathematics","score_opus":0.025084745881657163,"score_gpt":0.23972005889048253,"score_spread":0.21463531300882538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2406955896","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":[0.8507018,0.0020857444,0.12169852,0.00024288375,0.0003347499,0.00031887868,0.0006706389,0.0054707746,0.01847602],"genre_scores_gemma":[0.90058005,0.00034008315,0.09283619,0.00018243605,0.00005852857,0.00016087094,0.00092825905,0.0004885525,0.004425097],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9968611,0.00026961384,0.0002216882,0.0006484922,0.001610381,0.0003887174],"domain_scores_gemma":[0.9932334,0.002767608,0.0004752045,0.0017590183,0.0014209286,0.0003438618],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002464928,0.00067578006,0.00076132367,0.00073696807,0.0011140333,0.0013890114,0.0021296113,0.00069537025,0.0038638024],"category_scores_gemma":[0.0074399174,0.0004097286,0.00032726716,0.0012971368,0.00065100106,0.0033550346,0.0014087806,0.00085028814,0.0008797198],"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":[0.009230415,0.0035960022,0.022113604,0.0016128349,0.00041163096,0.0006497273,0.00131878,0.19573669,0.2969776,0.026904514,0.023526506,0.4179217],"study_design_scores_gemma":[0.0003800692,0.0033750548,0.006190443,0.000048447215,0.00012882962,0.0007282224,0.0006006904,0.56130135,0.39368117,0.011953583,0.021539314,0.00007278998],"about_ca_topic_score_codex":0.00090418995,"about_ca_topic_score_gemma":0.0013004905,"teacher_disagreement_score":0.0038638024,"about_ca_system_score_codex":0.00071123574,"about_ca_system_score_gemma":0.0012516505,"threshold_uncertainty_score":0.013035953},"labels":[],"label_agreement":null},{"id":"W2492590231","doi":"10.14778/2983200.2983203","title":"Distributed data deduplication","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":80,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Data deduplication; Tuple; Computer science; Blocking (statistics); Block (permutation group theory); Relation (database); Backup; Locality; Process (computing); Theoretical computer science; Data mining; Database; Mathematics","score_opus":0.27015668142943616,"score_gpt":0.4095847037039765,"score_spread":0.13942802227454032,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2492590231","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":[0.059925392,0.0018595963,0.92546886,0.0007775997,0.00047554227,0.0007247435,0.0017017503,0.003398426,0.0056680804],"genre_scores_gemma":[0.6663778,0.0010410012,0.31974822,0.00041169825,0.00023138057,0.00069858856,0.0033076024,0.0003751323,0.0078085666],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977374,0.0004630889,0.00026144093,0.0006499902,0.00067651557,0.00021157479],"domain_scores_gemma":[0.98988885,0.001886859,0.0005392479,0.0057534343,0.0016314646,0.0003000933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029006386,0.0009624009,0.0015229644,0.0009954371,0.0015690282,0.0021685874,0.0028828953,0.0007436742,0.0028813796],"category_scores_gemma":[0.008180099,0.00047495659,0.00063607696,0.0023445673,0.00083552446,0.0030634664,0.003867195,0.0010695858,0.0011926611],"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":[0.0019077276,0.00073909527,0.010536206,0.0015383656,0.00039144306,0.001142978,0.0010137161,0.16711324,0.06034495,0.0474415,0.07294006,0.6348907],"study_design_scores_gemma":[0.00038228702,0.00061904197,0.003919673,0.00013251779,0.0001595308,0.0025902395,0.0010795408,0.72922444,0.1134366,0.07963128,0.06867234,0.00015251598],"about_ca_topic_score_codex":0.00071855006,"about_ca_topic_score_gemma":0.0011535881,"teacher_disagreement_score":0.0029006386,"about_ca_system_score_codex":0.00053356314,"about_ca_system_score_gemma":0.0018595309,"threshold_uncertainty_score":0.015340269},"labels":[],"label_agreement":null},{"id":"W2512595308","doi":"10.14778/3067421.3067422","title":"Effective and complete discovery of order dependencies via set-based axiomatization","year":2017,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor; University of Waterloo; York University; Ontario Tech University","funders":"","keywords":"Computer science; Tuple; Completeness (order theory); Inference; Axiom; Rule of inference; Functional dependency; Set (abstract data type); Data mining; Theoretical computer science; Algorithm; Mathematics; Artificial intelligence; Relational database; Discrete mathematics","score_opus":0.10627989270277285,"score_gpt":0.36554846528466095,"score_spread":0.25926857258188807,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2512595308","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":[0.0070721395,0.00029209632,0.9850626,0.0009201721,0.000059455353,0.00033610157,0.002106992,0.0028715532,0.001278822],"genre_scores_gemma":[0.05584779,0.0003873443,0.9356506,0.00049575785,0.00009665486,0.00028764212,0.0057600504,0.0003096078,0.0011645985],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9805393,0.004525507,0.0026212211,0.004284365,0.007110383,0.00091914233],"domain_scores_gemma":[0.9591693,0.025202125,0.0020453515,0.007984961,0.0050581866,0.00054006005],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007920923,0.00194911,0.0023899614,0.0053123184,0.0019229841,0.0059317974,0.0049561486,0.0019330803,0.005811308],"category_scores_gemma":[0.039886728,0.0020539127,0.0068140407,0.0058106165,0.0027072642,0.010758898,0.00791899,0.0068782545,0.002311278],"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":[0.00041341668,0.000787165,0.007527675,0.0019942587,0.0007226984,0.001013832,0.0011301936,0.11793128,0.01834963,0.2625605,0.032860544,0.5547089],"study_design_scores_gemma":[0.00015739976,0.000096920696,0.0010702173,0.0001953658,0.00027310156,0.0007382197,0.00037164698,0.55566514,0.017277392,0.4031721,0.020845665,0.00013673717],"about_ca_topic_score_codex":0.008497953,"about_ca_topic_score_gemma":0.015585256,"teacher_disagreement_score":0.008497953,"about_ca_system_score_codex":0.003190144,"about_ca_system_score_gemma":0.009888368,"threshold_uncertainty_score":0.041890323},"labels":[],"label_agreement":null},{"id":"W2544486974","doi":"10.14778/2994509.2994518","title":"Detecting data errors","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":237,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"University of California Berkeley","keywords":"Computer science; Raw data; Outlier; Data mining; Set (abstract data type); Variety (cybernetics); Data quality; Ground truth; Anomaly detection; Quality (philosophy); Big data; Data science; Machine learning; Artificial intelligence; Engineering","score_opus":0.30976099780462824,"score_gpt":0.41262294928917714,"score_spread":0.10286195148454891,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2544486974","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":[0.11729096,0.002201726,0.810873,0.0032461092,0.0008943844,0.0019433573,0.015100043,0.039766625,0.008683854],"genre_scores_gemma":[0.26726305,0.00065665913,0.7069734,0.0014015376,0.00014395389,0.0009840429,0.015839912,0.0027570985,0.0039804354],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9522556,0.009320097,0.007339614,0.011260907,0.018240035,0.0015837299],"domain_scores_gemma":[0.7991547,0.07963275,0.01970252,0.064534865,0.035493746,0.001481378],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020980965,0.0022116809,0.001998074,0.009275529,0.0016451668,0.0051043616,0.004321513,0.002830164,0.002670607],"category_scores_gemma":[0.15179028,0.000865113,0.0020160691,0.007719268,0.0015979238,0.005799147,0.0065647718,0.0027270466,0.0028608919],"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":[0.00095121656,0.00058896106,0.118958615,0.0037388871,0.00070832815,0.0023220924,0.008237539,0.013977137,0.03586598,0.021077884,0.048595954,0.7449773],"study_design_scores_gemma":[0.00020180193,0.00089806615,0.07099519,0.0021148338,0.0008785389,0.004792019,0.008833255,0.21652001,0.31197852,0.07103218,0.31104565,0.00070994796],"about_ca_topic_score_codex":0.0023586159,"about_ca_topic_score_gemma":0.002255817,"teacher_disagreement_score":0.020980965,"about_ca_system_score_codex":0.0013152073,"about_ca_system_score_gemma":0.0032493733,"threshold_uncertainty_score":0.11095929},"labels":[],"label_agreement":null},{"id":"W2546973305","doi":"10.14778/3007263.3007267","title":"GraphJet","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"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; Exploit; Theoretical computer science; Clique-width; Line graph; Voltage graph","score_opus":0.006945037593673362,"score_gpt":0.19420157507786545,"score_spread":0.18725653748419208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2546973305","genre_codex":"software","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":[0.006090757,0.0017336098,0.2947027,0.0019296794,0.0017445185,0.0007078323,0.032835525,0.5331148,0.12714064],"genre_scores_gemma":[0.07931865,0.003195814,0.32920572,0.0035938285,0.00061564543,0.0009408201,0.20106828,0.08228715,0.29977414],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99917245,0.00009135281,0.00006563095,0.00023485151,0.00033993355,0.000095839605],"domain_scores_gemma":[0.9987173,0.00021068408,0.00005881018,0.0005993309,0.00028341857,0.00013036937],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006136853,0.001430846,0.000811318,0.0015686301,0.0010384743,0.003097087,0.0030908452,0.0015054389,0.11863486],"category_scores_gemma":[0.0033099896,0.00095480314,0.0012021506,0.0014695604,0.0004184545,0.0045795124,0.0037027223,0.0018791935,0.0964596],"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":[0.0005328859,0.00013665152,0.0013711568,0.00083081855,0.000116982155,0.00037444234,0.00024009716,0.004078311,0.009014096,0.034821607,0.6133787,0.3351042],"study_design_scores_gemma":[0.00009968495,0.0000859111,0.0005865994,0.00007071216,0.00003815639,0.000366856,0.00007292995,0.03492459,0.008412727,0.022779541,0.9324914,0.00007088906],"about_ca_topic_score_codex":0.005194614,"about_ca_topic_score_gemma":0.009669951,"teacher_disagreement_score":0.11863486,"about_ca_system_score_codex":0.0008633647,"about_ca_system_score_gemma":0.0013172608,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2547436379","doi":"10.14778/3007263.3007293","title":"Collaborative crowdsourcing with crowd4U","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Mobile Crowdsensing and Crowdsourcing","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Nautical Research Society","funders":"Army Research Office; Microsoft Research; Ministry of Education, Culture, Sports, Science and Technology; Agence Nationale de la Recherche; National Science Foundation","keywords":"Crowdsourcing; Software deployment; Computer science; Task (project management); Set (abstract data type); Data science; Crowdsourcing software development; Human–computer interaction; Knowledge management; Quality (philosophy); World Wide Web; Software engineering; Engineering; Software; Software development","score_opus":0.0036326253293779993,"score_gpt":0.18044141811212303,"score_spread":0.17680879278274503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2547436379","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":[0.027573995,0.0004249171,0.9105622,0.00078509876,0.00052401325,0.0011122428,0.000911513,0.030150713,0.02795539],"genre_scores_gemma":[0.386968,0.0002735152,0.5885065,0.0006141566,0.00016280297,0.001766262,0.0020671762,0.002586939,0.017054657],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99666923,0.0010777081,0.00017251635,0.00079572824,0.00094059814,0.0003441733],"domain_scores_gemma":[0.99681574,0.00096562126,0.00015693031,0.0011425425,0.00048266124,0.00043648412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036168138,0.0012897244,0.0013509452,0.0012509015,0.002954333,0.0028159984,0.0034007502,0.0022440497,0.0077320756],"category_scores_gemma":[0.00862917,0.00092902867,0.0013340997,0.0012907357,0.0019518592,0.0021192823,0.010662928,0.0017086106,0.003970591],"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":[0.0025559796,0.001071058,0.0055498527,0.0011101187,0.0004362583,0.0016358042,0.0055129426,0.34780142,0.054724984,0.093120165,0.09950152,0.38697988],"study_design_scores_gemma":[0.0003135824,0.00022513051,0.00074673066,0.00007313024,0.000048088274,0.0001860409,0.00051007984,0.79950446,0.017956095,0.066959575,0.11331092,0.00016615653],"about_ca_topic_score_codex":0.009531854,"about_ca_topic_score_gemma":0.008119761,"teacher_disagreement_score":0.009531854,"about_ca_system_score_codex":0.0012655834,"about_ca_system_score_gemma":0.0029707584,"threshold_uncertainty_score":0.02586633},"labels":[],"label_agreement":null},{"id":"W2548122763","doi":"10.14778/2994509.2994514","title":"ActiveClean","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":244,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"MNIST database; Computer science; Context (archaeology); Support vector machine; Data mining; Convergence (economics); Process (computing); Class (philosophy); Iterative and incremental development; Machine learning; Artificial intelligence; Deep learning","score_opus":0.009115700324434017,"score_gpt":0.2121947867126384,"score_spread":0.20307908638820438,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2548122763","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":[0.0034649696,0.0013612435,0.8787732,0.0012610811,0.00067428104,0.00035274064,0.006805339,0.09451578,0.012791311],"genre_scores_gemma":[0.06609979,0.0021483838,0.8232331,0.0026526062,0.00036561515,0.0011816679,0.044706933,0.030909756,0.028702075],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99515015,0.001209463,0.00040061979,0.001279485,0.001660123,0.00030015153],"domain_scores_gemma":[0.990415,0.0041446644,0.00035396885,0.0033464874,0.0014871,0.00025289875],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005524416,0.0032810848,0.0029106059,0.0033982706,0.0020984195,0.007996758,0.007369641,0.0035589868,0.04162381],"category_scores_gemma":[0.02294586,0.0024809486,0.0038605367,0.0033418317,0.0014045233,0.0076425574,0.007066833,0.0045604496,0.036416937],"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":[0.0005766489,0.0002924403,0.0043785237,0.0018551188,0.00068702863,0.00044681723,0.0008711849,0.05785158,0.006938002,0.041637167,0.40853354,0.475932],"study_design_scores_gemma":[0.0001804244,0.00011430245,0.0008681212,0.00025392274,0.00012965214,0.0006816341,0.00043683135,0.43555498,0.019381862,0.11413616,0.42810857,0.00015338916],"about_ca_topic_score_codex":0.0036909587,"about_ca_topic_score_gemma":0.0077412687,"teacher_disagreement_score":0.04162381,"about_ca_system_score_codex":0.0008763244,"about_ca_system_score_gemma":0.0030942822,"threshold_uncertainty_score":0.13924551},"labels":[],"label_agreement":null},{"id":"W2548429475","doi":"10.14778/3007263.3007289","title":"Sapphire","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"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; RDF; Information retrieval; Linked data; RDF Schema; Cloud computing; Vocabulary; Database; Semantic Web","score_opus":0.010229025287030399,"score_gpt":0.19948959821345455,"score_spread":0.18926057292642415,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2548429475","genre_codex":"other","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":[0.0055854362,0.0029715756,0.10348954,0.0028510985,0.0018787627,0.00063000864,0.028229,0.15006927,0.70429534],"genre_scores_gemma":[0.05599687,0.005002071,0.11148386,0.0039624358,0.00067453686,0.0011290485,0.11005626,0.052146442,0.6595485],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976829,0.00028076585,0.000120102966,0.00061489077,0.0009940448,0.0003071834],"domain_scores_gemma":[0.99784005,0.00034987464,0.00007762829,0.00063646847,0.00079732656,0.00029859794],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019240195,0.0018053091,0.0010997489,0.0022154842,0.0015959333,0.0058903955,0.0026999626,0.0021997446,0.34270936],"category_scores_gemma":[0.004031231,0.0010139004,0.0013254434,0.0025779966,0.0007937998,0.0056047346,0.0042882734,0.0027566429,0.26407373],"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":[0.0005189589,0.00008665728,0.00066522113,0.0008612247,0.00003585891,0.0005330447,0.0005053964,0.0010349392,0.009787456,0.031424053,0.7700187,0.18452857],"study_design_scores_gemma":[0.00004276027,0.000037779522,0.0004138778,0.00009841087,0.000013204123,0.00038404632,0.00012905689,0.0017115548,0.003056978,0.005730203,0.9883442,0.00003798947],"about_ca_topic_score_codex":0.004742703,"about_ca_topic_score_gemma":0.004425864,"teacher_disagreement_score":0.34270936,"about_ca_system_score_codex":0.001455672,"about_ca_system_score_gemma":0.0024696235,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2549035799","doi":"10.14778/3007263.3007320","title":"Qualitative data cleaning","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":40,"is_retracted":false,"has_abstract":true,"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; Scripting language; Data quality; Data science; Analytics; Big data; Data mining; Qualitative property; Taxonomy (biology); Human error; Data analysis; Machine learning; Engineering; Reliability engineering","score_opus":0.49719438527572407,"score_gpt":0.5118316654115125,"score_spread":0.01463728013578841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2549035799","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063292184,0.0024080246,0.95290375,0.006619236,0.0009322039,0.0020391364,0.007603204,0.003262329,0.017902784],"genre_scores_gemma":[0.09568972,0.004334341,0.86402375,0.0046153786,0.0005877364,0.0041344156,0.011801747,0.0015289113,0.01328402],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.92576873,0.029795036,0.0071179285,0.007476767,0.028182898,0.0016586866],"domain_scores_gemma":[0.8062847,0.06957751,0.014324901,0.052234415,0.05603293,0.0015454716],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.053398643,0.002031484,0.002210829,0.0070739063,0.0042020124,0.009652482,0.0054167253,0.0020544787,0.012347684],"category_scores_gemma":[0.1973461,0.0011848024,0.0029508083,0.011121048,0.0044459286,0.009192078,0.009725417,0.004343951,0.0056048883],"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":[0.0005591808,0.00025971598,0.013604731,0.010330168,0.0005347159,0.00054726924,0.011457157,0.009163247,0.01463893,0.29438984,0.090151325,0.55436367],"study_design_scores_gemma":[0.00010920025,0.0003041657,0.006333708,0.0041732583,0.00026294318,0.00097269646,0.007900896,0.021722533,0.030341312,0.25585318,0.6717395,0.0002865991],"about_ca_topic_score_codex":0.005870326,"about_ca_topic_score_gemma":0.004911764,"teacher_disagreement_score":0.94660133,"about_ca_system_score_codex":0.0051969755,"about_ca_system_score_gemma":0.011822782,"threshold_uncertainty_score":0.2824024},"labels":[],"label_agreement":null},{"id":"W2560800565","doi":"10.14778/3137628.3137635","title":"Revenue maximization in incentivized social advertising","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Optimization and Search Problems","field":"Computer Science","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Submodular set function; Incentive; Revenue; Online advertising; Viral marketing; Monetization; Budget constraint; Knapsack problem; Microeconomics; Computer science; Social graph; Bidding; Advertising; Maximization; Social media; Business; Economics; The Internet; Mathematical optimization; Mathematics","score_opus":0.019394091572590408,"score_gpt":0.2669975460336904,"score_spread":0.2476034544611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2560800565","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":[0.10234477,0.0020880033,0.8599838,0.0042467783,0.0002625308,0.0005522731,0.0012458974,0.0012594571,0.02801654],"genre_scores_gemma":[0.7783884,0.0013094222,0.20319638,0.00087146705,0.00029680645,0.00049303443,0.0010229023,0.00035699853,0.014064591],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99707615,0.0013905051,0.000084638166,0.000592655,0.00033310527,0.000523078],"domain_scores_gemma":[0.9932025,0.005103923,0.0003838311,0.00051968836,0.0003329012,0.0004571812],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003228744,0.0023119815,0.0029052044,0.0008404356,0.00089396595,0.0035407417,0.0032974598,0.0027610417,0.006620575],"category_scores_gemma":[0.011300501,0.0010882657,0.0016091993,0.0016497674,0.0019847306,0.0047348295,0.0025971124,0.00394913,0.0011597936],"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":[0.0007705103,0.00054482353,0.0010951337,0.00062934915,0.00014899507,0.0003419334,0.00038849065,0.7115955,0.003264885,0.20727336,0.014331527,0.059615497],"study_design_scores_gemma":[0.0000917566,0.00006427991,0.00017813234,0.000033116117,0.000030405234,0.00009525766,0.00006566053,0.89958805,0.0007892948,0.096408024,0.0026377495,0.000018382409],"about_ca_topic_score_codex":0.00347565,"about_ca_topic_score_gemma":0.0032681483,"teacher_disagreement_score":0.006620575,"about_ca_system_score_codex":0.0038850782,"about_ca_system_score_gemma":0.0025754867,"threshold_uncertainty_score":0.028188348},"labels":[],"label_agreement":null},{"id":"W2571118757","doi":"10.14778/3015274.3015276","title":"Mostly-optimistic concurrency control for highly contended dynamic workloads on a thousand cores","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":113,"is_retracted":false,"has_abstract":true,"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; Concurrency control; Server; Parallel computing; Concurrency; Cache; Distributed computing; Cache coherence; Lock (firearm); Deadlock; Out-of-order execution; Serialization; Operating system; CPU cache; Database transaction; Cache algorithms; Database","score_opus":0.010182189561394737,"score_gpt":0.23399981616808416,"score_spread":0.2238176266066894,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2571118757","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":[0.67711115,0.0018972154,0.30541518,0.0004502787,0.00013966481,0.00014751487,0.00007773902,0.00788448,0.0068768705],"genre_scores_gemma":[0.96931255,0.0001559821,0.029149342,0.00007615101,0.000022846765,0.000038292153,0.000062452105,0.00008965648,0.0010927449],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998418,0.00025544266,0.00012284039,0.00021494472,0.0007040303,0.00028485787],"domain_scores_gemma":[0.99663687,0.0009183764,0.00041362818,0.0011392161,0.0006372695,0.00025464626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015776413,0.000656544,0.00047593552,0.0004805995,0.0008446991,0.0011149192,0.0015775769,0.00028285594,0.0010311661],"category_scores_gemma":[0.004224886,0.00034733306,0.0002272773,0.00059301395,0.000811714,0.0013729386,0.0010154383,0.00075317227,0.0002181784],"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":[0.0019333636,0.00053676177,0.02545721,0.00037047194,0.00015871105,0.00054821017,0.0010619925,0.2789944,0.27800933,0.021835424,0.007989328,0.3831048],"study_design_scores_gemma":[0.00011757374,0.0003794762,0.0022557748,0.000025231659,0.000046602694,0.0002164148,0.00016611893,0.9101683,0.076738186,0.0054957103,0.0043422976,0.000048251382],"about_ca_topic_score_codex":0.0048182937,"about_ca_topic_score_gemma":0.0065444154,"teacher_disagreement_score":0.0048182937,"about_ca_system_score_codex":0.0007770901,"about_ca_system_score_gemma":0.0025507486,"threshold_uncertainty_score":0.009580493},"labels":[],"label_agreement":null},{"id":"W2572152291","doi":"10.14778/3021924.3021930","title":"Efficient computation of feedback arc set at web-scale","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Computer science; Scalability; Feedback arc set; Set (abstract data type); Randomized algorithm; Greedy algorithm; Theoretical computer science; Probabilistic logic; Algorithm; Graph; Artificial intelligence","score_opus":0.01020956528438349,"score_gpt":0.23808216134682195,"score_spread":0.22787259606243845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2572152291","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":[0.2238443,0.0007043446,0.75426596,0.0009689788,0.000091754184,0.00017295952,0.001965379,0.012511613,0.0054747276],"genre_scores_gemma":[0.6131679,0.00019651555,0.38151065,0.00014162573,0.000034509936,0.0001680683,0.0024304497,0.00059763115,0.0017527171],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99892837,0.0002609457,0.00005632443,0.0003025433,0.000317134,0.00013468918],"domain_scores_gemma":[0.994042,0.0037142416,0.00044702305,0.001001681,0.0005153734,0.00027973563],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014578251,0.0009134967,0.0012757775,0.0021046544,0.0009634785,0.0016856727,0.002350344,0.0015396425,0.0036262032],"category_scores_gemma":[0.013041795,0.0006042317,0.0008515591,0.0021218462,0.0009102586,0.0039256606,0.0017544784,0.0014674048,0.0009263195],"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":[0.00031990136,0.00016464661,0.003260303,0.0002661229,0.000069906244,0.00013136465,0.00017378609,0.87313676,0.0045811716,0.014962002,0.0075183813,0.09541567],"study_design_scores_gemma":[0.000016222932,0.000011545074,0.00024646183,0.000006514351,0.0000061921746,0.000027483284,0.000039974555,0.9814943,0.0011437584,0.016529413,0.00047332182,0.000004873401],"about_ca_topic_score_codex":0.0056691314,"about_ca_topic_score_gemma":0.010569925,"teacher_disagreement_score":0.0056691314,"about_ca_system_score_codex":0.0018792982,"about_ca_system_score_gemma":0.00167581,"threshold_uncertainty_score":0.013635397},"labels":[],"label_agreement":null},{"id":"W2574861468","doi":"10.14778/3025111.3025123","title":"Skipping-oriented partitioning for columnar layouts","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":true,"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; Flexibility (engineering); Workload; Column (typography); Big data; Analytics; Database; Data access; Distributed computing; Tuple; Data science; Data mining; Mathematics","score_opus":0.01434291533108535,"score_gpt":0.23476807476105574,"score_spread":0.2204251594299704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2574861468","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":[0.053381234,0.00077629276,0.93727297,0.00019974967,0.00011839587,0.00016937086,0.00042638098,0.00341576,0.0042398362],"genre_scores_gemma":[0.38271832,0.00055770297,0.60877097,0.0002826352,0.000070304166,0.0002860915,0.0013098498,0.0008158348,0.0051883017],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9994079,0.00011461544,0.00006349992,0.00010830615,0.00020688724,0.00009875514],"domain_scores_gemma":[0.9976084,0.0006016118,0.00017233721,0.0009871614,0.00051395514,0.0001164796],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00054955465,0.0007408674,0.00052556477,0.0008005743,0.00072964345,0.0011176426,0.001691543,0.00043870477,0.0044509736],"category_scores_gemma":[0.0026316089,0.00040944526,0.00046999264,0.001258619,0.00061265356,0.0019553606,0.00137061,0.0005801462,0.0010065502],"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":[0.00069416757,0.00032862564,0.0065043094,0.0007394088,0.00010795406,0.00054858055,0.0007769772,0.21038693,0.15812811,0.066115215,0.029915694,0.525754],"study_design_scores_gemma":[0.00009293999,0.00048260266,0.0018914462,0.000076459466,0.00008970136,0.0008113911,0.00039438388,0.81722575,0.07729391,0.047913965,0.05363211,0.00009526457],"about_ca_topic_score_codex":0.0018731662,"about_ca_topic_score_gemma":0.0052171974,"teacher_disagreement_score":0.0044509736,"about_ca_system_score_codex":0.00061287166,"about_ca_system_score_gemma":0.0009962721,"threshold_uncertainty_score":0.014890015},"labels":[],"label_agreement":null},{"id":"W2579368542","doi":"10.14778/3025111.3025122","title":"Persistent hybrid transactional memory for databases","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Transactional memory; Computer science; Scalability; Software transactional memory; Synchronization (alternating current); Concurrency; Implementation; Transactional leadership; Database transaction; Concurrency control; Embedded system; Operating system; Parallel computing; Database; Software engineering; Computer network","score_opus":0.023698588738854606,"score_gpt":0.2329220493524717,"score_spread":0.2092234606136171,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2579368542","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":[0.04103238,0.002719835,0.94939816,0.0003782783,0.00013463869,0.0001020751,0.00014150035,0.002748339,0.003344889],"genre_scores_gemma":[0.6082025,0.0013328879,0.38440448,0.00027818952,0.00010529872,0.0002658383,0.0003816769,0.000243506,0.0047856937],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9994261,0.0001265777,0.00006581379,0.00011539537,0.00020752139,0.000058562127],"domain_scores_gemma":[0.99854445,0.00042277056,0.0000953996,0.0006346301,0.00023754936,0.00006516752],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007427053,0.00038876056,0.00039472026,0.0005170881,0.000615366,0.0019631407,0.0022782353,0.0006718583,0.0022047784],"category_scores_gemma":[0.002523879,0.00030636106,0.00038529516,0.0009740925,0.0006611276,0.0027235097,0.0016555192,0.0010361297,0.00067017064],"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":[0.0012963921,0.00020308592,0.0036097243,0.0008778501,0.00015665182,0.0004067372,0.00044790746,0.08536195,0.05305982,0.2138369,0.012643249,0.62809986],"study_design_scores_gemma":[0.00015693554,0.0003654445,0.0005533812,0.00007077148,0.00009285292,0.0006310086,0.00019537576,0.7998106,0.0367536,0.1324384,0.028888572,0.000043100474],"about_ca_topic_score_codex":0.00095617457,"about_ca_topic_score_gemma":0.0012117928,"teacher_disagreement_score":0.0022782353,"about_ca_system_score_codex":0.0006601167,"about_ca_system_score_gemma":0.0008730361,"threshold_uncertainty_score":0.007375717},"labels":[],"label_agreement":null},{"id":"W2591700809","doi":"10.14778/3137628.3137631","title":"HoloClean","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":454,"is_retracted":false,"has_abstract":true,"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; Defense Advanced Research Projects Agency","keywords":"Leverage (statistics); Computer science; Probabilistic logic; Inference; Tuple; Data mining; Statistical model; Machine learning; Artificial intelligence; Mathematics","score_opus":0.22842113719041443,"score_gpt":0.4286142488382545,"score_spread":0.20019311164784007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2591700809","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":[0.002447318,0.0008144128,0.9284461,0.0006306557,0.00018847355,0.00025845022,0.0038877549,0.057100773,0.006226082],"genre_scores_gemma":[0.056501098,0.0006951344,0.90884465,0.0010321318,0.000120729215,0.00044922275,0.016860064,0.007929657,0.0075673508],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99493825,0.00079898984,0.00038757044,0.0014657937,0.0021026782,0.0003067198],"domain_scores_gemma":[0.9922168,0.0024181514,0.0004580509,0.0035613473,0.0011020446,0.00024355476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004578141,0.001504516,0.0014914782,0.00361551,0.0010163506,0.0050338036,0.0068213055,0.0015344174,0.01789622],"category_scores_gemma":[0.02542172,0.0012098736,0.0031228617,0.00255591,0.0018275846,0.006923518,0.007741638,0.003278982,0.00826311],"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":[0.00055648084,0.00020679584,0.006176967,0.001574929,0.00032541467,0.00046439565,0.000695684,0.06786154,0.004960692,0.11640966,0.123634756,0.6771326],"study_design_scores_gemma":[0.00010885186,0.000112434514,0.0010843189,0.00033627526,0.00009390851,0.00064940646,0.00017373175,0.46641612,0.012034581,0.2181794,0.3006937,0.00011725965],"about_ca_topic_score_codex":0.008731898,"about_ca_topic_score_gemma":0.015827257,"teacher_disagreement_score":0.01789622,"about_ca_system_score_codex":0.0021318803,"about_ca_system_score_gemma":0.0050244643,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2616147950","doi":"10.14778/3115404.3115409","title":"Auto-join","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Joins; Join (topology); Computer science; Transformation (genetics); Domain (mathematical analysis); String (physics); Database; Data mining; Theoretical computer science; Programming language; Mathematics","score_opus":0.19962649177423786,"score_gpt":0.4134582450415684,"score_spread":0.21383175326733056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2616147950","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":[0.013760381,0.00036082554,0.92473555,0.00021033104,0.00017048143,0.00039397788,0.0024952074,0.044915073,0.012958121],"genre_scores_gemma":[0.20912685,0.00036212246,0.75317687,0.00054457114,0.00018055574,0.00052862504,0.013611353,0.008758059,0.013711105],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9921705,0.0010297982,0.00072233396,0.0019501421,0.0037132483,0.00041398307],"domain_scores_gemma":[0.9898599,0.002834718,0.0004376196,0.004863498,0.0017167947,0.0002874784],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052976366,0.0010989477,0.0011468601,0.0019706234,0.0013674515,0.0032207447,0.0033764248,0.000937036,0.019776393],"category_scores_gemma":[0.013722933,0.0007492726,0.0017462831,0.0022578677,0.0012211773,0.0047265384,0.0064447154,0.001458352,0.0058915266],"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":[0.0014578191,0.00048754466,0.014582558,0.00087992614,0.00033297966,0.0005058463,0.0012513328,0.023803703,0.022435477,0.09813234,0.0789342,0.75719625],"study_design_scores_gemma":[0.00028262648,0.0004530248,0.0039828815,0.0001628235,0.00017958853,0.0017424339,0.0008230765,0.4322891,0.0912307,0.18324469,0.2854356,0.00017343699],"about_ca_topic_score_codex":0.0016211531,"about_ca_topic_score_gemma":0.0021431157,"teacher_disagreement_score":0.019776393,"about_ca_system_score_codex":0.0006244694,"about_ca_system_score_gemma":0.0017373752,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2619906413","doi":"10.14778/3099622.3099623","title":"Revisiting the stop-and-stare algorithms for influence maximization","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":132,"is_retracted":false,"has_abstract":true,"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; Ministry of Education, India","keywords":"Maximization; Scalability; Computer science; Set (abstract data type); Scaling; Order (exchange); Approximation algorithm; Mathematical optimization; Algorithm; Mathematics; Economics","score_opus":0.0194310404711445,"score_gpt":0.28423438600650436,"score_spread":0.26480334553535984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2619906413","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":[0.011282896,0.0008377076,0.9790261,0.0012129591,0.0001472214,0.00012277556,0.00021698092,0.0010570077,0.006096295],"genre_scores_gemma":[0.3302193,0.0011873444,0.65914196,0.0011110975,0.00062952674,0.0003410921,0.00080543273,0.00083130103,0.0057329233],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9953805,0.0021163665,0.00023489565,0.00086892216,0.0010441842,0.00035512183],"domain_scores_gemma":[0.9713637,0.020779418,0.00087850133,0.004399718,0.0019245219,0.0006540426],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069355466,0.0018514653,0.0024056002,0.0020994483,0.0015076358,0.003103946,0.0027378742,0.0022089446,0.005249097],"category_scores_gemma":[0.038718354,0.0008947405,0.0022179952,0.002561452,0.0024861665,0.0064101378,0.0035354546,0.004086678,0.0022209636],"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":[0.00048009277,0.0003362755,0.004663488,0.00051031634,0.00024415477,0.00017091204,0.0006191935,0.49530643,0.0040630265,0.17140748,0.01998568,0.30221283],"study_design_scores_gemma":[0.000035173493,0.00007071879,0.00022266316,0.000032818458,0.000023971257,0.0000773233,0.000046565012,0.9153221,0.0011105064,0.078592464,0.0044520446,0.000013635103],"about_ca_topic_score_codex":0.004252749,"about_ca_topic_score_gemma":0.00765274,"teacher_disagreement_score":0.0069355466,"about_ca_system_score_codex":0.001942327,"about_ca_system_score_gemma":0.0030999002,"threshold_uncertainty_score":0.03667915},"labels":[],"label_agreement":null},{"id":"W2621145626","doi":"10.14778/3099622.3099626","title":"Attribute-driven community search","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":241,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Node (physics); Theoretical computer science; Relevance (law); Graph; Community structure; Cohesion (chemistry); Data mining; Mathematics; Combinatorics","score_opus":0.04102465157361466,"score_gpt":0.3032241378313941,"score_spread":0.26219948625777945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2621145626","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":[0.02493704,0.0018112767,0.96791357,0.00074258546,0.00009328967,0.0003125884,0.00062445196,0.0006334327,0.002931734],"genre_scores_gemma":[0.41214323,0.0014873943,0.57607687,0.0007970922,0.00032598255,0.000573586,0.0029553215,0.00021621653,0.005424307],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996046,0.0014339989,0.00019043426,0.0010072389,0.0010888501,0.00023345047],"domain_scores_gemma":[0.9917973,0.0049389456,0.0006822974,0.0009126957,0.0012803628,0.000388345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003737529,0.0011843978,0.002556851,0.0049547777,0.0018290282,0.002030328,0.003495515,0.0025098906,0.0029397812],"category_scores_gemma":[0.014038618,0.0007495718,0.0016110406,0.007300162,0.0014044109,0.0048324633,0.0035250813,0.0015920254,0.0009970963],"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":[0.00042356257,0.00041963317,0.0050922018,0.0008740696,0.0005221784,0.00027148335,0.00076364743,0.50844556,0.003964516,0.119416185,0.01959518,0.34021175],"study_design_scores_gemma":[0.00007111157,0.000048698963,0.00024600583,0.000024059891,0.000039352788,0.00011486687,0.00007082218,0.9058746,0.00075515674,0.08918751,0.0035496382,0.000018164337],"about_ca_topic_score_codex":0.0048122113,"about_ca_topic_score_gemma":0.0056085386,"teacher_disagreement_score":0.0049547777,"about_ca_system_score_codex":0.0017597283,"about_ca_system_score_gemma":0.0017065883,"threshold_uncertainty_score":0.019766152},"labels":[],"label_agreement":null},{"id":"W2750991217","doi":"10.14778/3137765.3137788","title":"Interactive navigation of open data linkages","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"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; Cloud computing; The Internet; Big data; Interface (matter); Data mining; Linkage (software); Millisecond; Information retrieval; World Wide Web; Database; Operating system","score_opus":0.07352448097533695,"score_gpt":0.3467404140371007,"score_spread":0.2732159330617637,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2750991217","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":[0.061721157,0.0009932445,0.75103176,0.00097416947,0.00038835636,0.0005627416,0.009796159,0.157947,0.016585356],"genre_scores_gemma":[0.26139084,0.000746286,0.6988667,0.0005555399,0.00017721459,0.000543736,0.017022729,0.007960881,0.012736018],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980883,0.0003549771,0.00014291306,0.0005529305,0.00069847534,0.00016248984],"domain_scores_gemma":[0.99438995,0.0023996697,0.00026548235,0.0017467095,0.0005379831,0.00066025194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024335068,0.0010579554,0.001168158,0.0030892908,0.0017221548,0.004094742,0.0023002196,0.0012486338,0.0120203905],"category_scores_gemma":[0.009570195,0.00085357705,0.0010120663,0.0035983026,0.00079459633,0.0062334323,0.011356927,0.0015967414,0.0033324873],"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":[0.0024568187,0.0008278706,0.019556068,0.0012333754,0.0005359304,0.002573105,0.006143983,0.022959944,0.046810348,0.059502948,0.2085518,0.6288479],"study_design_scores_gemma":[0.0004604929,0.00036307116,0.0060185767,0.00027456955,0.00015028135,0.0011536957,0.0017711533,0.4267363,0.05318553,0.096822426,0.41269377,0.0003701915],"about_ca_topic_score_codex":0.005471734,"about_ca_topic_score_gemma":0.010074406,"teacher_disagreement_score":0.0120203905,"about_ca_system_score_codex":0.0007192351,"about_ca_system_score_gemma":0.0016997003,"threshold_uncertainty_score":0.040212154},"labels":[],"label_agreement":null},{"id":"W2751694342","doi":"10.14778/3137628.3137630","title":"Trajectory similarity join in spatial networks","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":179,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"Beijing Nova Program; King Abdullah University of Science and Technology; National Natural Science Foundation of China; Innovationsfonden","keywords":"Join (topology); Computer science; Pruning; Similarity (geometry); Trajectory; Nearest neighbor search; Heuristic; Matching (statistics); Data mining; Scheduling (production processes); Algorithm; Theoretical computer science; Artificial intelligence; Mathematics; Mathematical optimization","score_opus":0.017957525548891625,"score_gpt":0.2326463657509419,"score_spread":0.21468884020205029,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2751694342","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":[0.06919332,0.0013410475,0.922861,0.00048739646,0.00007873572,0.0002716003,0.001151843,0.001394408,0.003220666],"genre_scores_gemma":[0.50178534,0.00090675673,0.488082,0.0001706108,0.00014885096,0.0003003645,0.00370589,0.00019928208,0.0047009387],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99659014,0.00069061894,0.00026821945,0.0009711265,0.0012244239,0.00025546955],"domain_scores_gemma":[0.99618584,0.001619811,0.00052164146,0.000822759,0.0006227158,0.00022717286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023288047,0.0008499649,0.001586289,0.003559474,0.00172811,0.002500695,0.0022682035,0.0013841247,0.0028848972],"category_scores_gemma":[0.010027416,0.00052203273,0.0010924768,0.0058677355,0.0011286174,0.0050805095,0.0036601839,0.0010855193,0.00090601266],"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":[0.00076342176,0.00030915398,0.009325427,0.00040371553,0.00024609655,0.00047993887,0.0007790136,0.5982834,0.0075385193,0.09688129,0.0077548646,0.27723506],"study_design_scores_gemma":[0.00003932925,0.000098985336,0.0011507544,0.00002310965,0.000037664657,0.00025152828,0.00025792126,0.9164852,0.004638363,0.07045581,0.0065377415,0.000023610257],"about_ca_topic_score_codex":0.008056411,"about_ca_topic_score_gemma":0.006782762,"teacher_disagreement_score":0.008056411,"about_ca_system_score_codex":0.0018412189,"about_ca_system_score_gemma":0.0019557108,"threshold_uncertainty_score":0.016019046},"labels":[],"label_agreement":null},{"id":"W2752921427","doi":"10.14778/3137765.3137771","title":"Query-able Kafka","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Stream Mining Techniques","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"","keywords":"SPARK (programming language); Computer science; Analytics; Downstream (manufacturing); Overhead (engineering); Pipeline (software); Upstream (networking); Big data; Order (exchange); Computer network; Data science; Data mining; Operating system; Engineering","score_opus":0.016787425411015303,"score_gpt":0.25097056548130364,"score_spread":0.23418314007028834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2752921427","genre_codex":"software","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":[0.015656587,0.0011432859,0.32439342,0.0016242319,0.0005963613,0.00065286853,0.032786928,0.58021057,0.042935662],"genre_scores_gemma":[0.3616432,0.0012119268,0.40870643,0.0027736044,0.0004006816,0.00138543,0.12849447,0.049556114,0.045828175],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.996424,0.00040239224,0.0003598462,0.0011682747,0.0011390928,0.0005063632],"domain_scores_gemma":[0.9953832,0.0008177056,0.00016291674,0.0024615664,0.0009083482,0.00026627062],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020725233,0.0027697561,0.0013991605,0.0024227789,0.0015794232,0.0051941876,0.004857056,0.0017446813,0.04645725],"category_scores_gemma":[0.011501414,0.0010737036,0.0025032049,0.0027452074,0.0012422801,0.008411454,0.006741965,0.002582067,0.039102767],"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":[0.003197288,0.00036692168,0.0055427435,0.0012497343,0.00031507356,0.0006450918,0.00093370525,0.015470197,0.021082073,0.051463895,0.61457616,0.28515708],"study_design_scores_gemma":[0.0004254203,0.00021835248,0.0029971802,0.00018562659,0.00013256037,0.0010557143,0.00096815877,0.31510594,0.03559323,0.11457664,0.5284517,0.00028953305],"about_ca_topic_score_codex":0.009928129,"about_ca_topic_score_gemma":0.00974182,"teacher_disagreement_score":0.04645725,"about_ca_system_score_codex":0.0016501223,"about_ca_system_score_gemma":0.0025577242,"threshold_uncertainty_score":0.155415},"labels":[],"label_agreement":null},{"id":"W2753088425","doi":"10.14778/3137628.3137637","title":"I've seen \"enough\"","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":69,"is_retracted":false,"has_abstract":true,"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; Visualization; Usability; Sampling (signal processing); Interactivity; Data mining; Context (archaeology); Speedup; Creative visualization; Data visualization; Sample (material); Machine learning; Data science; Human–computer interaction; World Wide Web; Computer vision; Parallel computing","score_opus":0.026778950403935492,"score_gpt":0.29047185050860946,"score_spread":0.26369290010467394,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2753088425","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1722456,0.0059280437,0.5790727,0.02959044,0.0059890132,0.0007389318,0.024349835,0.06834939,0.113736115],"genre_scores_gemma":[0.5284244,0.0034718127,0.42049202,0.0039580814,0.0008539901,0.0005105298,0.01106728,0.0056992345,0.025522724],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995092,0.00011066015,0.000024731233,0.00013173578,0.00015736005,0.00006624041],"domain_scores_gemma":[0.99742174,0.0011918689,0.00020384455,0.0003826995,0.00056036224,0.00023953109],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009754866,0.00096503244,0.00059141376,0.00093149225,0.00079461833,0.0025239692,0.00075306895,0.001266006,0.026823893],"category_scores_gemma":[0.00956004,0.0003808057,0.0008603484,0.001274322,0.00053521065,0.0036129102,0.002032952,0.0017502218,0.0063875737],"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":[0.0017156977,0.00018129416,0.029162137,0.0013809426,0.00024443443,0.0009528795,0.0044492115,0.006648702,0.02038911,0.018324887,0.3134094,0.6031414],"study_design_scores_gemma":[0.00030799766,0.00070452905,0.05250166,0.0017438061,0.00065888575,0.0037152886,0.0065856697,0.11680489,0.03731681,0.10725749,0.6718316,0.00057127135],"about_ca_topic_score_codex":0.0026785543,"about_ca_topic_score_gemma":0.0058685564,"teacher_disagreement_score":0.026823893,"about_ca_system_score_codex":0.00038592526,"about_ca_system_score_gemma":0.00060169917,"threshold_uncertainty_score":0.08973491},"labels":[],"label_agreement":null},{"id":"W2765782779","doi":"10.14778/3151106.3151107","title":"Scalable replay-based replication for fast databases","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Backup; Computer science; Scalability; Bottleneck; Transaction log; Backup software; Replication (statistics); Database; Fault tolerance; Computer network; Database transaction; Bandwidth (computing); High availability; Throughput; Distributed computing; Operating system; Embedded system","score_opus":0.038431743002288236,"score_gpt":0.292418312224537,"score_spread":0.25398656922224877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765782779","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":[0.062307183,0.0021225128,0.9236784,0.0004686229,0.0002109362,0.00019312819,0.00027180975,0.0072370055,0.0035103587],"genre_scores_gemma":[0.70718056,0.00091629784,0.28626624,0.00016202162,0.00020856393,0.00024867788,0.0006093559,0.00031888604,0.0040894053],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983455,0.00037743006,0.00014214132,0.00029837876,0.0006284139,0.00020819192],"domain_scores_gemma":[0.99645954,0.00062854687,0.00025308115,0.001676694,0.0008395651,0.00014254304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016494234,0.0005955813,0.0009994035,0.0007903667,0.0010154718,0.0012118609,0.0021977066,0.00074349844,0.0023406155],"category_scores_gemma":[0.0041582654,0.0005048612,0.0006201468,0.0009858547,0.0005764005,0.002560643,0.0023462535,0.0011395246,0.0011735467],"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":[0.001544437,0.00040860428,0.0028537298,0.00071319303,0.00023434182,0.00084669294,0.0006327295,0.20675492,0.24168737,0.049650762,0.024688732,0.4699844],"study_design_scores_gemma":[0.00017710819,0.00043099048,0.0009643317,0.000044122182,0.00008961072,0.0006599533,0.00017472645,0.89764273,0.057683643,0.021484884,0.020552069,0.00009575277],"about_ca_topic_score_codex":0.0018382714,"about_ca_topic_score_gemma":0.0018843182,"teacher_disagreement_score":0.0023406155,"about_ca_system_score_codex":0.00082879787,"about_ca_system_score_gemma":0.0012329803,"threshold_uncertainty_score":0.00872308},"labels":[],"label_agreement":null},{"id":"W2765816511","doi":"10.14778/3151106.3151111","title":"Efficient mining of regional movement patterns in semantic trajectories","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Trajectory; Computer science; Focus (optics); Data mining; Movement (music); Scheme (mathematics); Semantics (computer science); Space (punctuation); Key (lock); Artificial intelligence; Mathematics","score_opus":0.023246962939578505,"score_gpt":0.242734101600947,"score_spread":0.21948713866136849,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765816511","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":[0.1807454,0.0008627071,0.806033,0.0005907059,0.00005239976,0.00023600197,0.0062940763,0.0024836406,0.0027020331],"genre_scores_gemma":[0.6544463,0.00064424064,0.33016998,0.00008845474,0.00005523491,0.00022406843,0.01222703,0.00018483821,0.0019598834],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990471,0.00014099908,0.00009243727,0.0003726448,0.00022851252,0.00011825321],"domain_scores_gemma":[0.99852717,0.0004374752,0.00033619843,0.00025722195,0.00036318455,0.00007879858],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006424194,0.00091191666,0.0010680051,0.0041263015,0.0007822433,0.0010274878,0.0015195763,0.00086508825,0.0012108306],"category_scores_gemma":[0.004652071,0.00040846635,0.001328083,0.0051142653,0.0006393169,0.0022576335,0.0016447251,0.00070711516,0.00072916725],"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":[0.00067189446,0.0003399781,0.08300839,0.00096175715,0.00045431862,0.0018024797,0.001255288,0.34275162,0.017411016,0.03148465,0.016060587,0.50379795],"study_design_scores_gemma":[0.000028193912,0.00008646768,0.008645934,0.00005205558,0.0000820879,0.0006027273,0.00076912035,0.9462473,0.005466589,0.029191932,0.008800244,0.000027332288],"about_ca_topic_score_codex":0.009537203,"about_ca_topic_score_gemma":0.01707091,"teacher_disagreement_score":0.009537203,"about_ca_system_score_codex":0.00068665814,"about_ca_system_score_gemma":0.0014911437,"threshold_uncertainty_score":0.018963337},"labels":[],"label_agreement":null},{"id":"W2786851308","doi":"10.14778/3199517.3199520","title":"Distributed evaluation of subgraph queries using worst-case optimal low-memory dataflows","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Dataflow; Computer science; Computation; Memory footprint; Joins; Massively parallel; Graph; Parallel computing; Theoretical computer science; Distributed computing; Algorithm","score_opus":0.03065908720145722,"score_gpt":0.27186278200001407,"score_spread":0.24120369479855686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2786851308","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":[0.29222766,0.00047908947,0.69869614,0.0010701901,0.00008152809,0.00018079224,0.00023022329,0.0035749497,0.0034594247],"genre_scores_gemma":[0.88404226,0.000066141,0.11472083,0.0001132869,0.00004586818,0.00009535775,0.00018812841,0.0001899793,0.00053824205],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9945194,0.0014629398,0.00032601503,0.0014855472,0.0013790103,0.0008270432],"domain_scores_gemma":[0.9841303,0.009797818,0.0011652568,0.003268433,0.0010741879,0.0005639633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051711127,0.0010677436,0.001618126,0.00097423175,0.0012556537,0.0027694218,0.0030585013,0.0013887198,0.0015918564],"category_scores_gemma":[0.021312904,0.0006748031,0.0008730162,0.0013365141,0.0029224132,0.005713324,0.002883933,0.0012594872,0.00030526306],"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":[0.0013296445,0.0002738374,0.0048997803,0.00014033649,0.00011458104,0.00018183282,0.00031675166,0.87225056,0.01727868,0.023750922,0.0024786885,0.07698438],"study_design_scores_gemma":[0.000032656702,0.00005097751,0.00016216401,0.0000034652207,0.000012611875,0.000023362965,0.00004155096,0.9792029,0.004240832,0.01601687,0.00020643981,0.0000061358073],"about_ca_topic_score_codex":0.0037516118,"about_ca_topic_score_gemma":0.0044871,"teacher_disagreement_score":0.0051711127,"about_ca_system_score_codex":0.0026276845,"about_ca_system_score_gemma":0.0026788306,"threshold_uncertainty_score":0.027347744},"labels":[],"label_agreement":null},{"id":"W2798664493","doi":"10.14778/3192965.3192973","title":"Table union search on open data","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":193,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Table (database); Computer science; Benchmark (surveying); Data mining; Set (abstract data type); Semantic search; Domain (mathematical analysis); Ontology; Decision table; Probabilistic logic; Information retrieval; Search engine; Artificial intelligence; Mathematics; Programming language","score_opus":0.48054255452095584,"score_gpt":0.4884253644300699,"score_spread":0.007882809909114052,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2798664493","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":[0.08211512,0.002759091,0.8976465,0.0013900466,0.00014061107,0.00033665702,0.0049871216,0.0058034114,0.004821426],"genre_scores_gemma":[0.3087517,0.0007467618,0.67798907,0.00040061743,0.00009948557,0.00035815348,0.008297247,0.0005445703,0.0028124906],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9944501,0.0014248968,0.00050301076,0.0013894216,0.001759829,0.00047275302],"domain_scores_gemma":[0.98208505,0.012250816,0.0011970783,0.0027038248,0.0012919145,0.00047134855],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00439759,0.0009174236,0.0021785507,0.0051318347,0.0019497307,0.0040698266,0.003392497,0.0019510234,0.0052774074],"category_scores_gemma":[0.028504001,0.00093093805,0.0025374359,0.01051983,0.0016488842,0.009212591,0.004871484,0.0015438154,0.0011078159],"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":[0.00087642454,0.0005574672,0.015572079,0.0012703029,0.00044382308,0.001172553,0.0013181301,0.40512142,0.004733785,0.10754552,0.033207204,0.42818123],"study_design_scores_gemma":[0.00009483257,0.00011524092,0.001094164,0.00011171024,0.000097161574,0.00049658254,0.00052198995,0.8407168,0.0041857557,0.14210503,0.010419548,0.000041256662],"about_ca_topic_score_codex":0.006355432,"about_ca_topic_score_gemma":0.006847159,"teacher_disagreement_score":0.006355432,"about_ca_system_score_codex":0.0016372895,"about_ca_system_score_gemma":0.0028072912,"threshold_uncertainty_score":0.023256958},"labels":[],"label_agreement":null},{"id":"W2808068568","doi":"10.14778/3213880.3213884","title":"Morton filters","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Caching and Content Delivery","field":"Computer Science","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Advanced Micro Devices (Canada)","funders":"","keywords":"Bloom filter; Computer science; Hash function; Filter (signal processing); Parallel computing; Set (abstract data type); Cache; Metadata; Data structure; Computer hardware; Algorithm; Operating system; Programming language","score_opus":0.010008721895973577,"score_gpt":0.20012860783806471,"score_spread":0.19011988594209114,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2808068568","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":[0.0370978,0.0022807966,0.89570165,0.0017075536,0.00084359787,0.00058270036,0.0031324457,0.013194701,0.045458723],"genre_scores_gemma":[0.30549216,0.0019495974,0.6029725,0.0017852342,0.0009097256,0.0009598236,0.006063655,0.0016905337,0.078176886],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99687225,0.0003618011,0.00027098178,0.00047899075,0.0016488029,0.00036717055],"domain_scores_gemma":[0.99357444,0.0015590184,0.00040352778,0.0028584609,0.0014356092,0.00016892485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020585114,0.00083008147,0.0012511333,0.0023243532,0.0017157438,0.003463093,0.0026314235,0.0016092162,0.023649668],"category_scores_gemma":[0.010924239,0.00065636286,0.0010320079,0.0027191201,0.0013301763,0.0064389165,0.0030331698,0.0014306982,0.009019312],"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":[0.0015480997,0.00024030452,0.0027694728,0.00057900633,0.000119937264,0.00023472155,0.00054311904,0.023036659,0.021110635,0.30149442,0.07544649,0.57287705],"study_design_scores_gemma":[0.00025476934,0.00069266505,0.001270368,0.0003038435,0.00014329368,0.0010936942,0.0005057782,0.26138753,0.086107284,0.24097511,0.40707472,0.00019097346],"about_ca_topic_score_codex":0.0034368313,"about_ca_topic_score_gemma":0.0051608593,"teacher_disagreement_score":0.023649668,"about_ca_system_score_codex":0.0025204879,"about_ca_system_score_gemma":0.0024968404,"threshold_uncertainty_score":0.07911599},"labels":[],"label_agreement":null},{"id":"W2809683060","doi":"10.14778/3231751.3231764","title":"Experimental analysis of distributed graph systems","year":2018,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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":"PageRank; Computer science; Scalability; Heuristics; Graph; Usability; SPARK (programming language); Distributed computing; Power graph analysis; Theoretical computer science; Database; Human–computer interaction; Operating system","score_opus":0.014885829664977758,"score_gpt":0.24282909842498754,"score_spread":0.22794326876000978,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809683060","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":[0.96361834,0.000705831,0.020584967,0.00071594416,0.00036191,0.00056009233,0.0031789655,0.0030807033,0.007193227],"genre_scores_gemma":[0.9654979,0.00030889385,0.024569703,0.0001605938,0.0001104004,0.0004484296,0.0070443377,0.00035971767,0.0014999247],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99135184,0.0032717602,0.0007378667,0.0016647258,0.0022721586,0.0007016224],"domain_scores_gemma":[0.96718544,0.015040848,0.001331463,0.008356257,0.0067812875,0.0013046545],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049434965,0.0010889875,0.00092530856,0.0016594345,0.001717818,0.0012721997,0.0026432953,0.0008901576,0.0032248392],"category_scores_gemma":[0.023942985,0.0004373534,0.00053696305,0.0033026969,0.0016182924,0.0029288018,0.0016899607,0.0014840327,0.00090928294],"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":[0.0076341876,0.010585439,0.054434918,0.004821795,0.0012731779,0.0011130349,0.00262308,0.5547334,0.082930416,0.019111393,0.06210752,0.19863163],"study_design_scores_gemma":[0.0017931634,0.0063975696,0.042815283,0.00017175716,0.0003345361,0.00085201976,0.0031087638,0.80391574,0.09328342,0.023327233,0.023790658,0.00020988607],"about_ca_topic_score_codex":0.004570648,"about_ca_topic_score_gemma":0.0037153086,"teacher_disagreement_score":0.0049434965,"about_ca_system_score_codex":0.0017186546,"about_ca_system_score_gemma":0.0012700701,"threshold_uncertainty_score":0.026143968},"labels":[],"label_agreement":null},{"id":"W2810219908","doi":"10.14778/3358701.3358704","title":"Online density bursting subgraph detection from temporal graphs","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Caching and Content Delivery","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Huawei Technologies (Canada)","funders":"","keywords":"Computer science; Scalability; Duration (music); Bounded function; Indecomposable module; Set (abstract data type); Burstiness; Combinatorics; Mathematics; Network packet; Computer network","score_opus":0.010226559624712495,"score_gpt":0.19241496182112253,"score_spread":0.18218840219641003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2810219908","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":[0.28442642,0.0012122801,0.7045707,0.0004875454,0.00005218476,0.0001710247,0.0030249483,0.0038638778,0.0021910323],"genre_scores_gemma":[0.80374235,0.0005378391,0.1882124,0.00016470515,0.000067570756,0.000111499096,0.0049309377,0.00022784043,0.0020048574],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995203,0.00007283796,0.00003032249,0.00015678733,0.00014728762,0.00007249284],"domain_scores_gemma":[0.9978453,0.00096782326,0.0003691029,0.00031775446,0.0003218547,0.00017807931],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003824087,0.0006886711,0.00078063895,0.0026825236,0.00049565407,0.00085008924,0.0010900052,0.00051932345,0.0007300569],"category_scores_gemma":[0.0039131637,0.0004239266,0.000573647,0.0024262236,0.00042117425,0.0013489685,0.0009329564,0.00056636834,0.00027828527],"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":[0.00066246616,0.0002763878,0.05394806,0.00079802354,0.00034737156,0.0012528754,0.00077633164,0.3434278,0.060669053,0.022980016,0.017104138,0.49775746],"study_design_scores_gemma":[0.000014238803,0.000031594354,0.0037688992,0.000013162626,0.000031999014,0.00034948005,0.00014487928,0.9709434,0.0050608125,0.017647108,0.001980725,0.000013676305],"about_ca_topic_score_codex":0.010218556,"about_ca_topic_score_gemma":0.020783115,"teacher_disagreement_score":0.010218556,"about_ca_system_score_codex":0.0008115612,"about_ca_system_score_gemma":0.0007931787,"threshold_uncertainty_score":0.02031815},"labels":[],"label_agreement":null},{"id":"W2888965704","doi":"10.14778/3236187.3236194","title":"AIDA","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Relational database management system; Software portability; Python (programming language); Relational database; Relational algebra; Programming language; Database; Interpreter","score_opus":0.00852893521103026,"score_gpt":0.21841857042476365,"score_spread":0.20988963521373338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2888965704","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00747207,0.0024893198,0.2056506,0.007660657,0.0050496613,0.00074673473,0.042618386,0.20260952,0.5257031],"genre_scores_gemma":[0.0705134,0.0032599121,0.21552445,0.007878522,0.001965403,0.0014924802,0.098367825,0.04472743,0.5562706],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99639857,0.000597229,0.00026659484,0.0007886833,0.0014610722,0.000487828],"domain_scores_gemma":[0.99374557,0.0008677373,0.00033279016,0.0021963534,0.0019166265,0.0009409485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003540876,0.0013170977,0.0010858942,0.00279815,0.0020263223,0.008149847,0.0043229554,0.0021592786,0.3098894],"category_scores_gemma":[0.010562313,0.001018634,0.0016467504,0.0025822327,0.0011098001,0.006514131,0.009118358,0.0027957894,0.27864206],"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":[0.0005312364,0.00015831884,0.0021330304,0.00076399616,0.00007108122,0.00040543906,0.0005305134,0.0012481669,0.004152074,0.072373606,0.67466784,0.24296477],"study_design_scores_gemma":[0.000028436607,0.000025872498,0.0004138921,0.00007755937,0.000018059109,0.0002650226,0.00006736766,0.0014577701,0.0016072784,0.009335069,0.98667836,0.000025305793],"about_ca_topic_score_codex":0.0022739258,"about_ca_topic_score_gemma":0.0021767707,"teacher_disagreement_score":0.3098894,"about_ca_system_score_codex":0.0017455793,"about_ca_system_score_gemma":0.004076868,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2889269664","doi":"10.14778/3229863.3236259","title":"MustaCHE","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Cluster analysis; Computer science; Set (abstract data type); Context (archaeology); Visualization; Hierarchical clustering; Range (aeronautics); Artificial intelligence; Data mining","score_opus":0.020014877002886716,"score_gpt":0.27329417948981405,"score_spread":0.2532793024869273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889269664","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":[0.0073809023,0.0019237888,0.5174622,0.0018476581,0.0010406551,0.00054878823,0.0376584,0.35756204,0.074575506],"genre_scores_gemma":[0.06253291,0.0019620252,0.69462097,0.0026234223,0.00031303536,0.0013432684,0.102953896,0.070719555,0.06293096],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99831283,0.00020379118,0.00011102877,0.0004223485,0.0007519231,0.00019813831],"domain_scores_gemma":[0.99777347,0.0007170049,0.00010214007,0.0006621577,0.00053861545,0.00020658202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015820534,0.0015755667,0.0012209191,0.0025272437,0.0013332086,0.00399404,0.0030813564,0.0017274258,0.10420681],"category_scores_gemma":[0.007597186,0.0011464243,0.0019119359,0.0023129168,0.000528184,0.0044121635,0.0052395947,0.0021442305,0.052638475],"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":[0.00070412975,0.00017783589,0.0025344621,0.0013680726,0.0001558495,0.0004153736,0.00073974027,0.0059158118,0.009817398,0.028391749,0.63990414,0.3098755],"study_design_scores_gemma":[0.00016553223,0.00008923255,0.002312621,0.0002998076,0.000048234728,0.00071708683,0.00022616166,0.03954933,0.010619411,0.038151424,0.90766495,0.00015620184],"about_ca_topic_score_codex":0.0051144157,"about_ca_topic_score_gemma":0.009405724,"teacher_disagreement_score":0.10420681,"about_ca_system_score_codex":0.00077282684,"about_ca_system_score_gemma":0.0015499139,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2889272789","doi":"10.14778/3229863.3236248","title":"Tooling framework for instantiating natural language querying system","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Schema (genetic algorithms); Natural language; Natural language user interface; Query language; Programming language; Database schema; Information retrieval; Database; Software engineering; Database design; Natural language processing","score_opus":0.014094250275580228,"score_gpt":0.26261730427254903,"score_spread":0.2485230539969688,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889272789","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":[0.0012517386,0.000095388816,0.9776469,0.00022140797,0.000036306923,0.00037595732,0.00029551642,0.016507808,0.0035689862],"genre_scores_gemma":[0.031734005,0.000281086,0.9599574,0.00022927375,0.00004237241,0.000710365,0.0015524109,0.0016994793,0.003793603],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954775,0.0011509176,0.0008014962,0.0009137275,0.0012973701,0.0003589983],"domain_scores_gemma":[0.99745804,0.0010578946,0.00012398277,0.00072009745,0.0004268577,0.00021309237],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075467196,0.0014260325,0.0013750015,0.0037845427,0.0018391203,0.006654073,0.0047799833,0.0026570042,0.009651029],"category_scores_gemma":[0.0069000334,0.0014510723,0.0031743676,0.0016408899,0.0024357487,0.0062529435,0.005607085,0.0032435623,0.004750265],"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":[0.00022814452,0.0004157272,0.0022607516,0.0009532354,0.0001548965,0.0034329174,0.0031183378,0.035801895,0.02301769,0.7306815,0.024439001,0.17549585],"study_design_scores_gemma":[0.00017080171,0.00016427407,0.0005197435,0.0004168978,0.00014992784,0.0021639278,0.00052912027,0.32789874,0.026967302,0.2596048,0.381184,0.00023044438],"about_ca_topic_score_codex":0.0054320158,"about_ca_topic_score_gemma":0.0042880718,"teacher_disagreement_score":0.009651029,"about_ca_system_score_codex":0.0018209646,"about_ca_system_score_gemma":0.0032646765,"threshold_uncertainty_score":0.03991139},"labels":[],"label_agreement":null},{"id":"W2889537237","doi":"10.14778/3229863.3236267","title":"ConTPL","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Institute of General Medical Sciences","keywords":"Differential privacy; Computer science; Bounding overwatch; Data stream mining; The Internet; Information privacy; Visualization; Data mining; Computer security; Artificial intelligence; World Wide Web","score_opus":0.0235639165118382,"score_gpt":0.25419733674221073,"score_spread":0.23063342023037253,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889537237","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":[0.008962355,0.00054091186,0.38582018,0.0012094771,0.00045134965,0.00091060955,0.009979687,0.5288963,0.06322921],"genre_scores_gemma":[0.28245088,0.0012615348,0.3308858,0.006546825,0.0006479101,0.0025773672,0.0688686,0.11945169,0.1873094],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969041,0.0005200449,0.00027184928,0.0009261864,0.0010930678,0.00028466826],"domain_scores_gemma":[0.9926926,0.0017004472,0.00039575,0.0037656154,0.0011891946,0.00025645163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003154729,0.0016112592,0.0007434577,0.0014299115,0.0010225158,0.0029574274,0.0035625547,0.0017844773,0.06985098],"category_scores_gemma":[0.012721355,0.00096555805,0.0010535883,0.0010057118,0.0010905717,0.007169995,0.006411064,0.0018553576,0.0451647],"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":[0.0022510176,0.00037461834,0.0053300005,0.0010974314,0.000111936555,0.0011344568,0.0011155637,0.004211862,0.01748929,0.06917938,0.49577615,0.40192822],"study_design_scores_gemma":[0.00020886716,0.0002891312,0.0019107015,0.0001848651,0.000052898908,0.0013556412,0.00016864567,0.05965608,0.037768483,0.046010178,0.852227,0.00016758367],"about_ca_topic_score_codex":0.0021828327,"about_ca_topic_score_gemma":0.0020527109,"teacher_disagreement_score":0.06985098,"about_ca_system_score_codex":0.0009997374,"about_ca_system_score_gemma":0.0013393143,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2909564896","doi":"10.14778/3352063.3352120","title":"Guided automated learning for query workload re-optimization","year":2019,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; IBM (Canada); Ontario Tech University","funders":"","keywords":"Computer science; Query optimization; Sargable; Web query classification; Query expansion; Web search query; Query language; SQL; Query plan; Query by Example; Knowledge base; View; SPARQL; Online aggregation; Spatial query; Database; Information retrieval; RDF query language; Data mining; RDF; World Wide Web; Semantic Web; Search engine; Database design","score_opus":0.025071032492887956,"score_gpt":0.2745905191098624,"score_spread":0.24951948661697443,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909564896","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":[0.060888655,0.0006808876,0.8891649,0.0009525241,0.00011522459,0.0005143054,0.0011826113,0.039802782,0.0066980165],"genre_scores_gemma":[0.38459757,0.0003330941,0.60272235,0.00089746574,0.00008069869,0.00048848614,0.005421785,0.0019080569,0.0035503905],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970124,0.0007770991,0.00018579412,0.0009084373,0.00083663425,0.00027971284],"domain_scores_gemma":[0.99431723,0.0032043639,0.00027565,0.0011985949,0.0008680216,0.00013610715],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002188574,0.0018740873,0.0011049664,0.0013377927,0.0007155031,0.001853166,0.0031317305,0.0010703584,0.004231282],"category_scores_gemma":[0.012162262,0.0007716123,0.0012859427,0.0012874232,0.0010043235,0.002806869,0.0025869124,0.002703627,0.0015009714],"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":[0.00040504598,0.0008187586,0.0061067916,0.00050008204,0.00014122947,0.0002721044,0.00056614185,0.27552703,0.018884527,0.008675764,0.02764999,0.66045254],"study_design_scores_gemma":[0.000041353225,0.00005801568,0.00041346072,0.000020021405,0.00003158484,0.000033880224,0.0000897405,0.97793853,0.006699469,0.01037872,0.004280543,0.0000147755345],"about_ca_topic_score_codex":0.0076842145,"about_ca_topic_score_gemma":0.015693054,"teacher_disagreement_score":0.0076842145,"about_ca_system_score_codex":0.0020008544,"about_ca_system_score_gemma":0.0032929138,"threshold_uncertainty_score":0.015278935},"labels":[],"label_agreement":null},{"id":"W2912891501","doi":"10.14778/3291264.3291270","title":"PS-tree-based efficient boolean expression matching for high-dimensional and dense workloads","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Network Packet Processing and Optimization","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"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; Disjoint sets; Matching (statistics); Predicate (mathematical logic); Memory footprint; Tree (set theory); Theoretical computer science; Algorithm; Parallel computing; Mathematics","score_opus":0.008289532133585992,"score_gpt":0.21845650172001987,"score_spread":0.21016696958643388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2912891501","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":[0.06944394,0.00074685377,0.91423273,0.00044862487,0.0001084581,0.00025479187,0.0016311089,0.008114597,0.0050189584],"genre_scores_gemma":[0.45673278,0.00054549804,0.53044397,0.00048968574,0.00009880958,0.00029944407,0.0061638197,0.00059846824,0.004627432],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987105,0.0001765309,0.00014690413,0.00021578767,0.00060130534,0.00014896643],"domain_scores_gemma":[0.99825627,0.00065861346,0.00014825171,0.0005082095,0.000345517,0.00008321134],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081603514,0.00056967,0.0009494238,0.0013421944,0.0006520706,0.0013699594,0.0014501282,0.00056113576,0.0030241867],"category_scores_gemma":[0.005174197,0.0002810902,0.0007781101,0.0034720728,0.0004505208,0.0034420008,0.0013155612,0.00076055207,0.0010052372],"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":[0.0006317106,0.00039564815,0.006845889,0.00043618196,0.00011699721,0.00038506946,0.0003410045,0.094106294,0.05619137,0.038505547,0.0291629,0.7728813],"study_design_scores_gemma":[0.00006102654,0.00015396731,0.00089560676,0.000014780218,0.000035843103,0.00028441282,0.00013034849,0.93731296,0.020015523,0.030394623,0.010676957,0.000024026267],"about_ca_topic_score_codex":0.0032648477,"about_ca_topic_score_gemma":0.004762508,"teacher_disagreement_score":0.0032648477,"about_ca_system_score_codex":0.0009113204,"about_ca_system_score_gemma":0.0017718518,"threshold_uncertainty_score":0.010116935},"labels":[],"label_agreement":null},{"id":"W2915016908","doi":"10.14778/3291264.3291274","title":"Shrinkwrap","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":true,"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; Differential privacy; SQL; Padding; Set (abstract data type); Cardinality (data modeling); Query optimization; Operator (biology); Sargable; Database; Web search query; Information retrieval; Data mining; Computer security; Search engine","score_opus":0.00835529035135328,"score_gpt":0.21258179589086773,"score_spread":0.20422650553951446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915016908","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":[0.08722324,0.0011092887,0.84083855,0.00162802,0.00042680613,0.00087127334,0.0059055127,0.03306601,0.028931312],"genre_scores_gemma":[0.653122,0.0007750051,0.30625153,0.0010430326,0.00021503265,0.0008986903,0.01275988,0.0020563095,0.022878533],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9954791,0.0006863637,0.00037191235,0.001180129,0.0016119716,0.0006704156],"domain_scores_gemma":[0.98805094,0.0009400927,0.00048334082,0.009171895,0.0009815223,0.00037218106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038023498,0.00060661806,0.0009623868,0.00092180335,0.0017181311,0.0030611735,0.002854032,0.000981797,0.00642528],"category_scores_gemma":[0.008791732,0.00066691905,0.0010736863,0.0019360901,0.0014531526,0.010344595,0.008603562,0.0020189572,0.0032568304],"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":[0.0026924233,0.00071186956,0.012577213,0.0008144046,0.0002976019,0.0007670246,0.0012333527,0.03470449,0.042451225,0.27445248,0.12302597,0.506272],"study_design_scores_gemma":[0.00024897404,0.0007762559,0.0041524973,0.00015641272,0.00016441387,0.0022819238,0.000697344,0.27039433,0.08645682,0.2616713,0.3727831,0.00021674976],"about_ca_topic_score_codex":0.0015639613,"about_ca_topic_score_gemma":0.001045898,"teacher_disagreement_score":0.00642528,"about_ca_system_score_codex":0.0009974274,"about_ca_system_score_gemma":0.0026784076,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2915915402","doi":"10.14778/3297753.3297758","title":"Cleaning crowdsourced labels using oracles for statistical classification","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Mobile Crowdsensing and Crowdsourcing","field":"Computer Science","cited_by":36,"is_retracted":false,"has_abstract":true,"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; Oracle; Crowdsourcing; Margin (machine learning); Ground truth; Artificial intelligence; Test data; Data mining; Machine learning; Estimator; Mathematics; Statistics","score_opus":0.04114410296757267,"score_gpt":0.28516068795633887,"score_spread":0.2440165849887662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915915402","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":[0.007256486,0.00036267395,0.986035,0.00062934845,0.0000918671,0.00011873065,0.00023996056,0.0043051536,0.00096091017],"genre_scores_gemma":[0.32211736,0.00032055314,0.66881216,0.0011829329,0.00028262744,0.00051380094,0.0021833468,0.00166612,0.0029210798],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98353344,0.007978245,0.00090561673,0.0030834444,0.0037607558,0.0007384848],"domain_scores_gemma":[0.94828296,0.028054629,0.003525972,0.013275508,0.005623104,0.0012377125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017249987,0.0028265305,0.0035943284,0.0027643458,0.0022224698,0.003571476,0.004950933,0.0030066965,0.0044172606],"category_scores_gemma":[0.074501805,0.0014182734,0.0025088177,0.002570615,0.004006381,0.0058133807,0.00705739,0.006403655,0.0024114628],"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":[0.0013261853,0.0004272452,0.011618638,0.0007614944,0.0002989742,0.0003726961,0.0013556761,0.44704285,0.0082681235,0.06215555,0.021051943,0.44532067],"study_design_scores_gemma":[0.000055553068,0.00008120497,0.0006638672,0.00005297545,0.000029372006,0.000056128367,0.000116648465,0.9328382,0.00418343,0.05791842,0.0039591775,0.00004510475],"about_ca_topic_score_codex":0.0074922484,"about_ca_topic_score_gemma":0.0092473,"teacher_disagreement_score":0.017249987,"about_ca_system_score_codex":0.0027277765,"about_ca_system_score_gemma":0.0039233547,"threshold_uncertainty_score":0.09122771},"labels":[],"label_agreement":null},{"id":"W2917106292","doi":"10.14778/3303753.3303756","title":"Correlation constraint shortest path over large multi-relation graphs","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Relation (database); Theoretical computer science; Computer science; Reachability; Vertex (graph theory); Tree traversal; Enhanced Data Rates for GSM Evolution; Shortest path problem; Longest path problem; Mathematics; Discrete mathematics; Graph; Algorithm; Data mining; Artificial intelligence","score_opus":0.010235153334248006,"score_gpt":0.2181015650240609,"score_spread":0.20786641168981287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2917106292","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":[0.120483294,0.0005549997,0.8736378,0.0010022703,0.00003247619,0.0001633146,0.00125459,0.00097401807,0.0018973973],"genre_scores_gemma":[0.5382194,0.0005780685,0.45610523,0.00025359786,0.000046812835,0.00022729397,0.00231388,0.00027025182,0.001985533],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9973374,0.000944526,0.00017678755,0.00072939094,0.00058059965,0.00023128049],"domain_scores_gemma":[0.9892029,0.00766222,0.0011237735,0.0010989957,0.0005613024,0.00035069045],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017030475,0.0008879293,0.0013327205,0.0014953669,0.0010516609,0.0015479303,0.0018672568,0.0012271485,0.0021864052],"category_scores_gemma":[0.011271031,0.00067295,0.0008618105,0.0048539685,0.0010705761,0.0062541673,0.0019506058,0.0017970351,0.00027433407],"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":[0.00020285508,0.00013407301,0.0025389732,0.0003905185,0.00010247033,0.0008799752,0.00037758108,0.8110193,0.005774776,0.09066295,0.0043819915,0.083534524],"study_design_scores_gemma":[0.000028427668,0.00003552161,0.00048458914,0.00001697136,0.000021469055,0.00022820075,0.00014210149,0.9050405,0.0018883197,0.08996712,0.0021310132,0.000015738437],"about_ca_topic_score_codex":0.006650717,"about_ca_topic_score_gemma":0.009261714,"teacher_disagreement_score":0.006650717,"about_ca_system_score_codex":0.002075499,"about_ca_system_score_gemma":0.001608752,"threshold_uncertainty_score":0.0150588155},"labels":[],"label_agreement":null},{"id":"W2925810066","doi":"10.14778/3342263.3342643","title":"Optimizing subgraph queries by combining binary and worst-case optimal joins","year":2019,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"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; Intersection (aeronautics); Binary number; Partition (number theory); Query plan; Vertex (graph theory); Query optimization; Matching (statistics); Theoretical computer science; Graph; Mathematics; Data mining; Sargable; Combinatorics; Search engine; Information retrieval","score_opus":0.011492111412630185,"score_gpt":0.21651894232646995,"score_spread":0.20502683091383977,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2925810066","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":[0.07612525,0.00048534793,0.91517514,0.0005021378,0.000033298595,0.0001871142,0.00033849175,0.0020392817,0.005113865],"genre_scores_gemma":[0.43606952,0.0002487486,0.56055015,0.00015596337,0.00004560248,0.00015690908,0.00078392465,0.0006559375,0.0013332999],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99598914,0.0010854651,0.0002369245,0.0005998036,0.0016312641,0.0004574387],"domain_scores_gemma":[0.9970925,0.0016106172,0.00027613746,0.00059544866,0.00029067465,0.00013466478],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030855648,0.0011165709,0.0011239804,0.0014560437,0.0007433302,0.0019458486,0.0014830445,0.000747141,0.0016819736],"category_scores_gemma":[0.0066724545,0.00056275976,0.001069676,0.0024688942,0.0014855235,0.0037444513,0.0021021112,0.0012296863,0.00031350553],"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":[0.00039778324,0.0003077616,0.0041110814,0.00017040868,0.000111334106,0.00013539716,0.00021435194,0.7894086,0.009425192,0.037484422,0.0042027165,0.15403098],"study_design_scores_gemma":[0.00003492396,0.00011963766,0.00039996798,0.000010285746,0.00003700939,0.00006235905,0.000083954175,0.9513783,0.0059392643,0.04003942,0.0018791937,0.0000157058],"about_ca_topic_score_codex":0.004872337,"about_ca_topic_score_gemma":0.00897921,"teacher_disagreement_score":0.004872337,"about_ca_system_score_codex":0.0017563791,"about_ca_system_score_gemma":0.0019502484,"threshold_uncertainty_score":0.016318262},"labels":[],"label_agreement":null},{"id":"W2963174348","doi":"10.14778/2994509.2994534","title":"LSH ensemble","year":2016,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":136,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Jaccard index; Computer science; Data mining; Domain (mathematical analysis); Locality-sensitive hashing; Data structure; Data set; Set (abstract data type); Hash function; Mathematics; Cluster analysis; Hash table; Artificial intelligence","score_opus":0.011317738621502458,"score_gpt":0.23601808838573757,"score_spread":0.2247003497642351,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2963174348","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":[0.080153085,0.0025213447,0.89945924,0.0008972928,0.00043859278,0.00031252584,0.0020675035,0.004608433,0.009542054],"genre_scores_gemma":[0.47334793,0.0008565969,0.50237757,0.00096702116,0.0005542141,0.00041187013,0.007565315,0.00062008516,0.01329939],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99796605,0.00040884106,0.000092230934,0.00050910603,0.00080253725,0.0002212636],"domain_scores_gemma":[0.99615246,0.0014610889,0.000194196,0.0011971825,0.00076208374,0.00023299608],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019409133,0.0010179277,0.0020864883,0.0022997153,0.0011301726,0.0014248603,0.0028149916,0.0016493473,0.006213341],"category_scores_gemma":[0.009290032,0.00042350142,0.0009942448,0.0024955203,0.0006770555,0.0035543425,0.00305159,0.001580052,0.0027500233],"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":[0.00029086389,0.00032857267,0.009156119,0.00025169028,0.00027290097,0.0001637288,0.00021498872,0.3901968,0.004409726,0.02080715,0.027332347,0.54657507],"study_design_scores_gemma":[0.000016078342,0.000098877084,0.00066305965,0.000013961767,0.000021507116,0.00012571331,0.000067299705,0.97545445,0.0020188638,0.015924515,0.0055799615,0.000015677067],"about_ca_topic_score_codex":0.004156455,"about_ca_topic_score_gemma":0.00725128,"teacher_disagreement_score":0.006213341,"about_ca_system_score_codex":0.0011206276,"about_ca_system_score_gemma":0.0015052698,"threshold_uncertainty_score":0.02078569},"labels":[],"label_agreement":null},{"id":"W2965684979","doi":"10.14778/3339490.3339492","title":"Finding theme communities from database networks","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Theme (computing); Computer science; Scalability; Database; Tree (set theory); World Wide Web; Mathematics","score_opus":0.0209118966385737,"score_gpt":0.2322965180983817,"score_spread":0.211384621459808,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2965684979","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":[0.20755139,0.0024862213,0.77524686,0.0011017679,0.00007053184,0.0005424679,0.0054738075,0.001783319,0.0057436908],"genre_scores_gemma":[0.49656573,0.0013169067,0.48753652,0.0002533591,0.000103381986,0.0004305969,0.010268036,0.00022585751,0.0032996219],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99846905,0.00027293377,0.00009698834,0.0004942977,0.00047231285,0.000194379],"domain_scores_gemma":[0.99560106,0.0019415978,0.0005544047,0.000510131,0.0010787349,0.0003141409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010758865,0.0007888487,0.00087080256,0.0064665647,0.0018378726,0.0023392576,0.0016795531,0.0011813118,0.0014100309],"category_scores_gemma":[0.010287098,0.00069517415,0.0010464002,0.005295313,0.0006192045,0.0043598125,0.0027564373,0.0009161823,0.00060143654],"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":[0.0007978403,0.0004441288,0.06862874,0.0018519835,0.00054348685,0.0028856485,0.004873841,0.16679767,0.03682191,0.077835426,0.045150157,0.5933691],"study_design_scores_gemma":[0.000047436588,0.000094239105,0.005899583,0.0001308876,0.000105448824,0.0014739529,0.002282386,0.8278164,0.0066270595,0.1352973,0.020183641,0.000041604817],"about_ca_topic_score_codex":0.0070493855,"about_ca_topic_score_gemma":0.012798343,"teacher_disagreement_score":0.0070493855,"about_ca_system_score_codex":0.0009137038,"about_ca_system_score_gemma":0.0012176259,"threshold_uncertainty_score":0.014016688},"labels":[],"label_agreement":null},{"id":"W2966581343","doi":"10.14778/3339490.3339501","title":"Ontology-based entity matching in attributed graphs","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Subgraph isomorphism problem; Computer science; Matching (statistics); Theoretical computer science; Graph; Node (physics); Ontology; Ontology alignment; Factor-critical graph; Induced subgraph isomorphism problem; Semantic Web; Line graph; Mathematics; Artificial intelligence; Voltage graph; Process ontology","score_opus":0.07418500935219877,"score_gpt":0.34623299728772483,"score_spread":0.2720479879355261,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2966581343","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":[0.029661192,0.00016555797,0.96555036,0.00049239467,0.000029023191,0.00029416502,0.001198708,0.0011055913,0.0015030033],"genre_scores_gemma":[0.34697852,0.0003421664,0.64732236,0.00017215926,0.000036153644,0.0002562406,0.0024600413,0.00026290872,0.0021694454],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9925403,0.0025510772,0.0008125678,0.0017694236,0.0016990452,0.00062756374],"domain_scores_gemma":[0.98238796,0.009776312,0.0014209778,0.004460885,0.0014702331,0.0004835869],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054618125,0.000884705,0.0015824331,0.0034211772,0.0019930322,0.004033792,0.0032112368,0.0018663761,0.003947645],"category_scores_gemma":[0.029762201,0.00085403025,0.0017049903,0.0065551293,0.0019511117,0.01642431,0.0058215023,0.0016546943,0.0007097323],"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":[0.0004408231,0.00029103874,0.004029928,0.0006342128,0.00014226984,0.0002658444,0.00073912495,0.38378888,0.0054912176,0.3332412,0.0068333205,0.2641022],"study_design_scores_gemma":[0.00006721368,0.000054217595,0.0005386161,0.000044522574,0.00006744947,0.00013615451,0.00040181848,0.6079054,0.006887747,0.37580192,0.008060287,0.000034775054],"about_ca_topic_score_codex":0.009379285,"about_ca_topic_score_gemma":0.013171856,"teacher_disagreement_score":0.009379285,"about_ca_system_score_codex":0.00303643,"about_ca_system_score_gemma":0.0039280104,"threshold_uncertainty_score":0.028885186},"labels":[],"label_agreement":null},{"id":"W2970397632","doi":"10.14778/3352063.3352102","title":"Making an RDBMS data scientist friendly","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Python (programming language); Relational database management system; Database; Implementation; Relational database; Programming language","score_opus":0.04182728214053458,"score_gpt":0.2994120196958087,"score_spread":0.2575847375552741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970397632","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":[0.022317387,0.0016695772,0.77621245,0.008700892,0.0027517017,0.0007398385,0.0034978327,0.14759688,0.036513466],"genre_scores_gemma":[0.102861024,0.0016623314,0.82311904,0.0043345424,0.000950229,0.00080234354,0.009538103,0.026387606,0.030344719],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9878499,0.0018939185,0.0008733616,0.0015880006,0.0069430633,0.00085178396],"domain_scores_gemma":[0.9801971,0.0024490952,0.00054049364,0.008741246,0.0060124015,0.002059558],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013656611,0.0011161115,0.0012308119,0.0013325169,0.0021274772,0.009221548,0.0075601884,0.00229433,0.012036926],"category_scores_gemma":[0.028268287,0.0017971222,0.0017406862,0.0016593159,0.002379476,0.016068272,0.015148168,0.0070228004,0.01124937],"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":[0.0019150074,0.0006693228,0.007908554,0.0012853859,0.00033915922,0.00090911984,0.0029965271,0.008192131,0.05679564,0.16183415,0.44369572,0.31345925],"study_design_scores_gemma":[0.00028678955,0.00010578054,0.0014404177,0.00020539972,0.000082314196,0.00041806983,0.00031082018,0.03402962,0.027532142,0.04586179,0.8895986,0.00012815984],"about_ca_topic_score_codex":0.0041517443,"about_ca_topic_score_gemma":0.0034376956,"teacher_disagreement_score":0.013656611,"about_ca_system_score_codex":0.0015734506,"about_ca_system_score_gemma":0.004530672,"threshold_uncertainty_score":0.07222396},"labels":[],"label_agreement":null},{"id":"W2970408474","doi":"10.14778/3342263.3342274","title":"PrivateSQL","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":108,"is_retracted":false,"has_abstract":true,"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; Differential privacy; SQL; Relational database; Schema (genetic algorithms); Conjunctive query; Information retrieval; Workload; View; Relation (database); Database; Data mining; Database design","score_opus":0.014256454438070635,"score_gpt":0.2277791953859026,"score_spread":0.21352274094783197,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970408474","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004051033,0.0020413327,0.53282773,0.002875245,0.0006489765,0.0008711866,0.05630909,0.36545685,0.03491852],"genre_scores_gemma":[0.15395464,0.004718132,0.37275976,0.011555698,0.00086954806,0.0026262437,0.2700798,0.10152819,0.081908055],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99180144,0.0017777727,0.0009584874,0.0012799908,0.0034331137,0.00074913184],"domain_scores_gemma":[0.9870236,0.0032249042,0.0006799468,0.0066989474,0.0019587781,0.0004138239],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007529411,0.0016856567,0.001520714,0.0018633718,0.0012185542,0.007394466,0.0068696607,0.0029123928,0.05990458],"category_scores_gemma":[0.02648951,0.001942877,0.0020136386,0.0026746145,0.0021081083,0.011564699,0.010809359,0.0042168414,0.049579754],"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":[0.0016524469,0.00020795902,0.0029013115,0.002102051,0.0002146552,0.00038753237,0.00078810676,0.0067959754,0.0072602862,0.17041811,0.5715935,0.23567806],"study_design_scores_gemma":[0.0003265774,0.00013435862,0.0007014884,0.00024106767,0.00004846329,0.00048113283,0.00016059425,0.026237117,0.011008636,0.1339788,0.8265503,0.00013154003],"about_ca_topic_score_codex":0.0052410536,"about_ca_topic_score_gemma":0.0037833904,"teacher_disagreement_score":0.05990458,"about_ca_system_score_codex":0.0024820855,"about_ca_system_score_gemma":0.0041051777,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2970546348","doi":"10.14778/3342263.33422629","title":"DimmStore","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"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; Testbed; Server; Exploit; Locality; Power (physics); Memory management; Embedded system; Power consumption; Operating system; Computer network; Semiconductor memory","score_opus":0.005575699143324562,"score_gpt":0.18550559531025376,"score_spread":0.1799298961669292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970546348","genre_codex":"other","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":[0.077188864,0.004808824,0.18081407,0.0024521796,0.0016531082,0.001013738,0.028021546,0.3262075,0.37784025],"genre_scores_gemma":[0.4547762,0.0024738156,0.15021062,0.0027383973,0.00046865706,0.0012495299,0.05292195,0.021900704,0.3132601],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993193,0.00007990901,0.000033927856,0.00015026814,0.0003224073,0.000094166164],"domain_scores_gemma":[0.9991053,0.00019713973,0.00004860601,0.00030143297,0.00022655744,0.00012105462],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007098561,0.00082304294,0.0006363919,0.00087025174,0.0004452736,0.001419883,0.0024790508,0.00074508047,0.13703486],"category_scores_gemma":[0.001787604,0.0004772327,0.0004217976,0.00088683306,0.0004260848,0.0020840762,0.0019054118,0.0012236125,0.042118903],"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":[0.0018662494,0.00035425814,0.0030092262,0.0009035107,0.00011241224,0.0005093,0.0002762646,0.0056516803,0.047775913,0.019227501,0.62372756,0.296586],"study_design_scores_gemma":[0.0005221894,0.0005856359,0.0033688068,0.00007957333,0.00006118206,0.0007971844,0.00010297711,0.054235417,0.078153744,0.007185168,0.85481185,0.00009624104],"about_ca_topic_score_codex":0.00087659666,"about_ca_topic_score_gemma":0.0016250497,"teacher_disagreement_score":0.13703486,"about_ca_system_score_codex":0.000710714,"about_ca_system_score_gemma":0.0006423022,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2970613315","doi":"10.14778/3352063.3352096","title":"ApproxML","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Bayesian Modeling and Causal Inference","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"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; Pipeline (software); Reuse; Machine learning; Artificial intelligence; Variety (cybernetics); Mixture model; Process (computing); Gaussian process; Gaussian","score_opus":0.007709354976284009,"score_gpt":0.19244915397182222,"score_spread":0.18473979899553822,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970613315","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":[0.0014145215,0.00048812907,0.9463077,0.0010085489,0.00020461461,0.00013447164,0.004344051,0.03734247,0.008755592],"genre_scores_gemma":[0.10024543,0.0012313512,0.8502906,0.0020897056,0.00047100414,0.0005352881,0.020562567,0.009915202,0.014658841],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9925761,0.0024232573,0.00063451135,0.0014160686,0.0026158015,0.00033431425],"domain_scores_gemma":[0.98231655,0.0076004737,0.0006665563,0.0067687556,0.0023224647,0.0003252115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005810534,0.0015387132,0.0014036361,0.0021940835,0.0011686984,0.007573531,0.0048183785,0.0028000441,0.05055661],"category_scores_gemma":[0.03722121,0.0011451052,0.002387057,0.002321703,0.0015371352,0.009824325,0.0067993472,0.004030102,0.021988098],"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":[0.00075489946,0.00022711759,0.0020482012,0.00090339896,0.00021317929,0.00040010267,0.00040614203,0.05949151,0.0028054016,0.28087112,0.1548818,0.49699703],"study_design_scores_gemma":[0.00011323589,0.00007190247,0.00023359637,0.00012845939,0.000056323337,0.00041791954,0.0000945618,0.45819914,0.0070679765,0.37645975,0.15709414,0.00006300129],"about_ca_topic_score_codex":0.0028832871,"about_ca_topic_score_gemma":0.004464619,"teacher_disagreement_score":0.05055661,"about_ca_system_score_codex":0.0020493348,"about_ca_system_score_gemma":0.0031665163,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2970675344","doi":"10.14778/3352063.3352117","title":"Combating fake news","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Disinformation; Fake news; Crowdsourcing; Social media; Popularity; Internet privacy; Misinformation; Computer science; News media; Political science; Journalism; Public relations; Data science; World Wide Web; Sociology; Media studies; Computer security","score_opus":0.01544656451373473,"score_gpt":0.2722623878711316,"score_spread":0.2568158233573969,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970675344","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":[0.110763535,0.05442415,0.4694032,0.05035423,0.0067037335,0.0013665837,0.004637243,0.014922989,0.2874244],"genre_scores_gemma":[0.6623434,0.030949004,0.2079311,0.0085644545,0.003974945,0.00073039025,0.0075351926,0.0009897964,0.076981746],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9922748,0.0018410863,0.0005272406,0.000852458,0.0037719796,0.0007323978],"domain_scores_gemma":[0.9711493,0.013595651,0.0027812242,0.006217361,0.0055100345,0.0007464139],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069572497,0.0011756506,0.0013299526,0.0053484538,0.002743901,0.009560248,0.0022485617,0.003669065,0.016425334],"category_scores_gemma":[0.036692195,0.00064937264,0.000895995,0.0030932613,0.0017903082,0.01123015,0.0040378626,0.002456519,0.010022854],"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":[0.00029290532,0.0002597191,0.0054449453,0.0013942835,0.00012132884,0.0004975922,0.0018641345,0.004392677,0.0085102385,0.06017018,0.1086363,0.8084157],"study_design_scores_gemma":[0.00009976991,0.0004884728,0.0076749967,0.0015231591,0.00028354634,0.0025242586,0.004621732,0.06570661,0.038217783,0.09201015,0.78666544,0.00018403617],"about_ca_topic_score_codex":0.0022442162,"about_ca_topic_score_gemma":0.0017722643,"teacher_disagreement_score":0.016425334,"about_ca_system_score_codex":0.0014724048,"about_ca_system_score_gemma":0.001970361,"threshold_uncertainty_score":0.05494827},"labels":[],"label_agreement":null},{"id":"W2970727798","doi":"10.14778/3342263.3342638","title":"Distributed implementations of dependency discovery algorithms","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"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; Implementation; Pruning; Dependency (UML); Focus (optics); Space (punctuation); Computation; Distributed computing; Big data; Theoretical computer science; Distributed algorithm; Algorithm; Parallel computing; Data mining; Artificial intelligence; Programming language","score_opus":0.07908823019991898,"score_gpt":0.3774095362212648,"score_spread":0.29832130602134577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970727798","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":[0.00914007,0.00015244464,0.9877407,0.00023237163,0.000034804478,0.00007865781,0.000050250008,0.0012293936,0.0013413499],"genre_scores_gemma":[0.26836255,0.00023237741,0.72867286,0.00017436994,0.00006416039,0.00030434722,0.00029938726,0.00026307817,0.0016267891],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99627787,0.0011380382,0.00031070455,0.0009744327,0.0010340584,0.00026491782],"domain_scores_gemma":[0.98868495,0.0056152805,0.0006219729,0.0037278729,0.001052319,0.00029756204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004650561,0.0010002266,0.0009893214,0.0009601292,0.0011156396,0.0027591505,0.003324919,0.0014127669,0.0037322042],"category_scores_gemma":[0.018046895,0.0006610148,0.0010203404,0.001818923,0.0012627352,0.0042113387,0.0027894261,0.0026792218,0.000973572],"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":[0.0006605982,0.00062834675,0.0057470915,0.0005433969,0.00031896983,0.00022105628,0.00053671486,0.41448835,0.00816999,0.15528922,0.008083122,0.40531322],"study_design_scores_gemma":[0.0001031578,0.000093266,0.00026664042,0.000024152203,0.00004619503,0.00010288457,0.000074512885,0.90708786,0.004508978,0.08386495,0.0038104164,0.000016959248],"about_ca_topic_score_codex":0.0016196619,"about_ca_topic_score_gemma":0.0030158297,"teacher_disagreement_score":0.004650561,"about_ca_system_score_codex":0.0011677019,"about_ca_system_score_gemma":0.0028346195,"threshold_uncertainty_score":0.024594843},"labels":[],"label_agreement":null},{"id":"W2970828623","doi":"10.14778/3342263.3342645","title":"Efficient algorithms for densest subgraph discovery","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":115,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Induced subgraph isomorphism problem; Intuition; Computer science; Subgraph isomorphism problem; Graph; Algorithm; Color-coding; Efficient algorithm; Theoretical computer science; Artificial intelligence; Line graph","score_opus":0.013070494118829084,"score_gpt":0.23755038924642918,"score_spread":0.2244798951276001,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970828623","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":[0.013130378,0.0009566081,0.97727793,0.0006265506,0.00007726126,0.0003665573,0.001011948,0.0039499993,0.0026027618],"genre_scores_gemma":[0.07665639,0.00054792356,0.914748,0.00024415384,0.00008888763,0.00043749897,0.004659056,0.00044483168,0.0021732524],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99614894,0.0007727728,0.0002957236,0.0011664355,0.0011720717,0.00044414419],"domain_scores_gemma":[0.99054503,0.0048018084,0.0007362175,0.0024066407,0.001161817,0.00034847402],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027628767,0.0024508866,0.0026835585,0.005454037,0.0018258239,0.002922295,0.004275587,0.0027045065,0.0063237124],"category_scores_gemma":[0.015302,0.0014699965,0.0027812729,0.007930484,0.0012770729,0.0061166007,0.004554974,0.0024068626,0.0030394227],"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":[0.0003835076,0.0006338007,0.004093499,0.0009383431,0.00028945622,0.00026695954,0.0005675493,0.20482634,0.0063864705,0.044349227,0.037354976,0.6999099],"study_design_scores_gemma":[0.00014947176,0.00008014928,0.00069081143,0.00004501609,0.000077036406,0.00035818852,0.0001911556,0.870224,0.0029184313,0.116618276,0.008617361,0.000030112758],"about_ca_topic_score_codex":0.0065972116,"about_ca_topic_score_gemma":0.013496765,"teacher_disagreement_score":0.0065972116,"about_ca_system_score_codex":0.002329322,"about_ca_system_score_gemma":0.004490763,"threshold_uncertainty_score":0.02115494},"labels":[],"label_agreement":null},{"id":"W2970882829","doi":"10.14778/3352063.3352146","title":"PNUTS to Sherpa","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Cloud computing; Computer science; Operating system","score_opus":0.0041628378496365685,"score_gpt":0.19134434637664263,"score_spread":0.18718150852700607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970882829","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":[0.03324667,0.009089656,0.40900347,0.08236482,0.013271174,0.00033644197,0.0026968636,0.0631459,0.386845],"genre_scores_gemma":[0.25394446,0.0101842,0.2619002,0.032340966,0.006458567,0.0006386358,0.0064016557,0.022710182,0.40542117],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971318,0.0006418346,0.000118820906,0.0005570102,0.0011249463,0.00042553636],"domain_scores_gemma":[0.9946571,0.001029089,0.00023659613,0.002274211,0.0011442014,0.00065880286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029301094,0.0006312766,0.00044795012,0.0011016657,0.001650714,0.0041634063,0.0024644134,0.001291406,0.057156738],"category_scores_gemma":[0.010300068,0.0006193373,0.0006014323,0.0019123371,0.0025904435,0.009544611,0.006048755,0.004476679,0.019108078],"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":[0.0004807139,0.00009761948,0.001266678,0.00027446027,0.000029237675,0.00026157644,0.0006558964,0.0024321154,0.002253409,0.32139426,0.3343387,0.3365153],"study_design_scores_gemma":[0.00005343618,0.000059290374,0.00038460476,0.00010833593,0.000013703849,0.00022322565,0.0002575091,0.004513705,0.0019415313,0.038036797,0.9543762,0.000031731826],"about_ca_topic_score_codex":0.0112444125,"about_ca_topic_score_gemma":0.007442125,"teacher_disagreement_score":0.057156738,"about_ca_system_score_codex":0.002071631,"about_ca_system_score_gemma":0.0033249413,"threshold_uncertainty_score":0.19120836},"labels":[],"label_agreement":null},{"id":"W2970956651","doi":"10.14778/3342263.3342627","title":"Ocean vista","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Natural Sciences and Engineering Research Council of Canada","funders":"","keywords":"Computer science; Computer network; Distributed computing; Gossip; Latency (audio); Concurrency control; Distributed transaction; Serializability; Transaction processing; Replication (statistics); Asynchronous communication; Database transaction; Database; Telecommunications","score_opus":0.004017234516262306,"score_gpt":0.1842977793246883,"score_spread":0.18028054480842598,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970956651","genre_codex":"other","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":[0.03573363,0.0034736102,0.20812142,0.002677773,0.0020775145,0.001014616,0.012393084,0.2254609,0.50904745],"genre_scores_gemma":[0.27276334,0.0030612773,0.24938487,0.0018941647,0.00055427046,0.0012435861,0.04653596,0.027585134,0.3969774],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991315,0.00010232338,0.000058854097,0.00018790092,0.00037446333,0.00014493447],"domain_scores_gemma":[0.9983541,0.00022314569,0.00012388166,0.00063006446,0.000409912,0.00025894464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009868338,0.0009067277,0.000733367,0.0011392349,0.00089168857,0.0028790243,0.0021082517,0.001162426,0.09179307],"category_scores_gemma":[0.0030124807,0.0006930445,0.00058921304,0.0011748329,0.0007637532,0.003783926,0.0039130147,0.0017944222,0.041170564],"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":[0.0033094604,0.0002714144,0.005665063,0.0012892159,0.00016870239,0.00073729904,0.0007097545,0.009979264,0.053193223,0.06308549,0.4180267,0.44356447],"study_design_scores_gemma":[0.0003583906,0.00029653526,0.0015425122,0.00016065415,0.000057028883,0.00047502207,0.00016337437,0.036845364,0.014606983,0.013791334,0.93162173,0.000081116326],"about_ca_topic_score_codex":0.0042229868,"about_ca_topic_score_gemma":0.0040454734,"teacher_disagreement_score":0.09179307,"about_ca_system_score_codex":0.0009861407,"about_ca_system_score_gemma":0.001985956,"threshold_uncertainty_score":0.30707836},"labels":[],"label_agreement":null},{"id":"W2970992672","doi":"10.14778/3352063.3352116","title":"Data lake management","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":236,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"TD Bank Group; University of Toronto","funders":"","keywords":"Metadata; Data management; Metadata management; Data science; Data management plan; Computer science; Data integration; Software versioning; Data mapping; Data extraction; Data element; Research data; Data virtualization; Data curation; Database; World Wide Web; Software","score_opus":0.19096554616353498,"score_gpt":0.3896089267691528,"score_spread":0.19864338060561784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970992672","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":[0.011691869,0.007111954,0.55940956,0.02324033,0.0028328507,0.0030808086,0.10107316,0.17754924,0.11401019],"genre_scores_gemma":[0.11187629,0.007822782,0.5482424,0.0051960056,0.001630319,0.0024580685,0.22207904,0.02403266,0.07666242],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9902188,0.0013882162,0.001680583,0.0014432638,0.0045852875,0.00068390084],"domain_scores_gemma":[0.97794515,0.0032139625,0.0015340964,0.009015027,0.006640497,0.001651296],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013512635,0.0010647658,0.0014221334,0.007892552,0.0031229446,0.012696608,0.005567789,0.0014988105,0.0280816],"category_scores_gemma":[0.041428905,0.0010784767,0.0014979772,0.009107835,0.0011268331,0.017185194,0.0112716155,0.0032642502,0.01830473],"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":[0.00021415568,0.00008546647,0.006912315,0.0009170364,0.00013103639,0.00025021736,0.0011389395,0.0024277007,0.003093446,0.07858947,0.5607018,0.34553835],"study_design_scores_gemma":[0.00003398857,0.00003280322,0.0018507765,0.00025515116,0.00004663851,0.00020707393,0.0004127493,0.011222766,0.0059730695,0.032935698,0.94695944,0.00006990021],"about_ca_topic_score_codex":0.0065349108,"about_ca_topic_score_gemma":0.004540147,"teacher_disagreement_score":0.0280816,"about_ca_system_score_codex":0.002542973,"about_ca_system_score_gemma":0.0071443515,"threshold_uncertainty_score":0.093942285},"labels":[],"label_agreement":null},{"id":"W2971000368","doi":"10.14778/3352063.3352080","title":"VISE","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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; Pipeline (software); Interface (matter); Convolutional neural network; Nearest neighbor search; Scalability; Feature (linguistics); Image retrieval; Artificial intelligence; Frame (networking); Search engine; Computer vision; Information retrieval; Image (mathematics); Database","score_opus":0.005005524439825163,"score_gpt":0.21708860347009692,"score_spread":0.21208307903027176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971000368","genre_codex":"software","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":[0.013558502,0.003127722,0.27051944,0.001183334,0.0013709861,0.00061836955,0.047670957,0.5072974,0.15465334],"genre_scores_gemma":[0.12877414,0.0028441926,0.3355346,0.0032186334,0.0006115645,0.0009094626,0.23798236,0.04530446,0.24482065],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994809,0.000055384124,0.000027947619,0.00013334895,0.00022533185,0.00007706995],"domain_scores_gemma":[0.99933726,0.00010346734,0.00002825922,0.00021063941,0.00021346554,0.00010695372],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006562337,0.0011107593,0.00075117353,0.0014016685,0.00046753522,0.002094226,0.0018860258,0.0011826949,0.09931005],"category_scores_gemma":[0.0020698986,0.00051202724,0.00068623875,0.0008719541,0.00034037774,0.0032352235,0.002463375,0.0011618653,0.06800454],"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":[0.0008353771,0.00020494545,0.0014883155,0.00051146705,0.000114706614,0.00028237127,0.00012509058,0.0023749627,0.016262066,0.012573122,0.7030102,0.26221743],"study_design_scores_gemma":[0.00018207327,0.00022026474,0.0015528549,0.000091985225,0.00003677208,0.00061697414,0.00006451461,0.035947405,0.021253739,0.009123646,0.93080187,0.0001078699],"about_ca_topic_score_codex":0.0027378506,"about_ca_topic_score_gemma":0.0047785137,"teacher_disagreement_score":0.09931005,"about_ca_system_score_codex":0.00049409573,"about_ca_system_score_gemma":0.00063693133,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2971120804","doi":"10.14778/3352063.3352064","title":"GALO","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; IBM (Canada)","funders":"","keywords":"Computer science; SPARQL; Knowledge base; SQL; Query optimization; Process (computing); Query plan; Plan (archaeology); Information retrieval; Base (topology); RDF; Web query classification; Sargable; Database; Data mining; Web search query; World Wide Web; Search engine; Semantic Web; Programming language","score_opus":0.00448911751621578,"score_gpt":0.1879691187956926,"score_spread":0.18348000127947683,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971120804","genre_codex":"software","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":[0.010525872,0.0019060544,0.21107273,0.0041572647,0.0013839058,0.0011346766,0.031998202,0.37604496,0.36177632],"genre_scores_gemma":[0.15662855,0.0026212423,0.29083353,0.007931165,0.0011530227,0.001276213,0.1750861,0.07154089,0.29292932],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99661726,0.00043733127,0.00017775806,0.000897536,0.0015097364,0.00036037245],"domain_scores_gemma":[0.9944537,0.0010831242,0.00019057078,0.0022193766,0.0015828061,0.0004703345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003044591,0.0014822516,0.0009758264,0.0026689996,0.0009647626,0.005127639,0.0036481216,0.0015259404,0.11917507],"category_scores_gemma":[0.009577267,0.0008245737,0.0011715345,0.0019537152,0.00091060164,0.005481177,0.0048860894,0.0026337474,0.087329365],"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":[0.0010085172,0.0002805994,0.002508872,0.00065258617,0.00008520794,0.00024293318,0.0003575059,0.0025153717,0.009650969,0.0384119,0.6089027,0.33538282],"study_design_scores_gemma":[0.00017910646,0.00011190967,0.0013275428,0.000118732394,0.00003655697,0.0002629044,0.00012855064,0.017820202,0.0065975506,0.02193326,0.9514161,0.000067657485],"about_ca_topic_score_codex":0.003790809,"about_ca_topic_score_gemma":0.0035779031,"teacher_disagreement_score":0.11917507,"about_ca_system_score_codex":0.001738885,"about_ca_system_score_gemma":0.0022438779,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2971290973","doi":"10.14778/3342263.3342633","title":"An intermediate representation for optimizing machine learning pipelines","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":48,"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; York University","keywords":"Computer science; Preprocessor; Pipeline transport; Domain (mathematical analysis); External Data Representation; Representation (politics); Semantics (computer science); Data pre-processing; Feature (linguistics); Programming language; Feature engineering; Artificial intelligence; Theoretical computer science; Machine learning; Deep learning","score_opus":0.08481634541627903,"score_gpt":0.37581740113264334,"score_spread":0.2910010557163643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971290973","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":[0.0038119282,0.00007099726,0.9846416,0.00017012723,0.00004003182,0.00005224551,0.00043557858,0.00887308,0.0019044034],"genre_scores_gemma":[0.104647085,0.00014830322,0.8854604,0.0001856129,0.00006115962,0.00036952665,0.0033201228,0.0028439169,0.0029638745],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977126,0.00050649757,0.00024727933,0.0004259828,0.0007855613,0.00032205658],"domain_scores_gemma":[0.9963546,0.0012266306,0.00019055269,0.0013305863,0.00077816215,0.00011952353],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002646128,0.0016301265,0.00096544035,0.0014950694,0.00095878815,0.0044449703,0.0035712437,0.0013432397,0.010614404],"category_scores_gemma":[0.009615853,0.00086006965,0.0022984375,0.002019137,0.0012009015,0.0046085506,0.002981791,0.0028784822,0.0041705533],"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":[0.00077290914,0.0003273059,0.002673076,0.0005844867,0.00010575488,0.0003450682,0.0005044057,0.37590447,0.017247815,0.27938142,0.033019856,0.28913337],"study_design_scores_gemma":[0.000061404855,0.00011728641,0.00028328618,0.0000667152,0.000055294546,0.00007428202,0.000090053356,0.82318753,0.020878745,0.1340856,0.021059321,0.00004039722],"about_ca_topic_score_codex":0.002789357,"about_ca_topic_score_gemma":0.004125796,"teacher_disagreement_score":0.010614404,"about_ca_system_score_codex":0.0016042962,"about_ca_system_score_gemma":0.002870207,"threshold_uncertainty_score":0.035508692},"labels":[],"label_agreement":null},{"id":"W2982295803","doi":"10.14778/3364324.3364330","title":"LINC","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Cluster analysis; Theoretical computer science; Graph; Algorithm; Data mining; Artificial intelligence","score_opus":0.003710575606360767,"score_gpt":0.20465746596577183,"score_spread":0.20094689035941107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2982295803","genre_codex":"other","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":[0.0069426573,0.0025722887,0.095322266,0.010765937,0.0048912535,0.0006232626,0.013306106,0.0068075815,0.8587686],"genre_scores_gemma":[0.112021185,0.003871215,0.06995794,0.011510713,0.0016660006,0.0012090622,0.026576504,0.0031284005,0.77005905],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99757177,0.0004209816,0.00015203636,0.0007644745,0.0008119173,0.00027889514],"domain_scores_gemma":[0.9966307,0.0006343748,0.00020908458,0.00093547,0.0012466302,0.00034364464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001560063,0.0009991009,0.0006394644,0.0014948971,0.002609862,0.0048620133,0.0021663257,0.002341452,0.29665402],"category_scores_gemma":[0.0077209077,0.00038749218,0.0007710654,0.0019860761,0.0012995148,0.00343682,0.0037207345,0.0018665104,0.1744835],"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":[0.00015735404,0.00010238794,0.002388041,0.00046675449,0.000032062297,0.0005254432,0.0005291301,0.0012240023,0.0026293215,0.150989,0.48915994,0.3517966],"study_design_scores_gemma":[0.000015750322,0.000027720764,0.0007773402,0.00010016446,0.0000115968805,0.00035905285,0.0001943132,0.0010433879,0.00089664693,0.026886493,0.96966994,0.00001762017],"about_ca_topic_score_codex":0.0040584104,"about_ca_topic_score_gemma":0.0064152814,"teacher_disagreement_score":0.29665402,"about_ca_system_score_codex":0.0014553498,"about_ca_system_score_gemma":0.0025649923,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2982581272","doi":"10.14778/3364324.3364332","title":"Secure multi-party functional dependency discovery","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"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; Profiling (computer programming); Computation; Secure two-party computation; Cryptography; Dependency (UML); Functional dependency; Access control; Distributed computing; Secure multi-party computation; Cryptographic protocol; Semantics (computer science); Data anonymization; Theoretical computer science; Computer security; Data mining; Information privacy; Algorithm; Relational database; Programming language; Artificial intelligence","score_opus":0.011561321195034262,"score_gpt":0.2036494168363956,"score_spread":0.19208809564136134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2982581272","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":[0.037540406,0.00019296694,0.9580129,0.0008948948,0.000043686603,0.00020567393,0.00028885782,0.000862561,0.0019581937],"genre_scores_gemma":[0.83522624,0.00022616966,0.16009943,0.00024522474,0.00006728341,0.0002826408,0.00064918166,0.00012186914,0.0030819413],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9896729,0.0028780333,0.0008797057,0.0015679928,0.0040412624,0.00096007297],"domain_scores_gemma":[0.9707474,0.009495327,0.0024609691,0.01523483,0.001550925,0.00051059836],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0072818934,0.00068291445,0.0014719934,0.0012353873,0.0024015605,0.0031854543,0.0026568044,0.0018669553,0.002437499],"category_scores_gemma":[0.018734915,0.00082868105,0.0017206216,0.001766126,0.0025091686,0.008888927,0.008121251,0.0030643665,0.0010456648],"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":[0.0016032466,0.0004312751,0.008699136,0.0005531913,0.00034412197,0.0014467945,0.0011519237,0.14102416,0.040613968,0.5046613,0.01086373,0.28860712],"study_design_scores_gemma":[0.00008090894,0.00014371568,0.0010295728,0.000048520254,0.00009006647,0.0009226298,0.00033053404,0.60544753,0.042025138,0.33798394,0.011829557,0.000067887515],"about_ca_topic_score_codex":0.0005193014,"about_ca_topic_score_gemma":0.0006742149,"teacher_disagreement_score":0.0072818934,"about_ca_system_score_codex":0.0017549619,"about_ca_system_score_gemma":0.0030597127,"threshold_uncertainty_score":0.0385108},"labels":[],"label_agreement":null},{"id":"W3000259868","doi":"10.14778/3372716.3372728","title":"Evaluating persistent memory range indexes","year":2019,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Parallel Computing and Optimization Techniques","field":"Computer Science","cited_by":94,"is_retracted":false,"has_abstract":true,"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; Benchmarking; Index (typography); Dram; Range (aeronautics); Tree (set theory); Key (lock); Data structure; Focus (optics); Persistent data structure; Database; Operating system; Computer hardware; Programming language","score_opus":0.03028429458173998,"score_gpt":0.2794155407627558,"score_spread":0.24913124618101584,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3000259868","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":[0.8573511,0.0044788024,0.10966148,0.00047145243,0.00022556975,0.00030454007,0.0031839933,0.005790073,0.018532883],"genre_scores_gemma":[0.9017313,0.0011340826,0.08987116,0.00010005652,0.000064673775,0.00016831797,0.004607538,0.00038434053,0.0019385939],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9956512,0.00080738927,0.00041052574,0.00036845784,0.0023933197,0.00036919402],"domain_scores_gemma":[0.9886262,0.0053705294,0.0006638476,0.0019624878,0.0030419389,0.00033499594],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028105553,0.0007211218,0.0005120956,0.0018247718,0.0006485537,0.0016563579,0.0021807493,0.0007501585,0.0015653259],"category_scores_gemma":[0.015155916,0.0002737911,0.0003610157,0.004264726,0.00071022555,0.0045504593,0.0012293104,0.0006675829,0.0005267627],"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":[0.0021926907,0.0010941185,0.047044825,0.0021937229,0.00031659717,0.00049771636,0.00085294596,0.33276114,0.0577754,0.036621228,0.03138889,0.48726076],"study_design_scores_gemma":[0.00015328226,0.0022138748,0.009857176,0.00011808542,0.00010888141,0.00048692853,0.0006992775,0.888905,0.069326304,0.011256164,0.016800173,0.00007492142],"about_ca_topic_score_codex":0.0030138385,"about_ca_topic_score_gemma":0.0029819116,"teacher_disagreement_score":0.0030138385,"about_ca_system_score_codex":0.0010441064,"about_ca_system_score_gemma":0.0012495534,"threshold_uncertainty_score":0.014863789},"labels":[],"label_agreement":null},{"id":"W3011592252","doi":"10.14778/3236187.3236199","title":"Efficient construction of approximate ad-hoc ML models through materialization and reuse","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"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 analytical processing; Reuse; Dimension (graph theory); Cluster analysis; Variety (cybernetics); Data warehouse; Data mining; Construct (python library); Mixture model; Machine learning; Artificial intelligence; Mathematics; Programming language","score_opus":0.0167162247919843,"score_gpt":0.22005666246918465,"score_spread":0.20334043767720034,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011592252","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":[0.014802088,0.00014003973,0.9811152,0.00027888792,0.000013750535,0.000090259135,0.00014772874,0.0026009036,0.0008110786],"genre_scores_gemma":[0.23310429,0.00020913318,0.7622724,0.00023556163,0.00006519612,0.0002665839,0.0013107461,0.00087935943,0.0016567322],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99415344,0.0019901572,0.00044691676,0.0010427558,0.0018483808,0.00051832374],"domain_scores_gemma":[0.98116106,0.010481738,0.0010443123,0.005667471,0.0012619224,0.00038350967],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058200895,0.002239894,0.0026905995,0.0020721988,0.0013524585,0.004736695,0.004772128,0.0020105545,0.0034789182],"category_scores_gemma":[0.028147131,0.0019003085,0.0033305017,0.003361834,0.0023380548,0.00886617,0.0070944736,0.0038882683,0.0017005646],"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":[0.00018448447,0.00020675806,0.0025815852,0.00013380825,0.00011871364,0.00021941777,0.0003984245,0.8080758,0.0027494635,0.03882231,0.004327176,0.1421821],"study_design_scores_gemma":[0.000013230606,0.00001956592,0.0000602112,0.000004352838,0.000011157769,0.00002889935,0.00005403761,0.9783036,0.001054648,0.019617595,0.0008241465,0.000008479864],"about_ca_topic_score_codex":0.009648849,"about_ca_topic_score_gemma":0.012222269,"teacher_disagreement_score":0.009648849,"about_ca_system_score_codex":0.00276346,"about_ca_system_score_gemma":0.0038679263,"threshold_uncertainty_score":0.030779958},"labels":[],"label_agreement":null},{"id":"W3012550338","doi":"10.14778/3389133.3389134","title":"Dash","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":135,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Hash function; Scalability; Emulation; Hash table; Factor (programming language); Double hashing; Table (database)","score_opus":0.021384455536146138,"score_gpt":0.22421011762339768,"score_spread":0.20282566208725156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3012550338","genre_codex":"other","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":[0.082473874,0.0037307704,0.3182985,0.0025802003,0.0038407529,0.0011792864,0.03173808,0.23667431,0.3194843],"genre_scores_gemma":[0.39170808,0.0028778778,0.15731063,0.0022186977,0.0005380669,0.0011739144,0.073355995,0.012210905,0.35860583],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99881834,0.00011705487,0.0001237667,0.00022602887,0.0005601695,0.00015457522],"domain_scores_gemma":[0.99769706,0.0002493751,0.00008245679,0.00098754,0.000789742,0.00019384158],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009530944,0.0008403062,0.0006575534,0.00067083084,0.00079258747,0.0025825643,0.0026931867,0.0008178233,0.065725066],"category_scores_gemma":[0.0029242854,0.0005792958,0.00042446822,0.000874974,0.0006065201,0.0040317443,0.0036548802,0.0013823991,0.041047893],"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":[0.0027579034,0.00047654225,0.00647753,0.0013871564,0.00016469457,0.00048743997,0.0005473629,0.005680885,0.07638193,0.07103008,0.42314625,0.41146225],"study_design_scores_gemma":[0.00018721916,0.0004052103,0.0012793358,0.00007340479,0.00005320411,0.0005147141,0.00025382114,0.02867669,0.08011591,0.017597258,0.87074155,0.000101752805],"about_ca_topic_score_codex":0.0009339538,"about_ca_topic_score_gemma":0.0014149023,"teacher_disagreement_score":0.065725066,"about_ca_system_score_codex":0.00067384366,"about_ca_system_score_gemma":0.0011990971,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3012623189","doi":"10.14778/3407790.3407814","title":"Efficient oblivious database joins","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Joins; Cloud computing; Encryption; Sorting; Join (topology); Logarithm; Data structure; Computational complexity theory","score_opus":0.018108163619771928,"score_gpt":0.21146538441976953,"score_spread":0.1933572207999976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3012623189","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":[0.07248522,0.0003801025,0.90821916,0.000678727,0.000063222404,0.0002093174,0.00031701408,0.003583484,0.014063794],"genre_scores_gemma":[0.7214936,0.00049975986,0.26161098,0.0003484457,0.00009634837,0.00042159963,0.0006793338,0.00050546334,0.014344476],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99634486,0.0006459061,0.00032698616,0.00056416314,0.0016029509,0.0005151501],"domain_scores_gemma":[0.99366885,0.0018901043,0.00056075904,0.003393351,0.0003538353,0.00013306421],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019823953,0.0005273394,0.0009840357,0.0006031894,0.0014605925,0.0033290049,0.0026537434,0.0010159783,0.005212275],"category_scores_gemma":[0.0055096606,0.000718838,0.0008787439,0.0011175505,0.0018479212,0.0067079477,0.005476726,0.002379631,0.00228931],"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":[0.00154744,0.00035987824,0.002042537,0.00049653056,0.00008908953,0.00018786384,0.0008306731,0.07277702,0.08316534,0.6367787,0.007943929,0.19378096],"study_design_scores_gemma":[0.00025554025,0.00032728573,0.00067619886,0.00005692669,0.00009044326,0.000456692,0.00025166676,0.38171637,0.13462794,0.45252207,0.028944792,0.000074098214],"about_ca_topic_score_codex":0.0004540916,"about_ca_topic_score_gemma":0.00064098404,"teacher_disagreement_score":0.005212275,"about_ca_system_score_codex":0.0014796681,"about_ca_system_score_gemma":0.0020867991,"threshold_uncertainty_score":0.017436802},"labels":[],"label_agreement":null},{"id":"W3025940890","doi":"10.14778/3401960.3401966","title":"Approximate denial constraints","year":2020,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Overfitting; Computer science; Feature (linguistics); Focus (optics); Axiom; Mathematical optimization; Theoretical computer science; Algorithm; Mathematics; Artificial intelligence","score_opus":0.024586798940756546,"score_gpt":0.24463647034998354,"score_spread":0.220049671409227,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3025940890","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":[0.037859015,0.0014451302,0.948762,0.0021058891,0.00016518004,0.0002571396,0.002626486,0.0010510933,0.005728091],"genre_scores_gemma":[0.41623122,0.0011630188,0.5707599,0.0012708536,0.0003070173,0.00036941748,0.0054368353,0.00038580573,0.004075877],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9869435,0.0029191803,0.0013596063,0.0031609861,0.0047953213,0.0008215211],"domain_scores_gemma":[0.95969594,0.026939133,0.003004768,0.006488798,0.0033657728,0.0005056863],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0060097137,0.0011027172,0.0021454582,0.002379176,0.001463537,0.004509238,0.0038208754,0.0024173716,0.00639202],"category_scores_gemma":[0.055442676,0.0009390979,0.0020229619,0.0044105905,0.0021537018,0.009873597,0.0035561111,0.00368178,0.0011327782],"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":[0.0008479169,0.00022828774,0.022496592,0.0016673758,0.000497687,0.0012940944,0.0010134522,0.1923077,0.0065413895,0.34489587,0.023378069,0.40483156],"study_design_scores_gemma":[0.00006294383,0.00008681204,0.0026219708,0.00017903466,0.00010918128,0.002258302,0.0005325764,0.5511032,0.008584257,0.40894493,0.025447346,0.00006947845],"about_ca_topic_score_codex":0.003562211,"about_ca_topic_score_gemma":0.0033696732,"teacher_disagreement_score":0.00639202,"about_ca_system_score_codex":0.0014511807,"about_ca_system_score_gemma":0.0023593013,"threshold_uncertainty_score":0.031782806},"labels":[],"label_agreement":null},{"id":"W3038067601","doi":"10.14778/3397230.3397239","title":"Scalable, near-zero loss disaster recovery for distributed data stores","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Backup; Computer science; Failover; Data loss; Scalability; Distributed computing; Replication (statistics); Fault tolerance; Backup software; Eventual consistency; Computer network; Disaster recovery; Data integrity; Data recovery; Asynchronous communication; Data consistency; Database; Consistency model; Operating system","score_opus":0.036865755455791,"score_gpt":0.2497641672253746,"score_spread":0.21289841176958357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3038067601","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":[0.3585028,0.0019403718,0.54779613,0.0012040278,0.00035131891,0.0008788985,0.0010506185,0.07921042,0.009065353],"genre_scores_gemma":[0.86950433,0.00032352863,0.12318856,0.00030555655,0.00006755142,0.00021421332,0.0010617407,0.000608224,0.004726274],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99860173,0.00016439882,0.00012283337,0.00026439436,0.000634439,0.00021217205],"domain_scores_gemma":[0.9963737,0.00058458536,0.0003908512,0.0017230073,0.0006384996,0.0002892036],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013274482,0.000752242,0.0006980517,0.000957335,0.0011440776,0.001496109,0.0036765,0.0006481403,0.0025427043],"category_scores_gemma":[0.0037793377,0.0006191665,0.00038434519,0.0010570388,0.000958694,0.004152371,0.003993035,0.0014421914,0.0010802535],"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":[0.0066575767,0.00171581,0.017223975,0.0014812769,0.00034157824,0.0016735293,0.0020211255,0.14793673,0.21788512,0.017911045,0.056322716,0.52882946],"study_design_scores_gemma":[0.0009788765,0.0023515855,0.005404462,0.00012387613,0.00019375925,0.0014200992,0.0012887327,0.77941066,0.14912583,0.014467758,0.04497017,0.00026421616],"about_ca_topic_score_codex":0.0023570354,"about_ca_topic_score_gemma":0.0032032323,"teacher_disagreement_score":0.0036765,"about_ca_system_score_codex":0.0008243459,"about_ca_system_score_gemma":0.0017086066,"threshold_uncertainty_score":0.008506179},"labels":[],"label_agreement":null},{"id":"W3082162291","doi":"10.14778/3407790.3407810","title":"<i>Pytheas</i>","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Information retrieval; Metadata; Table (database); File format; Data extraction; Data mining; World Wide Web; Database","score_opus":0.014954288133886124,"score_gpt":0.18961740079404488,"score_spread":0.17466311266015874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3082162291","genre_codex":"dataset","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":[0.007709616,0.0006335343,0.093025595,0.0023396006,0.0007328788,0.0006765134,0.44539255,0.42260516,0.026884604],"genre_scores_gemma":[0.041405845,0.0007234283,0.18481089,0.001402944,0.000226226,0.00094796956,0.7138702,0.035723258,0.020889195],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975503,0.0001925856,0.00023640976,0.0005374386,0.0012362697,0.00024701236],"domain_scores_gemma":[0.9831871,0.0037034417,0.0012174617,0.005718846,0.00540029,0.0007728932],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025992624,0.0016612809,0.00082611,0.007179614,0.0014173737,0.0040207724,0.0030476595,0.0013481894,0.050810087],"category_scores_gemma":[0.02578024,0.0009935838,0.0011780619,0.010749899,0.0015924514,0.0063315374,0.004091737,0.0017538213,0.04787798],"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":[0.0002907939,0.00004070865,0.0055712773,0.0005753144,0.000046450674,0.00019675629,0.00043950733,0.000829757,0.004446347,0.0043130973,0.8754288,0.1078212],"study_design_scores_gemma":[0.000045904213,0.000042722924,0.010077972,0.0002071694,0.000019884288,0.00035831044,0.00028124164,0.010022224,0.018866425,0.00504396,0.95491487,0.0001192311],"about_ca_topic_score_codex":0.10816597,"about_ca_topic_score_gemma":0.12076589,"teacher_disagreement_score":0.10816597,"about_ca_system_score_codex":0.0028784056,"about_ca_system_score_gemma":0.0058220155,"threshold_uncertainty_score":0.21507275},"labels":[],"label_agreement":null},{"id":"W3084763998","doi":"10.14778/3407790.3407854","title":"SAQE","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":true,"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; Differential privacy; Query optimization; Query expansion; Query plan; Overhead (engineering); Data mining; Private information retrieval; Information privacy; Leverage (statistics); Information retrieval; Sargable; Web search query; Computer security; Search engine","score_opus":0.033158732127586174,"score_gpt":0.23953265542545057,"score_spread":0.20637392329786441,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3084763998","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":[0.015902573,0.00047613998,0.92061675,0.001386097,0.000253752,0.0007238193,0.0023954804,0.037725244,0.020520292],"genre_scores_gemma":[0.33629704,0.00056763605,0.6122758,0.0014382936,0.0001511073,0.0010216805,0.009493716,0.0049274648,0.03382724],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9941375,0.0012848253,0.0004619087,0.00093265023,0.0025574218,0.0006255564],"domain_scores_gemma":[0.9924924,0.0014127295,0.00027061862,0.0040857615,0.0015141668,0.00022440465],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005894194,0.0009785211,0.0010197199,0.00075576635,0.0014073076,0.0036952167,0.0039055005,0.0014985907,0.021115389],"category_scores_gemma":[0.012632559,0.0008160283,0.0013172821,0.0012562512,0.001400731,0.006713987,0.0058164042,0.0028280078,0.008543978],"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":[0.002227781,0.0006626168,0.0068695853,0.00087776606,0.0002767564,0.00041373464,0.0012596798,0.060097996,0.030348599,0.29601488,0.18481646,0.41613412],"study_design_scores_gemma":[0.0002435244,0.00040410488,0.0014958108,0.000083710394,0.000058784717,0.00061275274,0.00044602668,0.6017722,0.03249227,0.12847994,0.23379108,0.00011978536],"about_ca_topic_score_codex":0.0027603274,"about_ca_topic_score_gemma":0.0029409467,"teacher_disagreement_score":0.021115389,"about_ca_system_score_codex":0.0013298823,"about_ca_system_score_gemma":0.0027287588,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3085364681","doi":"10.14778/3415478.3415562","title":"Data collection and quality challenges for deep learning","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":145,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Computer science; Machine learning; Artificial intelligence; Feature engineering; Deep learning; Data collection; Big data; Data science; Software; Data mining","score_opus":0.12845824892835148,"score_gpt":0.3110977474383105,"score_spread":0.182639498509959,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085364681","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":[0.019132115,0.011416847,0.90780926,0.03980681,0.0017412663,0.001254668,0.008394149,0.0037638068,0.006681067],"genre_scores_gemma":[0.1118938,0.009625044,0.8412076,0.008286663,0.0019942669,0.0037110525,0.016339885,0.0018474734,0.005094176],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.93783313,0.02391522,0.005539044,0.005116389,0.026424298,0.0011718151],"domain_scores_gemma":[0.7931804,0.10005246,0.009855264,0.039908245,0.053160097,0.0038434968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.057101943,0.0011265101,0.0020323908,0.0041692522,0.0028305373,0.009194133,0.0050856415,0.00294739,0.003655462],"category_scores_gemma":[0.20775719,0.00137614,0.0015453898,0.007496398,0.0040999604,0.009662429,0.008774134,0.008299758,0.002638957],"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":[0.0004928268,0.00043010004,0.029330969,0.0037405018,0.0004032459,0.00043819757,0.0020491257,0.020050593,0.0077145207,0.08695638,0.11575163,0.73264194],"study_design_scores_gemma":[0.00016101878,0.0004631849,0.021110073,0.003583948,0.00019362797,0.0010663788,0.0030581516,0.102932036,0.021288935,0.39958695,0.44629467,0.00026106334],"about_ca_topic_score_codex":0.0054403045,"about_ca_topic_score_gemma":0.006050148,"teacher_disagreement_score":0.057101943,"about_ca_system_score_codex":0.0038927752,"about_ca_system_score_gemma":0.008029495,"threshold_uncertainty_score":0.3019876},"labels":[],"label_agreement":null},{"id":"W3085986371","doi":"10.14778/3407790.3407845","title":"Do the best cloud configurations grow on trees?","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Cloud computing; Bayesian optimization; Computer science; Black box; Optimization problem; Mathematical optimization; Data mining; Algorithm; Machine learning; Artificial intelligence; Mathematics; Operating system","score_opus":0.025114293130988532,"score_gpt":0.22797825888268863,"score_spread":0.2028639657517001,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085986371","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":[0.47344568,0.010887665,0.4696642,0.01084117,0.0005983709,0.000391274,0.0029661704,0.004226051,0.026979495],"genre_scores_gemma":[0.8645487,0.0019042719,0.12701164,0.0012767668,0.00025113064,0.00014980344,0.0021520623,0.001053517,0.001652073],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99679226,0.0013020093,0.00014208187,0.00085143506,0.00048693383,0.0004252239],"domain_scores_gemma":[0.9789059,0.01582626,0.0017867802,0.0015786899,0.0013693285,0.0005330233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052538374,0.0011811575,0.0020886848,0.0019432905,0.0009459011,0.0027063,0.0016097755,0.0026917374,0.0036800855],"category_scores_gemma":[0.044599153,0.0011379182,0.0010199866,0.002103177,0.0012259334,0.0075384527,0.001038405,0.001711784,0.0025287587],"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":[0.00086872384,0.00039472105,0.027712548,0.0005044839,0.0002811934,0.00017959782,0.00040871985,0.605276,0.0023699023,0.019729653,0.034898557,0.30737576],"study_design_scores_gemma":[0.00008504822,0.00012400551,0.0036323888,0.00011868073,0.00006798357,0.00014160402,0.00029243456,0.92612725,0.0010517404,0.06483159,0.0034920636,0.00003505888],"about_ca_topic_score_codex":0.0041929795,"about_ca_topic_score_gemma":0.008063666,"teacher_disagreement_score":0.0052538374,"about_ca_system_score_codex":0.00092273386,"about_ca_system_score_gemma":0.0017778617,"threshold_uncertainty_score":0.027785301},"labels":[],"label_agreement":null},{"id":"W3086093300","doi":"10.14778/3415478.3415483","title":"PiBench online","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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; Benchmarking; Upload; Emulation; Key (lock); Implementation; Operating system; Dram; Interface (matter); Spec#; Embedded system; Software engineering; Computer hardware","score_opus":0.025558968550096615,"score_gpt":0.23712624469475133,"score_spread":0.2115672761446547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3086093300","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":[0.0021485968,0.0005074844,0.07434472,0.0010530225,0.00086259434,0.0004982334,0.054083217,0.777924,0.088578194],"genre_scores_gemma":[0.04442743,0.0010668571,0.09899297,0.0029743717,0.00067621534,0.0025828434,0.3702252,0.30536932,0.17368479],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99712914,0.00037593124,0.00018311695,0.0005861651,0.0013656192,0.0003599999],"domain_scores_gemma":[0.992149,0.0015877825,0.00029872652,0.0031383622,0.0020508582,0.000775242],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033923408,0.0027703412,0.0014960741,0.0027377713,0.0010647968,0.0060713068,0.0074475403,0.0020464074,0.36167195],"category_scores_gemma":[0.015527444,0.0016196836,0.0010870317,0.0025362002,0.00076368777,0.009302702,0.007621906,0.0035029708,0.2620883],"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":[0.000560103,0.00016476776,0.0008852457,0.0005072301,0.000049303726,0.000143076,0.00015955757,0.00060016307,0.0032822543,0.0075559197,0.91163486,0.074457474],"study_design_scores_gemma":[0.00021078087,0.00011101684,0.001157806,0.00012796313,0.000025429135,0.00019856816,0.00008453444,0.009175059,0.0096632885,0.011843203,0.96732336,0.00007905692],"about_ca_topic_score_codex":0.002424929,"about_ca_topic_score_gemma":0.0028427434,"teacher_disagreement_score":0.36167195,"about_ca_system_score_codex":0.0014953214,"about_ca_system_score_gemma":0.0025856136,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3086099593","doi":"10.14778/3407790.3407856","title":"Sentinel","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Software System Performance and Reliability","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Debugging; Computer science; Variety (cybernetics); Server; Distributed computing; Web application; Operating system; Database; Embedded system; Artificial intelligence","score_opus":0.013665081243024367,"score_gpt":0.20227277804738522,"score_spread":0.18860769680436085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3086099593","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":[0.034978256,0.0015098098,0.60010725,0.0017590262,0.00077128905,0.00057722337,0.011079112,0.25656646,0.092651494],"genre_scores_gemma":[0.28848207,0.002131729,0.5686124,0.0018770883,0.00032572617,0.00069255504,0.036165617,0.024295423,0.07741738],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971003,0.00046185136,0.00022458835,0.00054529373,0.001439688,0.00022834951],"domain_scores_gemma":[0.9943053,0.0018798,0.0005490454,0.0015313311,0.001481275,0.0002531909],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029751274,0.00097553636,0.0006672891,0.0019508898,0.0007993432,0.0029216658,0.0019926613,0.00081690465,0.02308565],"category_scores_gemma":[0.008129241,0.0007705248,0.00084066065,0.0012687824,0.0007561168,0.005219299,0.0020284392,0.0013588597,0.012533803],"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":[0.00080711197,0.00020364321,0.014624331,0.0014580719,0.00015795781,0.00058079977,0.0013462865,0.024187764,0.021691654,0.10900448,0.26633,0.5596079],"study_design_scores_gemma":[0.00013919959,0.00052085234,0.0062353555,0.00034604227,0.00012402554,0.0012530844,0.0005161784,0.18784328,0.022160988,0.05111838,0.7295914,0.00015123568],"about_ca_topic_score_codex":0.0033809028,"about_ca_topic_score_gemma":0.0059909145,"teacher_disagreement_score":0.02308565,"about_ca_system_score_codex":0.001112379,"about_ca_system_score_gemma":0.0025743048,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3086212533","doi":"10.14778/3407790.3407861","title":"Suffix rank","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; University of Toronto","funders":"","keywords":"Substring; Suffix array; Suffix; Generalized suffix tree; Computer science; Suffix tree; Compressed suffix array; Scalability; Rank (graph theory); Extension (predicate logic); Liveness; Algorithm; Context (archaeology); Parallelizable manifold; Auxiliary memory; Theoretical computer science; Data structure; Mathematics; Combinatorics; Database","score_opus":0.01641272735555164,"score_gpt":0.20674939338029258,"score_spread":0.19033666602474095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3086212533","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":[0.08433167,0.0024758205,0.8449347,0.002205421,0.0004948667,0.00047897387,0.005492691,0.0145609435,0.045024924],"genre_scores_gemma":[0.27593225,0.001334964,0.6801626,0.0007248614,0.0004983413,0.00038977177,0.009222846,0.001395304,0.030338997],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979773,0.00023829978,0.0001960593,0.000489657,0.0008640319,0.00023465673],"domain_scores_gemma":[0.99469006,0.0019883073,0.00036593692,0.0016475677,0.0010976031,0.00021054874],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001176269,0.00092675735,0.0013555036,0.0017609148,0.0010974888,0.0026846894,0.0021150229,0.00117613,0.018303366],"category_scores_gemma":[0.010887719,0.00037677595,0.0006677226,0.0040359553,0.0008124274,0.0060689044,0.0023834466,0.0012739374,0.009032955],"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":[0.0006139627,0.00022789944,0.0032059,0.0006130693,0.00006407866,0.00026909873,0.00024148254,0.045197472,0.013310955,0.0635201,0.060971156,0.8117649],"study_design_scores_gemma":[0.00015452038,0.0006445854,0.001104409,0.00012023241,0.00007585423,0.0013066281,0.00046067475,0.68058234,0.04814653,0.17528723,0.09205285,0.00006418587],"about_ca_topic_score_codex":0.0014725849,"about_ca_topic_score_gemma":0.002310262,"teacher_disagreement_score":0.018303366,"about_ca_system_score_codex":0.00081078894,"about_ca_system_score_gemma":0.0017932934,"threshold_uncertainty_score":0.06123084},"labels":[],"label_agreement":null},{"id":"W3086284023","doi":"10.14778/3415478.3415500","title":"ActiveDeeper","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Deep Web; Computer science; World Wide Web; State (computer science); Database; The Internet; Programming language","score_opus":0.017852573838847422,"score_gpt":0.1999878516851466,"score_spread":0.18213527784629918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3086284023","genre_codex":"software","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":[0.017696014,0.001270787,0.37481964,0.0008266026,0.00043928804,0.00046246315,0.043219727,0.5401314,0.02113399],"genre_scores_gemma":[0.2335577,0.0014104302,0.49121177,0.0031785336,0.00016240583,0.0011960944,0.18970205,0.031453565,0.04812748],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99911577,0.00008473951,0.00006653577,0.0003573285,0.00032059662,0.000054912824],"domain_scores_gemma":[0.9985285,0.00043919976,0.000059975464,0.00062880036,0.0002369462,0.00010649368],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012390758,0.0011410166,0.00083720684,0.0015613448,0.00048286052,0.0017078512,0.003067923,0.00092957757,0.019431055],"category_scores_gemma":[0.0041572624,0.00091471866,0.00094472343,0.001311914,0.0004287564,0.004761297,0.003721675,0.0016562996,0.01288255],"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":[0.0014745309,0.00062180473,0.007680128,0.0011486533,0.0004655102,0.0005036025,0.0003402035,0.010377942,0.016701635,0.0138461515,0.44002163,0.5068182],"study_design_scores_gemma":[0.00024199596,0.00030527302,0.004237163,0.00015050582,0.00013935904,0.0010377258,0.00014818742,0.29411298,0.05979939,0.043659747,0.5959935,0.0001740897],"about_ca_topic_score_codex":0.0026999793,"about_ca_topic_score_gemma":0.00550953,"teacher_disagreement_score":0.019431055,"about_ca_system_score_codex":0.000546101,"about_ca_system_score_gemma":0.0008770599,"threshold_uncertainty_score":0.065003335},"labels":[],"label_agreement":null},{"id":"W3086973390","doi":"10.14778/3407790.3407858","title":"ATHENA++","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"Computer science; Benchmark (surveying); SQL; Query language; Set (abstract data type); Ontology; Nesting (process); Information retrieval; Programming language; Database","score_opus":0.02196695787347267,"score_gpt":0.2013119211773401,"score_spread":0.17934496330386743,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3086973390","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":[0.01528395,0.0040659504,0.13688295,0.0019896338,0.0013467523,0.0011538768,0.07498315,0.5763773,0.18791655],"genre_scores_gemma":[0.09201,0.0030871811,0.31934872,0.0049300375,0.00048756428,0.0022893092,0.36436057,0.06311218,0.15037444],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9970306,0.00039749115,0.00031513246,0.000873029,0.0011697776,0.00021398404],"domain_scores_gemma":[0.99711907,0.0006264152,0.00014444433,0.0012656469,0.0006358124,0.00020860574],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015518425,0.0021075136,0.00136211,0.0022696112,0.0009962237,0.0044534444,0.00505697,0.0016990846,0.0873169],"category_scores_gemma":[0.0077642747,0.0013739818,0.0017210968,0.0022971195,0.0007891611,0.005833715,0.005592319,0.0026945446,0.1187564],"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":[0.0011327858,0.00024863816,0.0019826782,0.0014260635,0.00014785609,0.00044632197,0.00028616548,0.0028438242,0.009013368,0.018718738,0.7188429,0.2449107],"study_design_scores_gemma":[0.00024798786,0.00016693986,0.0015737526,0.00016125248,0.00005399438,0.0006902448,0.00007374149,0.015274426,0.0117131015,0.014830957,0.9550999,0.00011365261],"about_ca_topic_score_codex":0.004418154,"about_ca_topic_score_gemma":0.004405217,"teacher_disagreement_score":0.0873169,"about_ca_system_score_codex":0.0010518609,"about_ca_system_score_gemma":0.0017062129,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3094958164","doi":"10.14778/3424573.3424575","title":"An analysis of concurrency control protocols for in-memory databases with CCBench","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nautilus Environmental","funders":"","keywords":"Computer science; Workload; Garbage collection; Parallel computing; Thread (computing); Scalability; Cache; Concurrency control; Concurrency; Distributed computing; Cache coherence; Transaction processing; Database transaction; Operating system; CPU cache; Database; Programming language; Garbage; Cache algorithms","score_opus":0.029335185320784087,"score_gpt":0.29091004507903095,"score_spread":0.26157485975824685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3094958164","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":[0.3539605,0.0010410574,0.6290871,0.00036292922,0.0000841532,0.000512628,0.00051543844,0.00681316,0.0076230643],"genre_scores_gemma":[0.8345405,0.00033605748,0.16083603,0.00012309688,0.000046791778,0.00036350673,0.0009950148,0.0006108356,0.002148316],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9950983,0.0013010543,0.00033491632,0.0005321576,0.002145702,0.0005879195],"domain_scores_gemma":[0.9891768,0.0053388467,0.0006914116,0.0027802016,0.0017264788,0.0002862949],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043551764,0.0014120525,0.00078306487,0.0018896501,0.0009902522,0.0024107804,0.0021103625,0.00073693116,0.0019793408],"category_scores_gemma":[0.010534721,0.00075505406,0.0008790956,0.0016711417,0.0012397285,0.0046647927,0.0013173065,0.0016186428,0.00023765882],"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":[0.003406753,0.0016869782,0.043912753,0.000986339,0.00038120325,0.0008081991,0.0018178265,0.29897195,0.13635384,0.19416122,0.0064000054,0.3111129],"study_design_scores_gemma":[0.00009139623,0.0004569258,0.0025041352,0.00004107382,0.00011593178,0.0001903107,0.0001923659,0.9082106,0.06871248,0.012926746,0.006504915,0.00005306995],"about_ca_topic_score_codex":0.0058921915,"about_ca_topic_score_gemma":0.0035779218,"teacher_disagreement_score":0.0058921915,"about_ca_system_score_codex":0.0022472036,"about_ca_system_score_gemma":0.003186169,"threshold_uncertainty_score":0.023032665},"labels":[],"label_agreement":null},{"id":"W3095172747","doi":"10.14778/3424573.3424578","title":"MorphoSys","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"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; Replication (statistics); Distributed computing; Workload; Distributed database; Operating system","score_opus":0.01438667055517795,"score_gpt":0.19289990444754224,"score_spread":0.1785132338923643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3095172747","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":[0.07284612,0.0032014155,0.5184378,0.0010236049,0.0007729928,0.0011793866,0.011618843,0.30751243,0.08340736],"genre_scores_gemma":[0.46655688,0.0023821574,0.40705785,0.0011013476,0.00017514845,0.0012553949,0.030773586,0.019550864,0.07114669],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99946016,0.00006341852,0.000042020823,0.000118458796,0.00025919345,0.000056804747],"domain_scores_gemma":[0.99923265,0.00020683758,0.000082889695,0.00025280018,0.00016295098,0.00006186827],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006889353,0.0008337009,0.00044540965,0.00059936615,0.00036756048,0.0012156264,0.0023910564,0.00056925806,0.029420879],"category_scores_gemma":[0.0025990533,0.0005218086,0.00054887537,0.00048366396,0.0004699433,0.0017862642,0.0017462814,0.0009814654,0.0066523496],"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":[0.0017705585,0.0004134615,0.0062587224,0.0016919498,0.00019258211,0.00048807703,0.00042830073,0.062084384,0.06330409,0.058707345,0.2706497,0.5340107],"study_design_scores_gemma":[0.00045029254,0.0007085329,0.003694734,0.0001328175,0.00010793302,0.00091118395,0.00014351467,0.3754812,0.06149771,0.026930196,0.5298258,0.00011612467],"about_ca_topic_score_codex":0.0016930971,"about_ca_topic_score_gemma":0.0025874046,"teacher_disagreement_score":0.029420879,"about_ca_system_score_codex":0.0006108598,"about_ca_system_score_gemma":0.0010036908,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3096891463","doi":"10.14778/3430915.3430932","title":"CoroBase","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"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; Pointer (user interface); Database transaction; Asynchronous communication; Database; Transaction processing system; Software; Operating system; Transaction processing; Online transaction processing; Distributed computing; Computer network; Computer hardware","score_opus":0.014720176867610645,"score_gpt":0.19351069959949013,"score_spread":0.17879052273187948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3096891463","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":[0.027918983,0.0040511447,0.13497849,0.0018650804,0.0020181516,0.0010741531,0.018218137,0.5451409,0.264735],"genre_scores_gemma":[0.24261737,0.004071204,0.18072727,0.0033881564,0.00066489395,0.0014615243,0.095742434,0.12926611,0.34206104],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9978896,0.00016151316,0.00016084466,0.0005788409,0.000893207,0.00031604228],"domain_scores_gemma":[0.99585235,0.00058827666,0.00019308947,0.0016806569,0.0012386993,0.00044705035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018233766,0.0021429248,0.0012421345,0.0013404519,0.0010698457,0.0037304636,0.0054985834,0.0015454207,0.12369068],"category_scores_gemma":[0.0077555627,0.0011257537,0.0010230663,0.0014566109,0.0009903281,0.0066326056,0.0055755773,0.002863758,0.075562336],"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":[0.0051343227,0.0005515378,0.003264785,0.0022309478,0.00020726719,0.00070149294,0.00070708344,0.0033075134,0.047762297,0.047898926,0.61248976,0.2757442],"study_design_scores_gemma":[0.00057098904,0.00033975867,0.0010935575,0.0001856536,0.000059412232,0.0006308969,0.00011338813,0.01284989,0.024632823,0.013542721,0.945857,0.00012382986],"about_ca_topic_score_codex":0.0036130268,"about_ca_topic_score_gemma":0.0030448574,"teacher_disagreement_score":0.12369068,"about_ca_system_score_codex":0.0011359699,"about_ca_system_score_gemma":0.0023644755,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3111141572","doi":"10.14778/3461535.3461552","title":"Are we ready for learned cardinality estimation?","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":94,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Cardinality (data modeling); Inference; Focus (optics); Ask price; Domain (mathematical analysis); Workload; Data modeling","score_opus":0.07254020723674541,"score_gpt":0.3182425928154698,"score_spread":0.24570238557872437,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111141572","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":[0.0552291,0.0022674068,0.9302608,0.005571324,0.00022214995,0.00013366235,0.00066064514,0.0037181743,0.0019368448],"genre_scores_gemma":[0.51565546,0.0012035336,0.47727126,0.0015539164,0.00032224803,0.00019644896,0.0017512414,0.0007182777,0.0013276407],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99075186,0.004175003,0.0005388198,0.001923136,0.0020291833,0.0005819459],"domain_scores_gemma":[0.95412374,0.024429204,0.0030440444,0.013396651,0.003982592,0.0010237575],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010296117,0.0014130857,0.002231424,0.0009575972,0.00067134085,0.0040491675,0.0044897636,0.0021815235,0.0019685868],"category_scores_gemma":[0.070840836,0.0012063334,0.0011529257,0.0016133662,0.0020303146,0.017070213,0.0033729798,0.0052531376,0.0012415331],"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":[0.0006511849,0.00044301245,0.025706664,0.0006423184,0.0003528586,0.00017248809,0.0006470481,0.5115672,0.0056395573,0.039612167,0.019975323,0.39459026],"study_design_scores_gemma":[0.00003348662,0.0000715236,0.001021306,0.00005750674,0.000031941756,0.00008474244,0.0001983291,0.9613079,0.002418015,0.03167138,0.0030735321,0.000030409592],"about_ca_topic_score_codex":0.0053753257,"about_ca_topic_score_gemma":0.0068178466,"teacher_disagreement_score":0.010296117,"about_ca_system_score_codex":0.002344457,"about_ca_system_score_gemma":0.0031298331,"threshold_uncertainty_score":0.054451764},"labels":[],"label_agreement":null},{"id":"W3112994038","doi":"10.14778/3436905.3436920","title":"Maximizing social welfare in a competitive diffusion model","year":2020,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Mobile Crowdsensing and Crowdsourcing","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Competition (biology); Viral marketing; Computer science; Incentive; Maximization; Quality (philosophy); Social Welfare; Microeconomics; Focus (optics); Competitive equilibrium; Social network (sociolinguistics); Utility maximization; Utility maximization problem; Mathematical economics; Economics; Social media","score_opus":0.02382788169935284,"score_gpt":0.23324999598651466,"score_spread":0.20942211428716181,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3112994038","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":[0.06469744,0.0012419455,0.9070391,0.0029187603,0.00013546002,0.00030485017,0.0009965355,0.0005349213,0.022131022],"genre_scores_gemma":[0.81292695,0.0016544972,0.1586838,0.0008438052,0.00025391573,0.0007743915,0.0010414149,0.0002974345,0.023523811],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99832946,0.00070848037,0.000053166794,0.00035941004,0.00024371673,0.00030574552],"domain_scores_gemma":[0.99462765,0.004195867,0.00028472854,0.00018676887,0.00036459952,0.00034039217],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024671694,0.0021543487,0.0026525208,0.0014262736,0.0011423363,0.0026045463,0.0032950444,0.0032400123,0.0060638785],"category_scores_gemma":[0.008952772,0.000876846,0.001398998,0.0018366762,0.001705349,0.0029807081,0.0021479547,0.0023988765,0.0010431638],"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":[0.00013307374,0.00011909841,0.0006460025,0.00019449263,0.0000644717,0.0002777489,0.00018006355,0.8603491,0.00089913025,0.11539645,0.006108918,0.015631337],"study_design_scores_gemma":[0.000019926123,0.000013977593,0.000054433785,0.000008862899,0.000008900269,0.000021971353,0.000021432319,0.9663056,0.00009131275,0.032631323,0.00081574294,0.0000066018424],"about_ca_topic_score_codex":0.011689098,"about_ca_topic_score_gemma":0.0084362915,"teacher_disagreement_score":0.011689098,"about_ca_system_score_codex":0.0037306673,"about_ca_system_score_gemma":0.001849467,"threshold_uncertainty_score":0.027068019},"labels":[],"label_agreement":null},{"id":"W3116457585","doi":"10.14778/3184470.3184473","title":"Distributed evaluation of subgraph queries using worst-case optimal low-memory dataflows","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Dataflow; Computer science; Computation; Memory footprint; Joins; Massively parallel; Graph; Parallel computing; Theoretical computer science; Distributed computing; Algorithm","score_opus":0.03065908720145722,"score_gpt":0.27186278200001407,"score_spread":0.24120369479855686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3116457585","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":[0.29222766,0.00047908947,0.69869614,0.0010701901,0.00008152809,0.00018079224,0.00023022329,0.0035749497,0.0034594247],"genre_scores_gemma":[0.88404226,0.000066141,0.11472083,0.0001132869,0.00004586818,0.00009535775,0.00018812841,0.0001899793,0.00053824205],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9945194,0.0014629398,0.00032601503,0.0014855472,0.0013790103,0.0008270432],"domain_scores_gemma":[0.9841303,0.009797818,0.0011652568,0.003268433,0.0010741879,0.0005639633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051711127,0.0010677436,0.001618126,0.00097423175,0.0012556537,0.0027694218,0.0030585013,0.0013887198,0.0015918564],"category_scores_gemma":[0.021312904,0.0006748031,0.0008730162,0.0013365141,0.0029224132,0.005713324,0.002883933,0.0012594872,0.00030526306],"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":[0.0013296445,0.0002738374,0.0048997803,0.00014033649,0.00011458104,0.00018183282,0.00031675166,0.87225056,0.01727868,0.023750922,0.0024786885,0.07698438],"study_design_scores_gemma":[0.000032656702,0.00005097751,0.00016216401,0.0000034652207,0.000012611875,0.000023362965,0.00004155096,0.9792029,0.004240832,0.01601687,0.00020643981,0.0000061358073],"about_ca_topic_score_codex":0.0037516118,"about_ca_topic_score_gemma":0.0044871,"teacher_disagreement_score":0.0051711127,"about_ca_system_score_codex":0.0026276845,"about_ca_system_score_gemma":0.0026788306,"threshold_uncertainty_score":0.027347744},"labels":[],"label_agreement":null},{"id":"W3129744650","doi":"10.14778/3436905.3436907","title":"Astrid","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Substring; Computer science; Suffix; String (physics); Embedding; String searching algorithm; Suffix tree; Prefix; Theoretical computer science; Artificial intelligence; Algorithm; Pattern matching; Data structure; Mathematics","score_opus":0.23727257312021974,"score_gpt":0.3685803821593431,"score_spread":0.13130780903912334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3129744650","genre_codex":"other","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":[0.0059815133,0.0040311636,0.13743742,0.005060445,0.00357266,0.0008026523,0.07464096,0.26682124,0.5016518],"genre_scores_gemma":[0.069856025,0.0041447864,0.16708837,0.0077056787,0.0008750779,0.0010953606,0.26209033,0.05130776,0.43583658],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99788827,0.00024406456,0.00017685551,0.00049513136,0.00092730374,0.00026837786],"domain_scores_gemma":[0.9974376,0.0003924532,0.00013624957,0.0010959931,0.00070787675,0.00022984012],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012788738,0.0020897575,0.001457095,0.0017859176,0.0013945942,0.004763247,0.0043104617,0.0028900593,0.30434948],"category_scores_gemma":[0.0064688907,0.0010830284,0.0014613513,0.0019653619,0.00068538194,0.006202341,0.0051799067,0.002845559,0.34184963],"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":[0.0003834506,0.000094398674,0.00087080617,0.0006248831,0.000053724452,0.00022503323,0.00010807308,0.002163166,0.0027132763,0.01928807,0.8245985,0.14887668],"study_design_scores_gemma":[0.00007528558,0.000053655272,0.0003382589,0.00009183629,0.000024481797,0.00031755926,0.000073294985,0.007876948,0.0032998447,0.014252793,0.9735537,0.00004218449],"about_ca_topic_score_codex":0.0033777007,"about_ca_topic_score_gemma":0.004702401,"teacher_disagreement_score":0.30434948,"about_ca_system_score_codex":0.0013147659,"about_ca_system_score_gemma":0.0021034682,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3133302368","doi":"10.14778/3436905.3436916","title":"Scalable mining of maximal quasi-cliques","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"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; Clique; Load balancing (electrical power); Speedup; Vertex (graph theory); Graph; Theoretical computer science; Parallel computing; Combinatorics; Mathematics; Database","score_opus":0.02458840115704153,"score_gpt":0.23857404697814444,"score_spread":0.2139856458211029,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133302368","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":[0.22497405,0.000589487,0.7633046,0.00091235474,0.00008724962,0.00035100785,0.0025674126,0.004484023,0.0027299018],"genre_scores_gemma":[0.48519114,0.00022942225,0.50649726,0.0002146251,0.000059577076,0.00023403087,0.0051910067,0.00039476142,0.0019882452],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985026,0.00032633336,0.000078055746,0.0005943389,0.00030178457,0.00019691675],"domain_scores_gemma":[0.9961527,0.0017356564,0.00038903323,0.00081879913,0.0006338066,0.00027005145],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013874948,0.0008314549,0.00111341,0.0016421512,0.001132931,0.0015176501,0.0026242076,0.0008921878,0.0014818142],"category_scores_gemma":[0.0072971424,0.00081004336,0.00135172,0.0018038496,0.00068753853,0.0028809907,0.0018838788,0.00075633696,0.0005024457],"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":[0.0012175001,0.00050219556,0.03424282,0.0009978854,0.0005687472,0.0010598837,0.0009304655,0.5326649,0.049361095,0.04031944,0.027024895,0.31111014],"study_design_scores_gemma":[0.00005668528,0.00003601593,0.0012346809,0.0000119741335,0.000025649542,0.00011965601,0.0001331993,0.9720205,0.0034748157,0.020879641,0.0019963773,0.000010904549],"about_ca_topic_score_codex":0.0056494977,"about_ca_topic_score_gemma":0.013920185,"teacher_disagreement_score":0.0056494977,"about_ca_system_score_codex":0.0008097844,"about_ca_system_score_gemma":0.0023132963,"threshold_uncertainty_score":0.01123327},"labels":[],"label_agreement":null},{"id":"W3138240437","doi":"10.14778/3407790.3407806","title":"Knowledge translation","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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; Heuristics; SPARQL; Set (abstract data type); Ranking (information retrieval); Information retrieval; Translation (biology); Rank (graph theory); Semantic mapping; Natural language processing; Artificial intelligence; Data mining; Semantic Web; RDF; Programming language; Mathematics","score_opus":0.08498789382318941,"score_gpt":0.2627009697317941,"score_spread":0.17771307590860466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3138240437","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0069026733,0.0010739645,0.7895723,0.0016362921,0.0009606701,0.0012333372,0.012460714,0.03269364,0.15346637],"genre_scores_gemma":[0.093345165,0.0015302125,0.80689275,0.0015232309,0.000191604,0.0011826081,0.03941097,0.005621496,0.050301965],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99671006,0.0009230309,0.00041793514,0.0006296977,0.001073848,0.00024544128],"domain_scores_gemma":[0.9950441,0.001474154,0.00016776264,0.0020664698,0.0010809039,0.00016665852],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021569652,0.0013894526,0.0008219644,0.0038340841,0.0013497486,0.0052714706,0.002831423,0.001351386,0.05820124],"category_scores_gemma":[0.012835623,0.0007888001,0.0017035591,0.0036941215,0.0009768091,0.004581895,0.006228174,0.0017553346,0.026881317],"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":[0.00025736226,0.00031431502,0.001022171,0.0013094499,0.00016628572,0.0009141852,0.0012227552,0.0091300905,0.0055744974,0.17963459,0.13247654,0.6679778],"study_design_scores_gemma":[0.00007696357,0.00007677459,0.0005012499,0.0003806632,0.0000618671,0.0006998901,0.0008182205,0.027236156,0.011098335,0.16882679,0.7901568,0.00006626057],"about_ca_topic_score_codex":0.0029182548,"about_ca_topic_score_gemma":0.0031804678,"teacher_disagreement_score":0.05820124,"about_ca_system_score_codex":0.0013932979,"about_ca_system_score_gemma":0.0024245554,"threshold_uncertainty_score":0.19470257},"labels":[],"label_agreement":null},{"id":"W3142663085","doi":"10.14778/3447689.3447695","title":"On the string matching with <i>k</i> differences in DNA databases","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Winnipeg","funders":"","keywords":"Substring; Suffix tree; String (physics); String searching algorithm; Combinatorics; Pattern matching; Trie; Bounded function; Speedup; Sequence (biology); Tree (set theory); Computer science; Alphabet; Matching (statistics); Time complexity; Mathematics; Algorithm; Data structure; Artificial intelligence; Parallel computing; Biology","score_opus":0.024683335890819752,"score_gpt":0.22047070030775745,"score_spread":0.1957873644169377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3142663085","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":[0.019203395,0.0035676463,0.9724867,0.0006592704,0.00012713743,0.00017279896,0.00014705969,0.0016918165,0.0019441912],"genre_scores_gemma":[0.10399082,0.0029857967,0.8881436,0.00053732784,0.00030317446,0.00028823141,0.0006802765,0.00036910162,0.0027016918],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9931162,0.0016523938,0.0008150213,0.0017495478,0.0021066905,0.0005601773],"domain_scores_gemma":[0.9932326,0.0036904803,0.0006250975,0.001677863,0.0005588987,0.00021520295],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035066113,0.0014408372,0.0022919234,0.003326591,0.0016494007,0.003285393,0.004348255,0.002403203,0.0029717023],"category_scores_gemma":[0.011895614,0.0010549513,0.0020217127,0.008600943,0.002655583,0.01582088,0.0044114515,0.00247032,0.0019205889],"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":[0.0011484831,0.00044606568,0.0036942305,0.0014524038,0.00019175753,0.0005685095,0.00087514584,0.06753888,0.031693194,0.13381058,0.011859404,0.74672145],"study_design_scores_gemma":[0.00023920811,0.00071537244,0.0021011294,0.00022882568,0.0002315095,0.0025312768,0.0004304843,0.6883934,0.04084031,0.2229141,0.04116912,0.00020529413],"about_ca_topic_score_codex":0.0028818564,"about_ca_topic_score_gemma":0.0016671988,"teacher_disagreement_score":0.004348255,"about_ca_system_score_codex":0.0019688602,"about_ca_system_score_gemma":0.0016948639,"threshold_uncertainty_score":0.018544972},"labels":[],"label_agreement":null},{"id":"W3145740377","doi":"10.14778/3447689.3447700","title":"Dealer","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Sensitivity (control systems); Computer science; Revenue; Function (biology); Differential privacy; Value (mathematics); Set (abstract data type); Artificial intelligence; Machine learning; Algorithm; Economics","score_opus":0.02334067523738829,"score_gpt":0.2458582682426131,"score_spread":0.2225175930052248,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3145740377","genre_codex":"other","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":[0.027595079,0.0019782826,0.15335442,0.009084326,0.0032165158,0.0012081224,0.04232025,0.031465143,0.7297778],"genre_scores_gemma":[0.1153655,0.0009015733,0.056977052,0.0043206788,0.0009226691,0.0006474405,0.05729013,0.004668708,0.75890625],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99727005,0.00041572732,0.0001902772,0.0007982101,0.001020406,0.0003052513],"domain_scores_gemma":[0.9961754,0.0004923472,0.0002429588,0.0019798211,0.000677041,0.0004324587],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018081642,0.00088057935,0.0010850183,0.0014597762,0.0012266524,0.0042866957,0.0022771854,0.0017688324,0.4303056],"category_scores_gemma":[0.0064418735,0.0005733729,0.0010354198,0.0022626442,0.000521874,0.0075257174,0.004466285,0.001995174,0.26037169],"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":[0.00054731016,0.00029573415,0.0040093027,0.00031161963,0.000052547617,0.00038449623,0.0002819393,0.0015689384,0.0029395048,0.045112178,0.61021805,0.33427835],"study_design_scores_gemma":[0.00005385207,0.000053520147,0.0012338357,0.000038551432,0.000016288495,0.00040987114,0.00018510653,0.0075019784,0.001807444,0.011170861,0.9774983,0.000030385243],"about_ca_topic_score_codex":0.0018670176,"about_ca_topic_score_gemma":0.0023914352,"teacher_disagreement_score":0.4303056,"about_ca_system_score_codex":0.0010564785,"about_ca_system_score_gemma":0.0010967728,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3151791879","doi":"10.14778/3447689.3447713","title":"FREDE","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Optech (Canada)","funders":"","keywords":"Computer science","score_opus":0.007135979711530603,"score_gpt":0.22110637753027845,"score_spread":0.21397039781874785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3151791879","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0068996404,0.00088143826,0.9046724,0.0008257052,0.0006360628,0.00024209867,0.008393568,0.03578838,0.04166075],"genre_scores_gemma":[0.12133915,0.0016847838,0.7243572,0.0009636445,0.00025920762,0.00069514767,0.043882012,0.005311491,0.10150744],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993092,0.00011570065,0.000046986275,0.00019773096,0.0002734134,0.000057061545],"domain_scores_gemma":[0.99871254,0.0003009027,0.00006931251,0.00056591537,0.0002851241,0.00006627433],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00078032876,0.0009831594,0.00048131446,0.0013690413,0.00054325396,0.0016918379,0.001336616,0.0010164052,0.048364602],"category_scores_gemma":[0.0047875387,0.00043363724,0.0007870226,0.0010393541,0.00041724357,0.002779681,0.0021343841,0.0013617176,0.029003557],"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":[0.0002342967,0.00013671553,0.0014040571,0.00053713867,0.00010019918,0.00020513688,0.00019262484,0.025507765,0.012192322,0.114295036,0.1861675,0.65902716],"study_design_scores_gemma":[0.000090763584,0.00013191813,0.0010223435,0.000107994136,0.000037934456,0.00075693947,0.00013650798,0.22946878,0.023241432,0.12328864,0.62163144,0.00008538722],"about_ca_topic_score_codex":0.0014506964,"about_ca_topic_score_gemma":0.0030369023,"teacher_disagreement_score":0.048364602,"about_ca_system_score_codex":0.00050657097,"about_ca_system_score_gemma":0.00079689885,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3161742868","doi":"10.14778/3529337.3529339","title":"Accurate summary-based cardinality estimation through the lens of cardinality estimation graphs","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Estimator; Cardinality (data modeling); Graph; Mathematics; Computer science; Joins; Statistics; Mathematical optimization; Theoretical computer science; Data mining","score_opus":0.026187567913794885,"score_gpt":0.2648038244692834,"score_spread":0.23861625655548852,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3161742868","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":[0.0807599,0.0007284549,0.9125609,0.00065085833,0.000045080866,0.00008170006,0.00074326526,0.002106981,0.0023230275],"genre_scores_gemma":[0.710801,0.00036996137,0.28605378,0.00019785731,0.0000614448,0.00011692689,0.0011065239,0.00041577418,0.00087675446],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99035686,0.0039434135,0.000497622,0.0015554182,0.0031821865,0.0004645324],"domain_scores_gemma":[0.940362,0.03990454,0.0053742784,0.009975044,0.0036669183,0.0007172966],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009508972,0.0010286944,0.0013238619,0.002468712,0.00079796574,0.0041590454,0.002165312,0.0012417766,0.0017193026],"category_scores_gemma":[0.079818256,0.00074499525,0.00052172085,0.0028824336,0.0013712967,0.01021967,0.003056455,0.0023052283,0.000407356],"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":[0.0011342502,0.00019719865,0.022120662,0.00041682826,0.00020069472,0.00027267006,0.0015025572,0.47150725,0.014263001,0.26878235,0.0066028177,0.2129997],"study_design_scores_gemma":[0.000021532742,0.000075542746,0.001365363,0.000028428314,0.000028956272,0.00010516948,0.0001985632,0.91141003,0.006171082,0.077845745,0.002715219,0.000034351648],"about_ca_topic_score_codex":0.0031630443,"about_ca_topic_score_gemma":0.0021242576,"teacher_disagreement_score":0.009508972,"about_ca_system_score_codex":0.0022029171,"about_ca_system_score_gemma":0.0017766461,"threshold_uncertainty_score":0.050288856},"labels":[],"label_agreement":null},{"id":"W3164690045","doi":"10.14778/3457390.3457398","title":"CBench","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Topic Modeling","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Computer science; Question answering; Benchmark (surveying); Benchmarking; Suite; Set (abstract data type); Vocabulary; Information retrieval; Syntax; Graph; Artificial intelligence; Task (project management); Natural language processing; Theoretical computer science; Programming language","score_opus":0.01596362296791629,"score_gpt":0.2186539506523365,"score_spread":0.20269032768442022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3164690045","genre_codex":"software","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":[0.03932189,0.004870806,0.20525552,0.0023218761,0.0024000073,0.0025621438,0.084410466,0.41291228,0.24594496],"genre_scores_gemma":[0.1754389,0.0025394442,0.2789048,0.0029252986,0.0003828545,0.0036621531,0.3646212,0.051757623,0.11976772],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99043906,0.00178865,0.0009540799,0.0015757534,0.004353908,0.0008884288],"domain_scores_gemma":[0.9841962,0.003254206,0.00056178164,0.004424116,0.0068580564,0.0007057588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054514743,0.0024405655,0.0014960072,0.0042387904,0.0017061221,0.004822135,0.005067746,0.0021493717,0.054769482],"category_scores_gemma":[0.021978306,0.0009330835,0.0016604166,0.0043366174,0.0010329083,0.005819295,0.0052332585,0.003004603,0.049807444],"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":[0.0016898159,0.0007363506,0.0049563395,0.002487053,0.00015006108,0.00040986936,0.0010150596,0.009941691,0.010334826,0.032323323,0.5929157,0.3430398],"study_design_scores_gemma":[0.00023026543,0.0005296072,0.0042209933,0.00035669105,0.00008261956,0.0004934697,0.000688325,0.041283373,0.016735148,0.025531704,0.9096842,0.00016357783],"about_ca_topic_score_codex":0.010205412,"about_ca_topic_score_gemma":0.009110434,"teacher_disagreement_score":0.054769482,"about_ca_system_score_codex":0.0022638135,"about_ca_system_score_gemma":0.0033872374,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3168854329","doi":"10.14778/3467861.3467872","title":"Data acquisition for improving machine learning models","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; York University","funders":"","keywords":"Data acquisition; Process (computing); Data modeling; Online machine learning; Knowledge acquisition; Training set; Annotation; Data integration; Supervised learning","score_opus":0.28485107715764413,"score_gpt":0.3865902747486838,"score_spread":0.10173919759103967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3168854329","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":[0.043047808,0.0012389788,0.947251,0.0018106578,0.00009071598,0.00024696297,0.0006264637,0.0032114547,0.0024758754],"genre_scores_gemma":[0.42367253,0.00078296015,0.569904,0.00088127604,0.00021215368,0.00054181594,0.0021677278,0.0004238983,0.0014136683],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99238425,0.003929856,0.0004917698,0.0011884773,0.0017009819,0.00030476355],"domain_scores_gemma":[0.95497715,0.032147747,0.0014539228,0.0082481885,0.0028064572,0.00036666953],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012170153,0.0019846521,0.001995232,0.0020640495,0.0010006273,0.0026374136,0.0031444873,0.0022607816,0.004829378],"category_scores_gemma":[0.072211556,0.0010994332,0.0018313262,0.0028532327,0.0017169988,0.007616462,0.0042292792,0.00529235,0.0017943009],"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":[0.0011209685,0.0010026429,0.011551404,0.00045967265,0.0002915233,0.00019943861,0.00055250456,0.4463065,0.0072070523,0.049572986,0.009138313,0.47259694],"study_design_scores_gemma":[0.00006695604,0.00015168721,0.00060158176,0.000036100995,0.00003846759,0.00004951004,0.000079596095,0.9666936,0.005030031,0.02424877,0.0029832975,0.000020376201],"about_ca_topic_score_codex":0.004274661,"about_ca_topic_score_gemma":0.004387464,"teacher_disagreement_score":0.012170153,"about_ca_system_score_codex":0.0017089245,"about_ca_system_score_gemma":0.0029965218,"threshold_uncertainty_score":0.064362705},"labels":[],"label_agreement":null},{"id":"W3176539131","doi":"10.14778/3467861.3467876","title":"Kamino","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"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; Differential privacy; Tuple; Data mining; Schema (genetic algorithms); Publication; Data integrity; Data structure; Probabilistic logic; Database; Information retrieval; Artificial intelligence","score_opus":0.022339782567990962,"score_gpt":0.24248893966061258,"score_spread":0.22014915709262162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3176539131","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03760414,0.003600418,0.57940394,0.009373144,0.00240059,0.0009642669,0.013104133,0.056816336,0.29673308],"genre_scores_gemma":[0.23971502,0.004198555,0.33056378,0.004761426,0.0009849583,0.0011029344,0.029679703,0.0076391613,0.38135448],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974139,0.00039658844,0.00019717157,0.00080796523,0.0010293366,0.00015509891],"domain_scores_gemma":[0.9952644,0.0010087237,0.00029886572,0.002183105,0.00085056585,0.00039438615],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0023201967,0.0007066338,0.0007062999,0.0018570555,0.0014429209,0.0047999504,0.001851633,0.001468619,0.095545925],"category_scores_gemma":[0.008252991,0.0006738205,0.00072070654,0.002097475,0.0009651386,0.0065271053,0.005421986,0.0020889759,0.05593123],"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":[0.00089370727,0.00019896436,0.0041371253,0.00067856937,0.000077470635,0.0005423559,0.00104092,0.004789158,0.021171842,0.09234785,0.15379688,0.7203251],"study_design_scores_gemma":[0.00008427217,0.00016636332,0.001449732,0.00016047718,0.00004798978,0.0009127545,0.00031720335,0.020816078,0.019489244,0.04609546,0.9103857,0.00007481233],"about_ca_topic_score_codex":0.0010176746,"about_ca_topic_score_gemma":0.0012326405,"teacher_disagreement_score":0.90445405,"about_ca_system_score_codex":0.0011135134,"about_ca_system_score_gemma":0.0016197676,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3184011463","doi":"10.14778/3476249.3476263","title":"LES <sup>3</sup>","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; York University","funders":"","keywords":"Search engine indexing; Computer science; Set (abstract data type); Pruning; Similarity (geometry); Bitmap; Nearest neighbor search; Representation (politics); Data mining; Encoding (memory); Theoretical computer science; Artificial intelligence","score_opus":0.01984668302522739,"score_gpt":0.22997774859488967,"score_spread":0.2101310655696623,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3184011463","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038447995,0.0035028988,0.57583904,0.013732589,0.0059359586,0.0017225042,0.14951576,0.034780324,0.17652287],"genre_scores_gemma":[0.23565386,0.002734763,0.40790936,0.005795896,0.0025187836,0.0015388501,0.24167362,0.0054789153,0.09669588],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99806684,0.00026302878,0.00024810922,0.00026698087,0.0010224823,0.00013260765],"domain_scores_gemma":[0.9917476,0.002996085,0.0007039544,0.0024174114,0.0018562372,0.00027865617],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0015501657,0.0008745698,0.0011566316,0.001830933,0.0009556618,0.0038348415,0.0018592292,0.0016618052,0.0846277],"category_scores_gemma":[0.011188397,0.00044039317,0.00087334187,0.0032588325,0.0009019816,0.0032751537,0.002385258,0.001466443,0.03946854],"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":[0.00077040715,0.00015217613,0.0049218573,0.00081247685,0.000056787216,0.0006236722,0.00012498986,0.012523568,0.01023033,0.037915066,0.5812503,0.35061833],"study_design_scores_gemma":[0.00013568664,0.00029541127,0.0053920615,0.000263026,0.000041026233,0.0016435389,0.00042495533,0.17787586,0.036324147,0.07426441,0.70325416,0.000085691725],"about_ca_topic_score_codex":0.002071415,"about_ca_topic_score_gemma":0.005814899,"teacher_disagreement_score":0.9153723,"about_ca_system_score_codex":0.0012454435,"about_ca_system_score_gemma":0.0012969268,"threshold_uncertainty_score":0.28310788},"labels":[],"label_agreement":null},{"id":"W3193493045","doi":"10.14778/3510397.3510400","title":"Making RDBMSs efficient on graph workloads through predefined joins","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"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; Relational database management system; Hash function; Tuple; Graph; Theoretical computer science; Database; Relational database; Parallel computing; Data mining; Programming language","score_opus":0.03152427770464434,"score_gpt":0.2577984677144628,"score_spread":0.22627419000981847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3193493045","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":[0.6496421,0.0011598717,0.22561274,0.0012091051,0.00041561207,0.00069521787,0.002601195,0.0986813,0.019982908],"genre_scores_gemma":[0.7708418,0.00045524418,0.21415041,0.0003272854,0.00007800486,0.00027172855,0.0055374,0.0038939845,0.004444053],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99475116,0.0009704587,0.00045604454,0.00088638737,0.0021527787,0.000783181],"domain_scores_gemma":[0.9899132,0.0025596942,0.00042821481,0.004997568,0.0016709261,0.00043027243],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031008462,0.0012528289,0.0009567195,0.0012709784,0.0014835056,0.00379584,0.0041403924,0.0008514978,0.0033136997],"category_scores_gemma":[0.011128209,0.00093932654,0.000662365,0.0027056038,0.0012365244,0.0066332123,0.003033125,0.0011939723,0.002009158],"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":[0.004318158,0.0011898798,0.019197846,0.0011627675,0.00026912795,0.00078556,0.0014466778,0.17839316,0.16436715,0.026714059,0.07959966,0.52255607],"study_design_scores_gemma":[0.00058898004,0.0004928954,0.0072221407,0.000073332645,0.00012441952,0.00029230796,0.0009948546,0.77549404,0.15052573,0.02160552,0.042447466,0.00013834314],"about_ca_topic_score_codex":0.011502222,"about_ca_topic_score_gemma":0.01252631,"teacher_disagreement_score":0.011502222,"about_ca_system_score_codex":0.0019002199,"about_ca_system_score_gemma":0.0031587682,"threshold_uncertainty_score":0.0228706},"labels":[],"label_agreement":null},{"id":"W3196580506","doi":"10.14778/3476249.3476297","title":"Columnar storage and list-based processing for graph database management systems","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"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; Joins; Scalability; Column (typography); Query optimization; Graph; Block (permutation group theory); Parallel computing; Online aggregation; Database; Relational database management system; Theoretical computer science; Sargable; Relational database; Information retrieval; Web search query; Programming language; Search engine","score_opus":0.013402134967230724,"score_gpt":0.2220701710286801,"score_spread":0.20866803606144937,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196580506","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":[0.032992877,0.0015124129,0.953822,0.0010794945,0.00015416801,0.00015449504,0.0005201806,0.005105034,0.0046594],"genre_scores_gemma":[0.22235718,0.0015793771,0.7673827,0.0007778205,0.00024633622,0.00021364925,0.0012464112,0.0006452319,0.005551349],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99932325,0.0001231797,0.00005995003,0.000107532935,0.00029914503,0.000087022796],"domain_scores_gemma":[0.99770576,0.0006479514,0.00021135088,0.00096828886,0.0003910663,0.00007563711],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000583603,0.0005296627,0.0004315663,0.001385988,0.00078732864,0.002143348,0.0017669634,0.00043741718,0.003710844],"category_scores_gemma":[0.002871273,0.00040358174,0.0004248376,0.0033650198,0.0008326918,0.0036484457,0.0014058441,0.0009672403,0.0009807523],"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":[0.0004798562,0.00029971014,0.0037295476,0.0006806629,0.00009835382,0.00024369493,0.00071935303,0.0675694,0.07469368,0.1982266,0.03517022,0.61808896],"study_design_scores_gemma":[0.00008374858,0.00039099198,0.0016081545,0.000091891445,0.00010821502,0.0007497866,0.0003778181,0.6278543,0.11830019,0.17307921,0.07724177,0.000113899376],"about_ca_topic_score_codex":0.003211176,"about_ca_topic_score_gemma":0.0051962133,"teacher_disagreement_score":0.003710844,"about_ca_system_score_codex":0.0011380343,"about_ca_system_score_gemma":0.0013372902,"threshold_uncertainty_score":0.012413979},"labels":[],"label_agreement":null},{"id":"W3196792859","doi":"10.14778/3476311.3476335","title":"Demonstration of dealer","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Revenue; Variety (cybernetics); Quality (philosophy); Computer science; Business; Artificial intelligence; Finance","score_opus":0.023331209508722677,"score_gpt":0.24659098665221155,"score_spread":0.22325977714348888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196792859","genre_codex":"other","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":[0.15005662,0.00058342953,0.36056682,0.009386161,0.0013239204,0.0017586891,0.0053382977,0.044835784,0.42615032],"genre_scores_gemma":[0.41069078,0.00040881013,0.1868565,0.0026720432,0.00021541097,0.0015642946,0.0065959175,0.0038898143,0.38710642],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992428,0.00017732407,0.0000329629,0.00017143833,0.00024507035,0.0001304152],"domain_scores_gemma":[0.9990627,0.00029463717,0.00003577714,0.00019372265,0.00014559859,0.00026758184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011043046,0.0008436221,0.0004698309,0.00041009023,0.0007770235,0.0017586808,0.0019574054,0.002239408,0.15205061],"category_scores_gemma":[0.0027346315,0.0004281921,0.00059138134,0.00029175114,0.0007559472,0.0029471742,0.004081827,0.0016209132,0.03782637],"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":[0.00295242,0.0028341229,0.010103672,0.00081747025,0.00009175909,0.0075948676,0.0051754974,0.022519069,0.09026015,0.11176368,0.5205093,0.22537805],"study_design_scores_gemma":[0.0003213361,0.0009759825,0.00426432,0.00014781508,0.000038049042,0.002639887,0.0011766154,0.14760672,0.03306705,0.013508515,0.7961292,0.00012450044],"about_ca_topic_score_codex":0.0014962544,"about_ca_topic_score_gemma":0.0014789087,"teacher_disagreement_score":0.15205061,"about_ca_system_score_codex":0.0005066647,"about_ca_system_score_gemma":0.00061242067,"threshold_uncertainty_score":0.50865996},"labels":[],"label_agreement":null},{"id":"W3196867679","doi":"10.14778/3476311.3476364","title":"RONIN","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Tech University; University of Toronto","funders":"","keywords":"Computer science; Information retrieval; Set (abstract data type); Data mining; Focus (optics); Nearest neighbor search; Data set; Artificial intelligence","score_opus":0.1406382432299656,"score_gpt":0.3809184801512673,"score_spread":0.24028023692130168,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196867679","genre_codex":"other","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":[0.0066250614,0.005304344,0.15527174,0.008408885,0.003316893,0.0009021427,0.1276543,0.2559573,0.43655932],"genre_scores_gemma":[0.055321924,0.006303667,0.21683827,0.005851058,0.0014553231,0.001810399,0.3537599,0.044982698,0.31367674],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959908,0.00061278517,0.0003666948,0.0011961563,0.001549198,0.0002844021],"domain_scores_gemma":[0.9940521,0.0014468682,0.00036231725,0.0023886822,0.0012171814,0.0005327756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041632364,0.0016841319,0.001563735,0.0043173786,0.0014235706,0.0070056696,0.0033953905,0.0016326372,0.2390095],"category_scores_gemma":[0.015003483,0.000985006,0.0014958412,0.0039336877,0.00086852576,0.006416695,0.005325831,0.002052159,0.19751386],"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":[0.00053342275,0.00009758084,0.0018770734,0.0010351297,0.00008945593,0.0002897956,0.00028832752,0.0008264989,0.0028843232,0.029222308,0.7598981,0.20295797],"study_design_scores_gemma":[0.000058371217,0.00003684893,0.00073658064,0.00012631844,0.000021053254,0.00023215798,0.000065231856,0.0024476473,0.0018024186,0.011245233,0.98319525,0.00003292917],"about_ca_topic_score_codex":0.0027436686,"about_ca_topic_score_gemma":0.004540048,"teacher_disagreement_score":0.2390095,"about_ca_system_score_codex":0.0013118403,"about_ca_system_score_gemma":0.0024434612,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3196877232","doi":"10.14778/3476311.3476317","title":"A demonstration of KGLac","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; SPARQL; Metadata; Data science; Knowledge graph; Knowledge extraction; Data discovery; Information retrieval; Pipeline (software); Annotation; RDF; Graph; World Wide Web; Data mining; Semantic Web; Artificial intelligence; Theoretical computer science","score_opus":0.12604269999402856,"score_gpt":0.37172051449046417,"score_spread":0.2456778144964356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196877232","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03298176,0.00062855985,0.718341,0.0061590355,0.0005996736,0.00049765117,0.013788659,0.1899128,0.037090816],"genre_scores_gemma":[0.16201124,0.0005391289,0.79509395,0.0026354634,0.00015355548,0.00035378756,0.02173735,0.0065852976,0.010890225],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964064,0.0008294596,0.00026454675,0.0009199691,0.0013207182,0.00025886792],"domain_scores_gemma":[0.9848231,0.004744708,0.0006068921,0.0064322227,0.0027639319,0.0006291516],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00451425,0.00058818987,0.00050331565,0.0022224719,0.0015499907,0.00340523,0.002059994,0.0012652188,0.012383708],"category_scores_gemma":[0.016679376,0.0005324878,0.0009968437,0.004245347,0.001968003,0.0056165517,0.0051155756,0.001971572,0.010070193],"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":[0.0009640704,0.00078458304,0.02499083,0.0011635419,0.0001713883,0.0011842051,0.0035929775,0.0196317,0.019594625,0.15054314,0.3091299,0.46824902],"study_design_scores_gemma":[0.0001954815,0.0002713547,0.0060634436,0.0002491309,0.00009546252,0.0011036509,0.001124143,0.24025413,0.02697484,0.16858223,0.5549228,0.0001633281],"about_ca_topic_score_codex":0.013743342,"about_ca_topic_score_gemma":0.014902268,"teacher_disagreement_score":0.013743342,"about_ca_system_score_codex":0.0014199716,"about_ca_system_score_gemma":0.0040941276,"threshold_uncertainty_score":0.041427612},"labels":[],"label_agreement":null},{"id":"W3197016261","doi":"10.14778/3476311.3476326","title":"CBench","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Topic Modeling","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Computer science; Benchmarking; Suite; Benchmark (surveying); Question answering; Set (abstract data type); Task (project management); Quality (philosophy); Artificial intelligence; Information retrieval; Natural language processing; Programming language; Systems engineering","score_opus":0.01596362296791629,"score_gpt":0.2186539506523365,"score_spread":0.20269032768442022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3197016261","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":[0.0424864,0.005157063,0.18930519,0.0015975416,0.0016657104,0.0021853617,0.08803118,0.54420537,0.12536617],"genre_scores_gemma":[0.2062019,0.0026297627,0.288634,0.0021423271,0.00032669737,0.00365202,0.37554494,0.057456892,0.06341153],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99283785,0.0015303052,0.0008526623,0.0011998389,0.0029685453,0.000610826],"domain_scores_gemma":[0.9865362,0.003396722,0.00050821406,0.0039796494,0.0050370153,0.0005423375],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004941433,0.0022834754,0.0012854395,0.0044938484,0.0013302249,0.003980914,0.0041955146,0.0019064222,0.033811454],"category_scores_gemma":[0.020909846,0.0008568113,0.001444134,0.0044206884,0.00084607943,0.0048653595,0.0042488193,0.0025178492,0.031983517],"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":[0.0016439591,0.00090055936,0.0055253743,0.002940087,0.00022593599,0.0003418904,0.00091869925,0.010772586,0.012347705,0.027381452,0.64675266,0.29024905],"study_design_scores_gemma":[0.00042149136,0.00071544683,0.006729153,0.00050473673,0.00012825176,0.00051454996,0.0007034084,0.09204452,0.02749698,0.034175634,0.8363254,0.00024050692],"about_ca_topic_score_codex":0.009404302,"about_ca_topic_score_gemma":0.008403559,"teacher_disagreement_score":0.033811454,"about_ca_system_score_codex":0.001617888,"about_ca_system_score_gemma":0.0029009045,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3197536806","doi":"10.14778/3476311.3476394","title":"The future of data(base) education","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Popularity; Field (mathematics); Computer science; Mathematics education; Position (finance); Base (topology); Core (optical fiber); Data science; Psychology; Mathematics","score_opus":0.05319081169306056,"score_gpt":0.2876094141118519,"score_spread":0.23441860241879137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3197536806","genre_codex":"commentary","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":[0.003791274,0.036647603,0.006566168,0.85076,0.012066444,0.000033371656,0.00069555,0.00040999078,0.08902961],"genre_scores_gemma":[0.20255816,0.16024773,0.036610503,0.32923543,0.031940915,0.0003021104,0.0029977395,0.0007049829,0.23540251],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99324435,0.002158858,0.0004182739,0.00080798485,0.0019861674,0.0013842368],"domain_scores_gemma":[0.95635504,0.0149142435,0.0014857991,0.003155327,0.0074382043,0.01665128],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02353622,0.00036528273,0.0005240271,0.0029908274,0.0036348624,0.018399073,0.0024406791,0.005457181,0.049782787],"category_scores_gemma":[0.023390342,0.00038168306,0.0006192845,0.0044837277,0.00407529,0.021386396,0.009630009,0.007781354,0.008949819],"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":[0.000050164363,0.0001911175,0.0020366493,0.00035854327,0.000008911677,0.00006435749,0.0010743135,0.00025539118,0.00022466526,0.1921259,0.47124934,0.33236054],"study_design_scores_gemma":[0.000008731764,0.000024886345,0.00062596466,0.0003058877,0.0000023948887,0.000052618918,0.0008144991,0.0002025575,0.00011028467,0.02841516,0.9694286,0.000008381573],"about_ca_topic_score_codex":0.0069923704,"about_ca_topic_score_gemma":0.009205782,"teacher_disagreement_score":0.049782787,"about_ca_system_score_codex":0.0077480404,"about_ca_system_score_gemma":0.01937993,"threshold_uncertainty_score":0.16654003},"labels":[],"label_agreement":null},{"id":"W3198515637","doi":"10.14778/3476249.3476283","title":"SlimChain","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Blockchain Technology Applications and Security","field":"Computer Science","cited_by":143,"is_retracted":false,"has_abstract":true,"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; Blockchain; Scalability; Distributed computing; Database transaction; Computer network; Robustness (evolution); Distributed data store; Node (physics); Computer security; Database","score_opus":0.007848441708610996,"score_gpt":0.20753097551931157,"score_spread":0.19968253381070059,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3198515637","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":[0.05609741,0.0029332372,0.66745377,0.0018756109,0.0015445122,0.0017984207,0.005697579,0.037876707,0.22472262],"genre_scores_gemma":[0.6090364,0.0027980169,0.24904837,0.0012420021,0.0004242036,0.0015282483,0.016321847,0.0019326314,0.117668204],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99894506,0.0001769741,0.00007703304,0.0001763482,0.0004115542,0.00021302735],"domain_scores_gemma":[0.9982717,0.00022980715,0.00011969231,0.00069416536,0.00047070067,0.0002139533],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008915981,0.00054390135,0.00059619185,0.0008762743,0.0013305602,0.0016936653,0.0014524091,0.00085219956,0.035301678],"category_scores_gemma":[0.0025257932,0.0003702161,0.00041949592,0.0013018666,0.00065807626,0.0027640958,0.0026837038,0.0011505389,0.011538059],"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":[0.001481191,0.00042060416,0.0039235475,0.00137039,0.00013512938,0.0010411239,0.00053151906,0.052712776,0.028870724,0.22856615,0.17162977,0.50931716],"study_design_scores_gemma":[0.00043122945,0.00047814663,0.0007939258,0.00019758493,0.000070382914,0.0006505815,0.0001672633,0.29307017,0.027098028,0.11909952,0.5578431,0.00010001511],"about_ca_topic_score_codex":0.0026147545,"about_ca_topic_score_gemma":0.0034318748,"teacher_disagreement_score":0.035301678,"about_ca_system_score_codex":0.00084560836,"about_ca_system_score_gemma":0.00219982,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W3198645887","doi":"10.14778/3476249.3476304","title":"Explaining inference queries with bayesian optimization","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"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; Inference; Python (programming language); Bayesian inference; Query optimization; Predicate (mathematical logic); Bayesian probability; Data mining; Information retrieval; Theoretical computer science; Artificial intelligence; Programming language","score_opus":0.010849102484855025,"score_gpt":0.230534737405498,"score_spread":0.219685634920643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3198645887","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":[0.0077332016,0.00022605393,0.98489505,0.00088917697,0.000024396511,0.00007879015,0.00067318126,0.004253529,0.0012266359],"genre_scores_gemma":[0.20127235,0.00042519657,0.7895841,0.0007254795,0.00009665092,0.00028086468,0.0032986784,0.0016768668,0.0026398487],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99532497,0.0018415416,0.00027386015,0.00085220626,0.0014181612,0.00028925424],"domain_scores_gemma":[0.9906284,0.0069864816,0.00048175122,0.0010292548,0.00073623453,0.0001378922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0052287127,0.0017049635,0.0009810298,0.0019773112,0.0007940188,0.002839531,0.0022862495,0.0018317204,0.0069890767],"category_scores_gemma":[0.023529388,0.0008324522,0.002768903,0.0015833133,0.0016621945,0.004326524,0.0033511177,0.0033113996,0.00143395],"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":[0.00053685444,0.0002888012,0.008494598,0.000670722,0.00025646272,0.00047611317,0.0013416263,0.4082782,0.0057288674,0.1746369,0.031418666,0.36787218],"study_design_scores_gemma":[0.00003728382,0.000028621138,0.0004638783,0.000052858042,0.000040703908,0.000072146686,0.00010061083,0.8692558,0.0028590832,0.120798625,0.0062598884,0.000030440293],"about_ca_topic_score_codex":0.007088561,"about_ca_topic_score_gemma":0.012230774,"teacher_disagreement_score":0.007088561,"about_ca_system_score_codex":0.0020340444,"about_ca_system_score_gemma":0.002913699,"threshold_uncertainty_score":0.027652383},"labels":[],"label_agreement":null},{"id":"W3198782179","doi":"10.14778/3476311.3476363","title":"Catch a blowfish alive","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"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; Differential privacy; Domain (mathematical analysis); Bounded function; Information privacy; Privacy policy; Computer security; Data mining","score_opus":0.020958974059684658,"score_gpt":0.24404106155515926,"score_spread":0.2230820874954746,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3198782179","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":[0.05106233,0.0018498776,0.8462403,0.012495805,0.0008478481,0.00048140902,0.0036266146,0.060711157,0.022684745],"genre_scores_gemma":[0.4685129,0.001616662,0.4609666,0.010490119,0.00028398112,0.00060429785,0.006221015,0.01100912,0.04029535],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99176323,0.0015602869,0.0005697821,0.0021754378,0.0031205455,0.0008108043],"domain_scores_gemma":[0.9827319,0.0043180897,0.00046766418,0.010086438,0.0015771433,0.0008186912],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075087436,0.0009799948,0.00153159,0.0013348853,0.0020365492,0.005110385,0.003096944,0.0030430197,0.012649343],"category_scores_gemma":[0.024399787,0.001059579,0.0020228645,0.0014826152,0.003889973,0.015772786,0.017287998,0.006461273,0.0052507534],"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":[0.0037707426,0.00038189418,0.015454909,0.0011148535,0.0004931814,0.0021032705,0.008635815,0.022009192,0.06642472,0.25300136,0.17010638,0.45650366],"study_design_scores_gemma":[0.00018740442,0.00037206145,0.0025601534,0.0005308625,0.00014816358,0.0015315493,0.0023525998,0.14044623,0.06370771,0.35810515,0.4298189,0.00023912321],"about_ca_topic_score_codex":0.0033728485,"about_ca_topic_score_gemma":0.0022930577,"teacher_disagreement_score":0.012649343,"about_ca_system_score_codex":0.0015163681,"about_ca_system_score_gemma":0.0027589458,"threshold_uncertainty_score":0.042316318},"labels":[],"label_agreement":null},{"id":"W3200211247","doi":"10.14778/3485450.3485462","title":"Accelerating recommendation system training by leveraging popular choices","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Recommender Systems and Techniques","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Embedding; Recommender system; Categorical variable; Popularity; Parallel computing; Feature (linguistics); Representation (politics); Machine learning; Artificial intelligence","score_opus":0.04706617408481114,"score_gpt":0.2487423492667841,"score_spread":0.20167617518197295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3200211247","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":[0.2195842,0.0013268393,0.75651705,0.00066378264,0.00018076955,0.00017192081,0.0012657349,0.01496404,0.005325767],"genre_scores_gemma":[0.6155521,0.00042741655,0.37576738,0.00018010085,0.000068034686,0.00019542957,0.0026833387,0.0003764409,0.0047496753],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99919206,0.0001573817,0.000057104393,0.0002610768,0.00022940738,0.00010299947],"domain_scores_gemma":[0.997741,0.0009044828,0.000099026714,0.00077147206,0.00039676583,0.00008724104],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006723413,0.00068566087,0.0007941596,0.00062552374,0.0004818536,0.0010799515,0.0018186017,0.0009235611,0.0045666015],"category_scores_gemma":[0.0061319717,0.0007737358,0.0006769887,0.0012510058,0.00041712564,0.0025464164,0.0012671461,0.0013739491,0.0032445504],"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":[0.00094808376,0.0005984327,0.020104287,0.0002649529,0.00018999306,0.00027826894,0.00058266433,0.19240396,0.031239288,0.0041582985,0.013734405,0.7354973],"study_design_scores_gemma":[0.00005065117,0.00013878747,0.001902648,0.000012750807,0.000027340055,0.00009516026,0.00009530357,0.9803877,0.009598402,0.004281101,0.0033908326,0.000019323099],"about_ca_topic_score_codex":0.009717929,"about_ca_topic_score_gemma":0.02246052,"teacher_disagreement_score":0.009717929,"about_ca_system_score_codex":0.0005322853,"about_ca_system_score_gemma":0.0009258174,"threshold_uncertainty_score":0.019322753},"labels":[],"label_agreement":null},{"id":"W3210479060","doi":"10.14778/3461535.3461546","title":"Comprehensible counterfactual explanation on Kolmogorov-Smirnov test","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Simon Fraser University","funders":"","keywords":"Counterfactual thinking; Benchmark (surveying); Scalability; Test (biology); Set (abstract data type); Test set; Computer science; Exponential function; Algorithm; Artificial intelligence; Mathematics; Machine learning; Psychology; Database","score_opus":0.019585693011753096,"score_gpt":0.23513753534632073,"score_spread":0.21555184233456764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3210479060","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":[0.109482646,0.0025965876,0.87275547,0.0021665115,0.00017780316,0.0005857209,0.0027850356,0.0046305205,0.004819587],"genre_scores_gemma":[0.6354214,0.000658698,0.35716775,0.00047368236,0.00020637718,0.0007641529,0.0043437392,0.0002769177,0.00068716984],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9700495,0.014975138,0.0026326962,0.004722533,0.0069654346,0.0006547037],"domain_scores_gemma":[0.7059104,0.25907308,0.010386664,0.0155712245,0.0078594005,0.0011992671],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.023928791,0.0016412407,0.0017341991,0.005373106,0.0013377864,0.0040943082,0.0030886566,0.0027802095,0.005696033],"category_scores_gemma":[0.21381177,0.0006107489,0.0027510198,0.002968196,0.003606318,0.006776911,0.0031983545,0.003159241,0.00087580894],"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":[0.0024393054,0.00058108266,0.07398145,0.0024930376,0.0012754198,0.0013407879,0.0014379135,0.19605355,0.006298952,0.12270384,0.019096764,0.57229793],"study_design_scores_gemma":[0.00027958004,0.00059917534,0.01595409,0.00031818033,0.00027312286,0.00078531913,0.0005289227,0.6716638,0.0080195265,0.29341984,0.007968711,0.0001897202],"about_ca_topic_score_codex":0.001674141,"about_ca_topic_score_gemma":0.0015568845,"teacher_disagreement_score":0.023928791,"about_ca_system_score_codex":0.0019316798,"about_ca_system_score_gemma":0.0039206664,"threshold_uncertainty_score":0.12654907},"labels":[],"label_agreement":null},{"id":"W3211025647","doi":"10.14778/3484224.3484231","title":"Quantifying identifiability to choose and audit ϵ in differentially private deep learning","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Differential privacy; Identifiability; Computer science; Adversary; Bounding overwatch; Upper and lower bounds; Inference; Bayesian inference; Audit; Information privacy; Obfuscation; Data mining; Machine learning; Computer security; Bayesian probability; Artificial intelligence; Mathematics","score_opus":0.029575709233698576,"score_gpt":0.2709407707795119,"score_spread":0.24136506154581333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3211025647","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":[0.03902716,0.0002567199,0.9546271,0.0025497961,0.000041544346,0.00018212711,0.00027102875,0.00036811834,0.0026764136],"genre_scores_gemma":[0.8400007,0.00043249995,0.15469316,0.0010447433,0.000101112004,0.0005832514,0.0004132654,0.0001605892,0.002570707],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.96664,0.015934166,0.0021185693,0.005189112,0.0073887953,0.0027293686],"domain_scores_gemma":[0.8700028,0.08803424,0.0068155346,0.029197823,0.0042754826,0.001674164],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.035337143,0.0017188011,0.002243125,0.0013512115,0.0023595854,0.0069122417,0.0051489165,0.0047349287,0.0025942456],"category_scores_gemma":[0.15063782,0.001746525,0.002000185,0.0021941022,0.011568428,0.017765798,0.013127331,0.009890192,0.0006344289],"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":[0.00085294375,0.00025557083,0.006358959,0.00028998687,0.00017030444,0.00044259467,0.0015975104,0.30480745,0.00439761,0.6251955,0.0020536985,0.053577866],"study_design_scores_gemma":[0.000061050916,0.00009350063,0.0003844223,0.000077421064,0.00004491531,0.00016401972,0.00011944893,0.42240837,0.00463502,0.57083464,0.0011320588,0.0000451191],"about_ca_topic_score_codex":0.0019857793,"about_ca_topic_score_gemma":0.0019485209,"teacher_disagreement_score":0.035337143,"about_ca_system_score_codex":0.0071371635,"about_ca_system_score_gemma":0.0063410397,"threshold_uncertainty_score":0.18688285},"labels":[],"label_agreement":null},{"id":"W3211229017","doi":"10.14778/3551793.3551804","title":"A scalable AutoML approach based on graph neural networks","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Scalability; Scripting language; Metadata; Pipeline (software); Artificial intelligence; Machine learning; Graph; Pipeline transport; Theoretical computer science; Database; Programming language; World Wide Web","score_opus":0.012770768817358303,"score_gpt":0.21496839349981606,"score_spread":0.20219762468245775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3211229017","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":[0.009084394,0.00042343928,0.9274058,0.0005678299,0.000082260485,0.00020655687,0.0013748799,0.056166373,0.0046885144],"genre_scores_gemma":[0.16054498,0.00029374182,0.82229084,0.00065873313,0.00007730336,0.00040195798,0.005019492,0.002503995,0.008208968],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99901474,0.00017408059,0.00004117305,0.00039182685,0.00029929265,0.00007879962],"domain_scores_gemma":[0.99857426,0.0004437688,0.00011094563,0.00052171166,0.0002953014,0.000053957363],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012437665,0.0015714108,0.0010760342,0.0027522487,0.0009588877,0.0020097797,0.003575057,0.0015785738,0.008274923],"category_scores_gemma":[0.0040085413,0.0009734,0.0014588264,0.0022420485,0.00081520394,0.004682596,0.002071349,0.0020258971,0.004337719],"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":[0.00014829925,0.00027733255,0.0022664838,0.00030471315,0.00016444092,0.00013160946,0.000105166844,0.25345188,0.005905196,0.01869638,0.038461834,0.6800867],"study_design_scores_gemma":[0.000012467606,0.00001830431,0.00012115239,0.000010322928,0.000011476434,0.000025446097,0.0000174932,0.97958416,0.0017539255,0.014143389,0.004291725,0.00001010064],"about_ca_topic_score_codex":0.0132641485,"about_ca_topic_score_gemma":0.029765079,"teacher_disagreement_score":0.0132641485,"about_ca_system_score_codex":0.0023507355,"about_ca_system_score_gemma":0.0021480068,"threshold_uncertainty_score":0.027682364},"labels":[],"label_agreement":null},{"id":"W4206548428","doi":"10.14778/3494124.3494141","title":"APEX","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Liquid Crystal Research Advancements","field":"Materials Science","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Apex (geometry); Medicine; Anatomy","score_opus":0.019380075659541512,"score_gpt":0.2763615835010787,"score_spread":0.25698150784153717,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206548428","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13385017,0.012413606,0.2638547,0.0023559376,0.0030630596,0.0009900614,0.0250676,0.23577283,0.32263196],"genre_scores_gemma":[0.55236965,0.004507512,0.19779527,0.0017908076,0.000637928,0.00065472256,0.07438095,0.013099893,0.15476315],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99933726,0.00003870494,0.000052114094,0.00012767607,0.00033436352,0.00010987803],"domain_scores_gemma":[0.9990012,0.000099303485,0.00005136422,0.00043170407,0.00031245663,0.00010402659],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00043587567,0.0006541661,0.0004928835,0.00060164207,0.0006526122,0.0020791562,0.002292171,0.0005639983,0.029491564],"category_scores_gemma":[0.0019722118,0.0003877189,0.0003797301,0.0009130773,0.0003540976,0.0030716558,0.002431211,0.0012579066,0.011400903],"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":[0.0014314748,0.00023239371,0.0046015405,0.00088751846,0.00012352252,0.00027813113,0.00022755653,0.008799028,0.04139818,0.039108396,0.36808243,0.5348299],"study_design_scores_gemma":[0.00019311809,0.0005763924,0.002648625,0.00010466108,0.00010451077,0.0005967389,0.00013767729,0.08293745,0.08338598,0.016724754,0.8124907,0.000099505894],"about_ca_topic_score_codex":0.0022385737,"about_ca_topic_score_gemma":0.003139178,"teacher_disagreement_score":0.97050846,"about_ca_system_score_codex":0.00080584054,"about_ca_system_score_gemma":0.0014184599,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4210797700","doi":"10.14778/3489496.3489504","title":"LargeEA","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Scalability; Benchmark (surveying); Exploit; Process (computing); Channel (broadcasting); Partition (number theory); Competitor analysis; Feature (linguistics); Data mining; Artificial intelligence; Database; Programming language","score_opus":0.007208047789126523,"score_gpt":0.2039609482731143,"score_spread":0.1967529004839878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210797700","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":[0.00930729,0.0025778308,0.860348,0.0011677024,0.0005649751,0.00076394883,0.013042344,0.07794819,0.034279723],"genre_scores_gemma":[0.066601336,0.001213289,0.83352095,0.0012453196,0.00019692845,0.0007520067,0.06387875,0.0075013116,0.025090098],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99782664,0.0005161927,0.00011700669,0.0008600304,0.00047870874,0.00020146002],"domain_scores_gemma":[0.9972881,0.0007766431,0.0001127296,0.0012396916,0.00047066042,0.00011225135],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018956339,0.0022814267,0.0012274501,0.0036485947,0.0012453059,0.0027079594,0.0044562654,0.0017655855,0.037737668],"category_scores_gemma":[0.008922828,0.00082121795,0.002343537,0.0032950835,0.0007285713,0.0058752336,0.004218842,0.002382686,0.020990206],"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":[0.0003588291,0.00031086337,0.0018516362,0.0012671008,0.00033737384,0.00033737544,0.0001953886,0.043076906,0.006372924,0.04456819,0.19341016,0.70791334],"study_design_scores_gemma":[0.00022305551,0.00027571892,0.0017428654,0.00026898188,0.00025141495,0.0008150476,0.0003596174,0.45986193,0.017439445,0.1475069,0.37115973,0.00009535076],"about_ca_topic_score_codex":0.0047347397,"about_ca_topic_score_gemma":0.0116076805,"teacher_disagreement_score":0.037737668,"about_ca_system_score_codex":0.0014892331,"about_ca_system_score_gemma":0.002156917,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4226196214","doi":"10.14778/3503585.3503597","title":"COMET","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Lossy compression; Computer science; Speedup; Overhead (engineering); Convolutional neural network; Compression ratio; Bandwidth (computing); Process (computing); Computer engineering; Artificial neural network; Compression (physics); Bounded function; Data compression; Algorithm; Parallel computing; Artificial intelligence; Telecommunications","score_opus":0.0152569447225827,"score_gpt":0.24091374650963057,"score_spread":0.22565680178704786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226196214","genre_codex":"other","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":[0.0053970027,0.00416828,0.15902291,0.0053975843,0.0046799155,0.00084179005,0.05080109,0.27870503,0.49098635],"genre_scores_gemma":[0.05803754,0.005041895,0.1384894,0.0073452285,0.0016718139,0.0015118669,0.2097307,0.068550214,0.50962126],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983406,0.00015927559,0.00009035534,0.0003887293,0.00079919386,0.00022187839],"domain_scores_gemma":[0.99825424,0.00023678536,0.00009380072,0.00059049804,0.0005389367,0.0002857648],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012395378,0.0018195533,0.0012391071,0.0017572824,0.0013322784,0.0042115143,0.004163404,0.0031359969,0.32348168],"category_scores_gemma":[0.0045260377,0.00085460127,0.0013668479,0.001612414,0.0006345922,0.005290518,0.004841652,0.0029973055,0.286431],"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":[0.00039768766,0.00012173158,0.00071980484,0.00059184246,0.000053200176,0.0002554295,0.00011444156,0.0022388725,0.004943126,0.0187351,0.76892114,0.20290759],"study_design_scores_gemma":[0.00008839219,0.000066764194,0.000452059,0.00007475837,0.000018897694,0.00028417984,0.00003744332,0.007645812,0.004237392,0.010190916,0.9768593,0.00004407132],"about_ca_topic_score_codex":0.0026470856,"about_ca_topic_score_gemma":0.0034556922,"teacher_disagreement_score":0.32348168,"about_ca_system_score_codex":0.0010965968,"about_ca_system_score_gemma":0.0017038707,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4240133548","doi":"10.1145/3186728.3164137","title":"Query fresh","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"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; Backup; Overhead (engineering); Operating system; Database; Computer network; Replication (statistics)","score_opus":0.013816238538806296,"score_gpt":0.2359611939109308,"score_spread":0.2221449553721245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4240133548","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":[0.029990593,0.0046180044,0.37629375,0.005284704,0.0035462484,0.0031504387,0.045952007,0.28589496,0.24526936],"genre_scores_gemma":[0.26827714,0.0035229637,0.22263494,0.012403552,0.0021000197,0.0018018668,0.12678656,0.05914813,0.30332482],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99616635,0.00037277458,0.00032886106,0.0009922829,0.0015617874,0.0005779799],"domain_scores_gemma":[0.99365246,0.0009090312,0.00020861809,0.0032309692,0.0017386143,0.00026022882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024993762,0.001935225,0.0015981551,0.001762956,0.0018027647,0.005875465,0.0036995134,0.0017761016,0.12161238],"category_scores_gemma":[0.008294077,0.0010331909,0.0013061471,0.0017368227,0.0010274339,0.011169813,0.006834961,0.0023094916,0.059560664],"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":[0.0018769426,0.00030132246,0.0041951505,0.0015682875,0.00021285182,0.00070097035,0.0011292741,0.002534809,0.029243201,0.0726414,0.60121024,0.28438565],"study_design_scores_gemma":[0.00021342366,0.00020187613,0.0017546018,0.000119144184,0.00009694282,0.00074416556,0.00057607546,0.01561983,0.024361758,0.029708443,0.92646176,0.000141921],"about_ca_topic_score_codex":0.0049508205,"about_ca_topic_score_gemma":0.0045411573,"teacher_disagreement_score":0.12161238,"about_ca_system_score_codex":0.0018791755,"about_ca_system_score_gemma":0.0022107332,"threshold_uncertainty_score":0.4068339},"labels":[],"label_agreement":null},{"id":"W4245107420","doi":"10.14778/3236187.3269462","title":"Efficient construction of approximate ad-hoc ML models through materialization and reuse","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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 analytical processing; Reuse; Dimension (graph theory); Cluster analysis; Data mining; Variety (cybernetics); Construct (python library); Data warehouse; Mixture model; Machine learning; Artificial intelligence; Mathematics","score_opus":0.0167162247919843,"score_gpt":0.22005666246918465,"score_spread":0.20334043767720034,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4245107420","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":[0.014802088,0.00014003973,0.9811152,0.00027888792,0.000013750535,0.000090259135,0.00014772874,0.0026009036,0.0008110786],"genre_scores_gemma":[0.23310429,0.00020913318,0.7622724,0.00023556163,0.00006519612,0.0002665839,0.0013107461,0.00087935943,0.0016567322],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99415344,0.0019901572,0.00044691676,0.0010427558,0.0018483808,0.00051832374],"domain_scores_gemma":[0.98116106,0.010481738,0.0010443123,0.005667471,0.0012619224,0.00038350967],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058200895,0.002239894,0.0026905995,0.0020721988,0.0013524585,0.004736695,0.004772128,0.0020105545,0.0034789182],"category_scores_gemma":[0.028147131,0.0019003085,0.0033305017,0.003361834,0.0023380548,0.00886617,0.0070944736,0.0038882683,0.0017005646],"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":[0.00018448447,0.00020675806,0.0025815852,0.00013380825,0.00011871364,0.00021941777,0.0003984245,0.8080758,0.0027494635,0.03882231,0.004327176,0.1421821],"study_design_scores_gemma":[0.000013230606,0.00001956592,0.0000602112,0.000004352838,0.000011157769,0.00002889935,0.00005403761,0.9783036,0.001054648,0.019617595,0.0008241465,0.000008479864],"about_ca_topic_score_codex":0.009648849,"about_ca_topic_score_gemma":0.012222269,"teacher_disagreement_score":0.009648849,"about_ca_system_score_codex":0.00276346,"about_ca_system_score_gemma":0.0038679263,"threshold_uncertainty_score":0.030779958},"labels":[],"label_agreement":null},{"id":"W4246281707","doi":"10.1145/3187009.3164147","title":"Bztree","year":2018,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":true,"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; Throughput; Block (permutation group theory); Non-volatile memory; Embedded system; Computer hardware; Tree (set theory); Code (set theory); Parallel computing; Operating system; Programming language","score_opus":0.01413647543843038,"score_gpt":0.23617268233670108,"score_spread":0.2220362068982707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4246281707","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":[0.018353008,0.0060681454,0.39856017,0.0013964608,0.0015436243,0.0007174215,0.044397246,0.3352894,0.19367455],"genre_scores_gemma":[0.1537532,0.0047457325,0.43448636,0.0028620919,0.00047966576,0.0014525543,0.19007099,0.050400402,0.16174902],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989575,0.00008217567,0.000099027355,0.00018232928,0.00051477534,0.00016413136],"domain_scores_gemma":[0.9986564,0.0001716889,0.00009152132,0.00050094456,0.00047535068,0.000104016864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00063394295,0.0010087659,0.0008275282,0.0013385765,0.0009607441,0.0032579016,0.003539043,0.0009999827,0.06849849],"category_scores_gemma":[0.0027543092,0.0008534876,0.00085472583,0.0022751354,0.00046021902,0.005128909,0.0034252764,0.0015691044,0.05795289],"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":[0.0013351823,0.00016523487,0.0020448775,0.0014898124,0.00014865106,0.0003114685,0.00043570058,0.0031933938,0.032628525,0.06018229,0.52514124,0.37292364],"study_design_scores_gemma":[0.00017582893,0.00016754528,0.0010179882,0.00011419441,0.000058809826,0.00052934984,0.0001300827,0.015807658,0.02789436,0.027050523,0.92696625,0.00008748878],"about_ca_topic_score_codex":0.0030353493,"about_ca_topic_score_gemma":0.0036268358,"teacher_disagreement_score":0.06849849,"about_ca_system_score_codex":0.0010022077,"about_ca_system_score_gemma":0.0015785204,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4255889623","doi":"10.1145/3186728.3164139","title":"The ubiquity of large graphs and surprising challenges of graph processing","year":2017,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":145,"is_retracted":false,"has_abstract":true,"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; Suite; Scalability; Visualization; Graph; Software; Data science; Theoretical computer science; World Wide Web; Data mining; Programming language; Database","score_opus":0.03165215848905461,"score_gpt":0.305877498930536,"score_spread":0.2742253404414814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4255889623","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":[0.5449855,0.024137532,0.3449883,0.055874463,0.0007666642,0.0003722096,0.002189488,0.00424297,0.022442883],"genre_scores_gemma":[0.7618042,0.012735474,0.2118086,0.0042535937,0.00088567653,0.00030182936,0.0022687442,0.002382689,0.0035591272],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98168415,0.009347416,0.00072699296,0.002397612,0.005373306,0.00047046068],"domain_scores_gemma":[0.81012666,0.16145663,0.006602038,0.011109344,0.008404965,0.0023003416],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017004652,0.00059406104,0.0006192609,0.003853207,0.002008102,0.0053452714,0.0019583623,0.002045353,0.002530016],"category_scores_gemma":[0.09851131,0.0009708655,0.0007336288,0.005026787,0.0038724227,0.014602751,0.0032484343,0.002793205,0.0008190684],"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":[0.0004139372,0.00018633633,0.07200332,0.0072700023,0.0003252301,0.0030596734,0.09332798,0.010814671,0.019997561,0.096209586,0.068151176,0.6282406],"study_design_scores_gemma":[0.00005189896,0.00032623517,0.05978908,0.0016406276,0.0001238301,0.009185734,0.085120216,0.03606024,0.010217674,0.30083004,0.4962796,0.00037476062],"about_ca_topic_score_codex":0.001565127,"about_ca_topic_score_gemma":0.0033732362,"teacher_disagreement_score":0.017004652,"about_ca_system_score_codex":0.0013990613,"about_ca_system_score_gemma":0.001119767,"threshold_uncertainty_score":0.089930296},"labels":[],"label_agreement":null},{"id":"W4284974261","doi":"10.14778/3551793.3551848","title":"Are updatable learned indexes ready?","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":true,"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; Concurrency; Robustness (evolution); Software deployment; Space (punctuation); Data science; Software engineering; Distributed computing; Operating system","score_opus":0.026758491041296073,"score_gpt":0.24009395904235964,"score_spread":0.21333546800106357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4284974261","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":[0.3633968,0.022354197,0.47088975,0.019404318,0.004094361,0.0012436664,0.007905696,0.07137203,0.039339148],"genre_scores_gemma":[0.6488403,0.0045995754,0.31407735,0.0024724293,0.00091977464,0.00039642514,0.013423552,0.0042915177,0.010979048],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99203134,0.0017564328,0.0008897298,0.0012836772,0.0033264346,0.00071238825],"domain_scores_gemma":[0.9561834,0.010562959,0.0018254367,0.024061821,0.006403298,0.00096315204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070725833,0.0012371158,0.0016393788,0.0015189121,0.0012283548,0.0069376035,0.004508027,0.0015324638,0.006525167],"category_scores_gemma":[0.064660005,0.0010963598,0.0007364938,0.0034062525,0.0022029162,0.024533208,0.00370661,0.0029133165,0.0048400895],"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":[0.002071271,0.00060837035,0.019656643,0.000811663,0.000277145,0.000506092,0.0009396094,0.02866819,0.016903318,0.031251177,0.07640499,0.8219015],"study_design_scores_gemma":[0.0006659661,0.0016793202,0.012348234,0.00076643925,0.00035352918,0.0022290351,0.0030200088,0.48837098,0.089437634,0.12124235,0.2794678,0.00041870872],"about_ca_topic_score_codex":0.0031056392,"about_ca_topic_score_gemma":0.005492373,"teacher_disagreement_score":0.0070725833,"about_ca_system_score_codex":0.0014968885,"about_ca_system_score_gemma":0.0035186468,"threshold_uncertainty_score":0.037403822},"labels":[],"label_agreement":null},{"id":"W4285335450","doi":"10.14778/3494124.3494149","title":"Ember","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"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; Joins; Schema (genetic algorithms); Context (archaeology); Code (set theory); Information retrieval; Theoretical computer science; Artificial intelligence; Programming language","score_opus":0.14569448175843022,"score_gpt":0.38625044455308555,"score_spread":0.24055596279465533,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285335450","genre_codex":"software","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004436628,0.0011767813,0.3334717,0.0023189273,0.0010292926,0.0007672585,0.059703633,0.48294666,0.11414911],"genre_scores_gemma":[0.05525114,0.0017935757,0.3960297,0.0034340778,0.000576931,0.0012961315,0.2602469,0.07820662,0.20316488],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9968394,0.0004563312,0.00028203416,0.00093134365,0.0012310536,0.00025986688],"domain_scores_gemma":[0.99337196,0.0012475577,0.0002680411,0.0036941748,0.0010322666,0.00038606662],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0034408106,0.0018048898,0.0011420785,0.002454439,0.0009337822,0.0047387206,0.004336457,0.0019047309,0.17387053],"category_scores_gemma":[0.014936291,0.0013457085,0.0015037619,0.0022514402,0.0008296848,0.0096372655,0.0075410786,0.002998531,0.1854739],"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":[0.0004884751,0.00016516085,0.001516293,0.00079568214,0.00009612227,0.00023641996,0.00037447797,0.0019327546,0.0057805404,0.033133537,0.67034644,0.28513417],"study_design_scores_gemma":[0.0000941572,0.000087527194,0.00097437965,0.00012576421,0.000031765343,0.00033528547,0.00009984496,0.014804937,0.009047331,0.030308118,0.944025,0.000065955486],"about_ca_topic_score_codex":0.0020466002,"about_ca_topic_score_gemma":0.0032870409,"teacher_disagreement_score":0.82612944,"about_ca_system_score_codex":0.00096914003,"about_ca_system_score_gemma":0.0019141985,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4285337687","doi":"10.14778/3494124.3494125","title":"Enabling SQL-based training data debugging for federated learning","year":2021,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Debugging; Computer science; SQL; SQL injection; Protocol (science); Machine learning; Federated learning; Database; Software engineering; Artificial intelligence; Data mining; Query by Example; Programming language; World Wide Web; Search engine","score_opus":0.09386716120745424,"score_gpt":0.2972390531859824,"score_spread":0.20337189197852817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285337687","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":[0.03841531,0.000148958,0.939566,0.0009851665,0.000062488616,0.0001520832,0.00023168986,0.01973739,0.0007009164],"genre_scores_gemma":[0.68191445,0.00010501915,0.3139822,0.00093211216,0.000042545187,0.00025647169,0.0006671821,0.0009599884,0.0011400477],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98646635,0.005979837,0.001367416,0.0025983069,0.002676567,0.00091152755],"domain_scores_gemma":[0.94586873,0.019117897,0.0027974974,0.028084999,0.0032201037,0.00091078307],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01780811,0.0013183891,0.0013220003,0.0009961706,0.0011561342,0.0031315756,0.005266622,0.0023527995,0.0021093155],"category_scores_gemma":[0.057930846,0.00097451394,0.0012829914,0.000834533,0.003515557,0.010694651,0.0075515425,0.005871793,0.0009281465],"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":[0.0032467067,0.0012312795,0.049176954,0.00047981317,0.00036968285,0.0014055833,0.003012472,0.2871442,0.026593711,0.14189035,0.016813366,0.4686358],"study_design_scores_gemma":[0.000109084984,0.00019402264,0.0011014435,0.000055396365,0.00003784371,0.00034854203,0.00020486662,0.88043344,0.030427184,0.082894936,0.004129854,0.0000633804],"about_ca_topic_score_codex":0.0020774147,"about_ca_topic_score_gemma":0.0018218837,"teacher_disagreement_score":0.01780811,"about_ca_system_score_codex":0.0017613766,"about_ca_system_score_gemma":0.002955308,"threshold_uncertainty_score":0.09417939},"labels":[],"label_agreement":null},{"id":"W4288080130","doi":"10.14778/3407790.3407823","title":"Hypergraph motifs","year":2020,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"National Supercomputing Center, Korea Institute of Science and Technology Information; Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea; Korea Advanced Institute of Science and Technology; Iran Telecommunication Research Center; National Research Foundation of Korea","keywords":"Hypergraph; Computer science; Theoretical computer science; Motif (music); Combinatorics; Mathematics","score_opus":0.010777552812729326,"score_gpt":0.20393807135097922,"score_spread":0.1931605185382499,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288080130","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":[0.031575706,0.0017793908,0.9501931,0.0007542873,0.0001138075,0.00025059114,0.0040750806,0.0036325925,0.0076254103],"genre_scores_gemma":[0.28080243,0.001941721,0.6989393,0.0006375334,0.00025789242,0.00052009226,0.008517799,0.00074641017,0.007636861],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980354,0.0005304133,0.00015512186,0.0007041317,0.00046501512,0.000109924884],"domain_scores_gemma":[0.9932006,0.0036189456,0.00095118774,0.0013838216,0.0006214741,0.00022393986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009751962,0.00095564855,0.0009986003,0.005233152,0.0010066848,0.0035784743,0.0018345435,0.0017207969,0.007229231],"category_scores_gemma":[0.010371959,0.00086558424,0.0013538,0.005785132,0.0014035405,0.007990806,0.0017390712,0.0015863925,0.0016042121],"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":[0.00025437985,0.00019141173,0.013098912,0.0011813758,0.00039617484,0.00044134393,0.0010665271,0.07948206,0.011524917,0.5503453,0.028266978,0.3137506],"study_design_scores_gemma":[0.000033908345,0.00006104533,0.0021192548,0.00014372953,0.00008593302,0.000966339,0.0003062092,0.25941887,0.0044993185,0.68639445,0.045916013,0.000054919976],"about_ca_topic_score_codex":0.002632773,"about_ca_topic_score_gemma":0.003910217,"teacher_disagreement_score":0.007229231,"about_ca_system_score_codex":0.001714471,"about_ca_system_score_gemma":0.0009483434,"threshold_uncertainty_score":0.024184227},"labels":[],"label_agreement":null},{"id":"W4289550979","doi":"10.14778/3551793.3551862","title":"Optimizing differentially-maintained recursive queries on dynamic graphs","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"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; Dataflow; Computation; Overhead (engineering); Graph; Probabilistic logic; Parallel computing; Differential (mechanical device); Theoretical computer science; Algorithm; Programming language","score_opus":0.006113840765845322,"score_gpt":0.20260877880646236,"score_spread":0.19649493804061705,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289550979","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":[0.45470476,0.0004855754,0.530319,0.00049811887,0.00005660763,0.00018066754,0.00048471347,0.009462462,0.00380809],"genre_scores_gemma":[0.82943827,0.00011333669,0.1673791,0.00012345162,0.000027339342,0.00008994128,0.00072192564,0.00043036466,0.0016762568],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983014,0.00033367972,0.00013654662,0.00036733242,0.0005321595,0.00032887168],"domain_scores_gemma":[0.99487525,0.0023896478,0.00032261477,0.0016695895,0.0005984526,0.00014449388],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014353764,0.0006404418,0.00083086727,0.00059456716,0.0006881561,0.0015860428,0.0029057665,0.00066994625,0.0013459796],"category_scores_gemma":[0.006833139,0.0004469022,0.000501299,0.001704797,0.0010575273,0.0043434463,0.0020710328,0.0010391397,0.00028733857],"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":[0.0020481185,0.00049998367,0.010727127,0.0003765768,0.00012787562,0.00048637867,0.0011048486,0.3858628,0.100468375,0.059219033,0.013188185,0.42589074],"study_design_scores_gemma":[0.00010824379,0.00016899518,0.000930892,0.000007949253,0.00004213504,0.000098996155,0.00020354793,0.94541043,0.03184085,0.019101784,0.002063633,0.00002245],"about_ca_topic_score_codex":0.007123776,"about_ca_topic_score_gemma":0.009598757,"teacher_disagreement_score":0.007123776,"about_ca_system_score_codex":0.0013636429,"about_ca_system_score_gemma":0.0016121162,"threshold_uncertainty_score":0.014164627},"labels":[],"label_agreement":null},{"id":"W4289785654","doi":"10.14778/3551793.3551829","title":"On shapley value in data assemblage under independent utility","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Shapley value; Computer science; Benchmark (surveying); Revenue; Computation; Value (mathematics); Assemblage (archaeology); Mathematical economics; Game theory; Economics; Algorithm; Machine learning; Finance","score_opus":0.06382497208843492,"score_gpt":0.2859229840051806,"score_spread":0.22209801191674572,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289785654","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":[0.039535027,0.00051331753,0.94944084,0.00065681123,0.0001061955,0.00018017078,0.0002889414,0.0002872591,0.008991483],"genre_scores_gemma":[0.6282623,0.0011517794,0.36317006,0.00042956718,0.00026039226,0.00037119383,0.0008070428,0.00027799356,0.005269633],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9956868,0.0018813352,0.00019180268,0.00083639496,0.0009803149,0.00042334702],"domain_scores_gemma":[0.99022466,0.006710355,0.0005606439,0.0014474026,0.00065069547,0.00040628985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0058942665,0.0014994543,0.0022921872,0.002027888,0.0017522242,0.0039862436,0.0027008897,0.0015182702,0.007568002],"category_scores_gemma":[0.025852479,0.0006769703,0.0014341534,0.0040320433,0.0038420383,0.0127682965,0.0032316397,0.0039030395,0.0010387033],"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":[0.00026521896,0.00018122049,0.0013800189,0.00023434906,0.000095186704,0.00016007468,0.00047702,0.25902227,0.0018120122,0.65352845,0.005280838,0.07756335],"study_design_scores_gemma":[0.000034876393,0.000052557578,0.00016333236,0.000029065324,0.000016010088,0.000075655415,0.000058345395,0.33207595,0.00064213766,0.66522324,0.0016059559,0.000022852833],"about_ca_topic_score_codex":0.0023624832,"about_ca_topic_score_gemma":0.0020105066,"teacher_disagreement_score":0.007568002,"about_ca_system_score_codex":0.0032982847,"about_ca_system_score_gemma":0.0023841201,"threshold_uncertainty_score":0.031172276},"labels":[],"label_agreement":null},{"id":"W4294903937","doi":"10.14778/3547305.3547324","title":"Misinformation mitigation under differential propagation rates and temporal penalties","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Submodular set function; Bounding overwatch; Computer science; Misinformation; Mathematical optimization; Reachability; Theoretical computer science; Algorithm; Mathematics; Artificial intelligence","score_opus":0.01896916786759804,"score_gpt":0.2716387004994955,"score_spread":0.25266953263189745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294903937","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":[0.10866026,0.00070711254,0.8856777,0.0009756786,0.00005080317,0.00013693578,0.00032619375,0.0007512903,0.0027139494],"genre_scores_gemma":[0.885198,0.00041898031,0.11135883,0.0001951009,0.00009915769,0.00017825486,0.00042375622,0.00013219065,0.0019957612],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960751,0.0017352821,0.0001690149,0.0007796179,0.00081014616,0.00043082482],"domain_scores_gemma":[0.97130615,0.021102207,0.0027133776,0.0030858074,0.0011911492,0.0006013605],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0069583897,0.0015231436,0.0019844736,0.001628569,0.0008901937,0.0025445544,0.0028373324,0.0022156916,0.0017075383],"category_scores_gemma":[0.03582032,0.000859348,0.0011016652,0.0015609969,0.0019229613,0.0063250368,0.002881066,0.0030512868,0.00036685157],"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":[0.00036545197,0.00013135954,0.003399238,0.00020504817,0.000094648116,0.00019210576,0.00027586776,0.88580745,0.0040435204,0.04880964,0.0017991532,0.054876488],"study_design_scores_gemma":[0.000016548549,0.000057642952,0.00043210824,0.000014010443,0.000024390902,0.00008540527,0.000041087194,0.97196704,0.0018674386,0.02495717,0.00052184,0.000015342903],"about_ca_topic_score_codex":0.0036135353,"about_ca_topic_score_gemma":0.0032795381,"teacher_disagreement_score":0.0069583897,"about_ca_system_score_codex":0.0025009045,"about_ca_system_score_gemma":0.0020801984,"threshold_uncertainty_score":0.036799967},"labels":[],"label_agreement":null},{"id":"W4294904134","doi":"10.14778/3547305.3547321","title":"Improving matrix-vector multiplication via lossless grammar-compressed matrices","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Lossless compression; Computer science; Matrix (chemical analysis); Algorithm; Matrix multiplication; Linear algebra; Data compression ratio; Data compression; Theoretical computer science; Mathematics; Image compression; Artificial intelligence; Image processing","score_opus":0.008123292950760374,"score_gpt":0.222613248707195,"score_spread":0.21448995575643465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294904134","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":[0.069741935,0.0006324595,0.91415507,0.00051469763,0.0001481108,0.00013098786,0.00048939773,0.0096247215,0.0045625614],"genre_scores_gemma":[0.37059373,0.00050469977,0.6182547,0.00035887994,0.0001763467,0.00022577867,0.0019636054,0.0009036587,0.007018499],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987173,0.00018591671,0.00007035677,0.00016099634,0.0007443072,0.00012114433],"domain_scores_gemma":[0.9977029,0.0008188065,0.00018202547,0.0007477517,0.0004715554,0.00007704166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006714058,0.0010308675,0.00057863473,0.0011443065,0.0005004141,0.0010909396,0.0011368255,0.0005480724,0.004060425],"category_scores_gemma":[0.0049965107,0.00025869216,0.00043061867,0.0015969899,0.00088942004,0.0031196184,0.0015542014,0.0011386664,0.0019203529],"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":[0.0010241633,0.00034790064,0.0016089373,0.00043566892,0.00007487089,0.00049830787,0.00047741595,0.10054779,0.12282374,0.070163995,0.020889066,0.6811082],"study_design_scores_gemma":[0.00012360334,0.00041720906,0.0006147608,0.00004099345,0.000036778474,0.00057130505,0.00016201925,0.7710693,0.16792466,0.04370766,0.015287629,0.000044076292],"about_ca_topic_score_codex":0.0020212221,"about_ca_topic_score_gemma":0.00314099,"teacher_disagreement_score":0.004060425,"about_ca_system_score_codex":0.0006304985,"about_ca_system_score_gemma":0.0012227811,"threshold_uncertainty_score":0.013583422},"labels":[],"label_agreement":null},{"id":"W4312277017","doi":"10.14778/3551793.3551808","title":"Evaluating persistent memory range indexes","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Memory and Neural Computing","field":"Engineering","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Dram; Computer science; Scalability; Latency (audio); CAS latency; Range (aeronautics); Embedded system; Memory controller; Computer hardware; Operating system; Semiconductor memory; Telecommunications; Engineering","score_opus":0.037182770645599446,"score_gpt":0.2610079709039207,"score_spread":0.22382520025832126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312277017","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":[0.8159396,0.016526116,0.109135844,0.000734514,0.00104772,0.0005620658,0.0029639078,0.011426021,0.04166418],"genre_scores_gemma":[0.9233061,0.002193021,0.064650916,0.00020116777,0.00013752731,0.00023348695,0.0036075811,0.00049907964,0.0051711956],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.995613,0.000603112,0.0004837843,0.00044990174,0.0023986218,0.00045161397],"domain_scores_gemma":[0.99047583,0.0035189248,0.0006099953,0.0021770482,0.0029401213,0.00027806778],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002557223,0.0008145009,0.00067782734,0.0019069106,0.00086760183,0.002399561,0.0026564077,0.00095249346,0.0035817078],"category_scores_gemma":[0.014303263,0.00030591246,0.00033989962,0.0026110501,0.0009455974,0.008166856,0.0020371026,0.0008024731,0.0011522844],"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":[0.0037487147,0.0010858387,0.023828646,0.0041657765,0.00036242683,0.0006201415,0.0011681955,0.07664687,0.13024111,0.03914743,0.04630696,0.672678],"study_design_scores_gemma":[0.0003333147,0.006122679,0.011264533,0.00047475376,0.00047249807,0.0015350253,0.0016311485,0.3976631,0.45612612,0.016732361,0.10739105,0.0002535362],"about_ca_topic_score_codex":0.0017181967,"about_ca_topic_score_gemma":0.0018401886,"teacher_disagreement_score":0.0035817078,"about_ca_system_score_codex":0.0014420596,"about_ca_system_score_gemma":0.001349039,"threshold_uncertainty_score":0.013523996},"labels":[],"label_agreement":null},{"id":"W4312312306","doi":"10.14778/3554821.3554894","title":"Modern techniques for querying graph-structured relations","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"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; Graph database; Theoretical computer science; Query optimization; Distributed computing; Information retrieval","score_opus":0.012329831228488347,"score_gpt":0.23290806312094914,"score_spread":0.22057823189246079,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312312306","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":[0.0034911109,0.0045411247,0.9731663,0.0014001549,0.000152634,0.00028870965,0.0030198775,0.0080908155,0.0058492185],"genre_scores_gemma":[0.053044736,0.0124683175,0.91854334,0.0008785646,0.0004202713,0.00046041506,0.009319064,0.0010322398,0.0038331172],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99422425,0.00082288316,0.0005965268,0.0010402174,0.0030689777,0.00024715753],"domain_scores_gemma":[0.99549073,0.0017234745,0.0002700414,0.0017894589,0.00063953764,0.000086796106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021522413,0.0018596257,0.0017714639,0.005239758,0.001146183,0.005082619,0.0046092616,0.0015569653,0.010118665],"category_scores_gemma":[0.010377114,0.0013370017,0.002194356,0.013376918,0.0016036081,0.012012816,0.0041449266,0.003002394,0.0058956225],"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":[0.00019831867,0.00019461123,0.001646116,0.0027213302,0.0002868239,0.000315734,0.0010848602,0.023249716,0.01975613,0.35555285,0.06527062,0.52972287],"study_design_scores_gemma":[0.000097200464,0.00010721009,0.0009896352,0.0003022836,0.00011945639,0.0013124917,0.0006402634,0.16770935,0.01427056,0.63042885,0.18391125,0.00011146065],"about_ca_topic_score_codex":0.0040946463,"about_ca_topic_score_gemma":0.0056875623,"teacher_disagreement_score":0.010118665,"about_ca_system_score_codex":0.001592321,"about_ca_system_score_gemma":0.0016293577,"threshold_uncertainty_score":0.033850312},"labels":[],"label_agreement":null},{"id":"W4312535937","doi":"10.14778/3551793.3551805","title":"Don't be a tattle-tale","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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; Inference; Computation; Data mining; Information sensitivity; Information retrieval; Computer security; Algorithm; Artificial intelligence","score_opus":0.02477296912415364,"score_gpt":0.24534879267563733,"score_spread":0.2205758235514837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312535937","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":[0.044432655,0.002651173,0.76426893,0.10538601,0.0027328427,0.000354815,0.0013222703,0.0027025293,0.07614885],"genre_scores_gemma":[0.6275264,0.0023635789,0.24026231,0.025894819,0.001551099,0.00075853785,0.001460534,0.0018235953,0.098359056],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99258643,0.0024703573,0.00035419254,0.002293049,0.0014495534,0.0008464182],"domain_scores_gemma":[0.9732675,0.014473607,0.00090711826,0.008474756,0.0019893346,0.0008877903],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009080102,0.0008700519,0.0017031553,0.0011216587,0.0052117878,0.006722107,0.0029498783,0.0040920186,0.022531684],"category_scores_gemma":[0.049844608,0.0010793338,0.002723533,0.0014990809,0.011714482,0.025524756,0.007598502,0.008158378,0.006624053],"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":[0.00026373417,0.00008581422,0.0018316988,0.00019291084,0.00013747824,0.00037545274,0.0014468244,0.00684992,0.0010027974,0.90162355,0.043051947,0.043138016],"study_design_scores_gemma":[0.000039650793,0.000060109367,0.00020259264,0.00006395254,0.000045337245,0.00035295266,0.0003416119,0.019674545,0.0011725604,0.9294212,0.048568204,0.000057205983],"about_ca_topic_score_codex":0.0036662263,"about_ca_topic_score_gemma":0.003180029,"teacher_disagreement_score":0.022531684,"about_ca_system_score_codex":0.0020617582,"about_ca_system_score_gemma":0.0027198896,"threshold_uncertainty_score":0.075376034},"labels":[],"label_agreement":null},{"id":"W4312578340","doi":"10.14778/3554821.3554897","title":"The past, present and future of indexing on persistent memory","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Memory and Neural Computing","field":"Engineering","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Search engine indexing; Computer science; Hash function; Contrast (vision); Data science; Information retrieval; Artificial intelligence; Computer security","score_opus":0.009660911960263854,"score_gpt":0.19781378855625145,"score_spread":0.1881528765959876,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312578340","genre_codex":"methods","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":[0.021976529,0.4362053,0.4478361,0.01585503,0.0040353797,0.00021234874,0.0005742341,0.0036790345,0.069625966],"genre_scores_gemma":[0.20168717,0.36405674,0.3842607,0.00612515,0.008143834,0.00049147074,0.0011305605,0.00093109056,0.03317328],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9975586,0.00047462215,0.00026185514,0.0003405429,0.0010391715,0.00032511388],"domain_scores_gemma":[0.9947208,0.0024559856,0.00021297118,0.0013065665,0.0010484769,0.00025521274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045363684,0.0007706129,0.0010417779,0.0024971885,0.0013746016,0.007094591,0.0036670242,0.0024267961,0.009445797],"category_scores_gemma":[0.008511387,0.001120862,0.00068777247,0.0058509395,0.003913777,0.02710447,0.0036569543,0.0040666056,0.0038395312],"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":[0.00025400924,0.00009827109,0.0006620403,0.0017008231,0.000033840566,0.000110201974,0.0007406867,0.0034065058,0.0055995374,0.42081138,0.02791294,0.53866976],"study_design_scores_gemma":[0.00005623235,0.00036085726,0.0006558079,0.001340833,0.000074167765,0.0009993413,0.0006588605,0.027217189,0.013249847,0.3407302,0.61449087,0.00016578945],"about_ca_topic_score_codex":0.0020525947,"about_ca_topic_score_gemma":0.0014892769,"teacher_disagreement_score":0.009445797,"about_ca_system_score_codex":0.002123834,"about_ca_system_score_gemma":0.0021025774,"threshold_uncertainty_score":0.031599283},"labels":[],"label_agreement":null},{"id":"W4312611216","doi":"10.14778/3551793.3551865","title":"Spatial and temporal constrained ranked retrieval over videos","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; York University","funders":"","keywords":"Computer science; Graph; Matching (statistics); Artificial intelligence; Construct (python library); Pattern recognition (psychology); Window (computing); Data mining; Theoretical computer science; Mathematics","score_opus":0.011406514400310977,"score_gpt":0.24518063879892166,"score_spread":0.23377412439861067,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312611216","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":[0.036649276,0.0010590998,0.9571571,0.0001629328,0.0000342228,0.00015530887,0.0005740439,0.0026086362,0.0015994149],"genre_scores_gemma":[0.46784243,0.00092839735,0.5241456,0.0001420133,0.00012600918,0.00017196956,0.0027967573,0.00025164502,0.003595221],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985744,0.00035221016,0.00010019443,0.000355275,0.00046570908,0.00015219382],"domain_scores_gemma":[0.99841344,0.0005542861,0.00018636971,0.00046149056,0.00031775807,0.00006663208],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008334225,0.0008680898,0.001206272,0.0021703038,0.00044033583,0.0011415146,0.0016309711,0.0008132992,0.00241413],"category_scores_gemma":[0.0038176228,0.00030219107,0.0008094909,0.0039377892,0.00047507999,0.0023439652,0.0011642855,0.00047428178,0.0009682637],"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":[0.00056654075,0.00023779158,0.0018145668,0.0005951435,0.00012395784,0.0002686284,0.00020606739,0.16059454,0.0525738,0.024233505,0.011193545,0.7475919],"study_design_scores_gemma":[0.00005407408,0.00019764091,0.0013966524,0.000014324324,0.000048812555,0.00024368113,0.00013995857,0.95068395,0.01670199,0.026903506,0.0035730803,0.000042381347],"about_ca_topic_score_codex":0.013410996,"about_ca_topic_score_gemma":0.016578378,"teacher_disagreement_score":0.013410996,"about_ca_system_score_codex":0.0009266121,"about_ca_system_score_gemma":0.0012770358,"threshold_uncertainty_score":0.026665866},"labels":[],"label_agreement":null},{"id":"W4312621027","doi":"10.14778/3561261.3561270","title":"TreeLine","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"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; Merge (version control); Associative array; Workload; Key (lock); Point (geometry); Parallel computing; Operating system; Artificial intelligence","score_opus":0.012707166413659191,"score_gpt":0.22132474582873737,"score_spread":0.2086175794150782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312621027","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":[0.03547453,0.005205904,0.15810998,0.0023515366,0.0014530249,0.0007664534,0.07351161,0.42490557,0.29822144],"genre_scores_gemma":[0.18078935,0.003726952,0.21069889,0.0036408412,0.00051635306,0.0009720343,0.25318182,0.045105156,0.30136856],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99944645,0.000048379592,0.000052504583,0.00013016889,0.00022169434,0.00010072456],"domain_scores_gemma":[0.9989317,0.00014408366,0.00006837387,0.00034653445,0.00041380303,0.0000955056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052125804,0.0007108505,0.0005463132,0.00081876916,0.0007240584,0.0026654822,0.002321659,0.0008349948,0.09730415],"category_scores_gemma":[0.0019520127,0.0005553111,0.0005447761,0.001652771,0.00034288527,0.0040342403,0.0018192875,0.0009991275,0.071692206],"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":[0.0009933518,0.0001608028,0.0025508772,0.0008946854,0.00007771868,0.00021041979,0.00041917985,0.0013842438,0.016108586,0.022288658,0.7201733,0.23473819],"study_design_scores_gemma":[0.00014485695,0.00015578946,0.0010104235,0.00007593005,0.000031083604,0.00032088553,0.0001496015,0.008125511,0.014896174,0.008412334,0.9666282,0.00004928839],"about_ca_topic_score_codex":0.003246931,"about_ca_topic_score_gemma":0.005589445,"teacher_disagreement_score":0.09730415,"about_ca_system_score_codex":0.0007351177,"about_ca_system_score_gemma":0.0010914312,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312691443","doi":"10.14778/3554821.3554869","title":"SmartBench","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Benchmark (surveying); Computer science; Question answering; Usability; Cover (algebra); Task (project management); Software deployment; Natural language; Information retrieval; Quality (philosophy); Natural language processing; Artificial intelligence; Software engineering; Human–computer interaction; Systems engineering; Engineering","score_opus":0.014372961389645013,"score_gpt":0.20559853763416233,"score_spread":0.19122557624451733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312691443","genre_codex":"software","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":[0.012956902,0.0024833588,0.21339303,0.0013929046,0.000843357,0.001489803,0.122251615,0.57015604,0.07503302],"genre_scores_gemma":[0.07148253,0.0017575945,0.24619932,0.0017695762,0.00024308586,0.002089349,0.57529485,0.058703825,0.04245991],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971499,0.0007481547,0.00036457076,0.0007594156,0.0007755948,0.00020239853],"domain_scores_gemma":[0.993634,0.0029544844,0.00024699216,0.00155558,0.0013067517,0.0003022793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00327499,0.0018173169,0.0010354827,0.003552251,0.00068371807,0.0034657368,0.0045073424,0.0015957503,0.05794492],"category_scores_gemma":[0.014447127,0.0010860441,0.0015619916,0.003222921,0.00063701026,0.005446407,0.0038457667,0.0018066921,0.041099686],"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":[0.0012416388,0.0003884413,0.0024806727,0.0026706385,0.00012484692,0.00046651333,0.0008909164,0.0056400816,0.008594096,0.03274715,0.69026244,0.25449258],"study_design_scores_gemma":[0.00038500482,0.00033154106,0.0019208969,0.00030153518,0.00007524446,0.00044279077,0.00038275294,0.0491253,0.01721393,0.030996472,0.8987173,0.000107189],"about_ca_topic_score_codex":0.004169803,"about_ca_topic_score_gemma":0.0052753477,"teacher_disagreement_score":0.05794492,"about_ca_system_score_codex":0.0011460982,"about_ca_system_score_gemma":0.001470749,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312701113","doi":"10.14778/3551793.3551847","title":"ConnectorX","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"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; Overhead (engineering); Bridge (graph theory); Database; Process (computing); Interface (matter); Distributed computing; Operating system","score_opus":0.007993055508866007,"score_gpt":0.19490780951306425,"score_spread":0.18691475400419824,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312701113","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":[0.0053043487,0.0014185964,0.19293667,0.0008809072,0.00073713256,0.0007744188,0.022099042,0.68228364,0.09356519],"genre_scores_gemma":[0.06961286,0.0032204876,0.20451187,0.003754862,0.00066421105,0.0018971192,0.26586214,0.20220779,0.24826859],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99639314,0.00048827962,0.00034225298,0.00074949727,0.0015870353,0.0004398288],"domain_scores_gemma":[0.99555296,0.000829061,0.00023475157,0.0019533876,0.001023228,0.00040663395],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003107788,0.0026840498,0.0016016039,0.002301936,0.0012339278,0.0061694128,0.0060127596,0.0018876322,0.15285912],"category_scores_gemma":[0.009569264,0.0017835202,0.0019154405,0.0036001855,0.0009868264,0.009604281,0.008015667,0.0029464513,0.13033104],"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":[0.0013615052,0.00024122788,0.0022635753,0.0011592505,0.00015209756,0.00034726443,0.00037911092,0.0013263345,0.008796874,0.037102748,0.7396827,0.20718738],"study_design_scores_gemma":[0.00017390342,0.00014557858,0.0013345665,0.00011918879,0.00004823802,0.00038166606,0.00009257457,0.0077321553,0.0098891435,0.011303293,0.9686997,0.00007997752],"about_ca_topic_score_codex":0.003236249,"about_ca_topic_score_gemma":0.0032201814,"teacher_disagreement_score":0.15285912,"about_ca_system_score_codex":0.0012607024,"about_ca_system_score_gemma":0.0026697498,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312844330","doi":"10.14778/3551793.3551821","title":"MIDE","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"False positive paradox; Computer science; False positives and false negatives; Bounded function; Context (archaeology); Data mining; True positive rate; Machine learning; Artificial intelligence; Mathematics","score_opus":0.020862250324350327,"score_gpt":0.2353562044626464,"score_spread":0.21449395413829608,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312844330","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022085294,0.004848193,0.6581594,0.0054578283,0.002343719,0.0011271734,0.017714389,0.041843466,0.24642046],"genre_scores_gemma":[0.22638734,0.002963754,0.52330065,0.0059429742,0.0007905522,0.0013214886,0.04671118,0.004872695,0.18770945],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99750215,0.00046885904,0.00017830622,0.0005058231,0.0010608408,0.00028408898],"domain_scores_gemma":[0.9963207,0.00086681615,0.00018258013,0.0016098855,0.00077624724,0.00024368953],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0026181603,0.00068957504,0.0008091083,0.001636871,0.0011615211,0.0036524502,0.0030594117,0.0016629598,0.04220008],"category_scores_gemma":[0.01132307,0.00044724252,0.0007614495,0.0013121645,0.00057968136,0.005067573,0.005088882,0.0020314057,0.028754877],"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":[0.0012761201,0.0003879869,0.0032513363,0.00061134686,0.0000933914,0.00040259774,0.00027574407,0.008490884,0.007676967,0.18332842,0.2614164,0.5327889],"study_design_scores_gemma":[0.0001134354,0.00022191535,0.0011303703,0.00014909693,0.00003359171,0.0012705381,0.00014524281,0.06799895,0.011696599,0.074957766,0.8422139,0.00006848429],"about_ca_topic_score_codex":0.001020232,"about_ca_topic_score_gemma":0.0015546734,"teacher_disagreement_score":0.9577999,"about_ca_system_score_codex":0.0010224778,"about_ca_system_score_gemma":0.0013614511,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312887229","doi":"10.14778/3551793.3551857","title":"Tiresias","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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; Latency (audio); Search engine indexing; Adaptation (eye); Row; Parallel computing; Artificial intelligence; Database","score_opus":0.009267484417925695,"score_gpt":0.19480690833967898,"score_spread":0.1855394239217533,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312887229","genre_codex":"software","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":[0.017660465,0.0028978116,0.33051872,0.0023283192,0.0011787403,0.00090755854,0.019841215,0.5220715,0.10259575],"genre_scores_gemma":[0.31301615,0.0042091636,0.39249387,0.0028911948,0.00065145286,0.0014041736,0.100864924,0.03067664,0.15379255],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982065,0.0001966632,0.000116176,0.0005037574,0.00075387023,0.00022307016],"domain_scores_gemma":[0.9974382,0.000477762,0.00018079122,0.0010474905,0.0006521423,0.00020358138],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017872857,0.0017362193,0.0008667429,0.0010347161,0.00067801116,0.003369396,0.004133083,0.0011488267,0.050830174],"category_scores_gemma":[0.0077971187,0.0009858424,0.0011171381,0.0011033666,0.00054561533,0.0044231946,0.0027850259,0.003069475,0.038962703],"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":[0.0016971799,0.00045216756,0.0062965397,0.00087564444,0.00024797788,0.00031718714,0.00030170934,0.02869581,0.011375725,0.03761979,0.5301471,0.38197324],"study_design_scores_gemma":[0.0002480665,0.0003686341,0.00236619,0.0001655593,0.00011369828,0.0006047152,0.0000950117,0.34426147,0.018553384,0.03190106,0.6011443,0.00017781867],"about_ca_topic_score_codex":0.0054445425,"about_ca_topic_score_gemma":0.005951337,"teacher_disagreement_score":0.050830174,"about_ca_system_score_codex":0.0009937737,"about_ca_system_score_gemma":0.0024186852,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312987091","doi":"10.14778/3565816.3565818","title":"Online schema evolution is (almost) free for snapshot databases","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"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; Schema evolution; Schema migration; Database schema; Schema (genetic algorithms); Snapshot (computer storage); Database; Information retrieval; Semi-structured model; Database design","score_opus":0.02804371284440938,"score_gpt":0.25714350070823433,"score_spread":0.22909978786382496,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312987091","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":[0.052099213,0.0011739982,0.9027015,0.0016941582,0.0004048297,0.00027525966,0.00019817174,0.027310781,0.014142047],"genre_scores_gemma":[0.68614566,0.00076439325,0.2990362,0.0016107216,0.00033484644,0.0002783783,0.0006182994,0.00285807,0.008353362],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99315274,0.0016634944,0.0006840112,0.0012082447,0.002709825,0.0005817625],"domain_scores_gemma":[0.95689046,0.0044767912,0.0015423042,0.034243844,0.001583253,0.0012632849],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004900127,0.0008152549,0.0009752129,0.00072654913,0.0015596056,0.0042120204,0.00343618,0.0015105896,0.0035671943],"category_scores_gemma":[0.017986024,0.0010991765,0.00066745694,0.00090165075,0.0023116462,0.009747918,0.007871645,0.0046505253,0.0018608273],"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":[0.0019835788,0.0008616253,0.010552475,0.00071487436,0.00032068035,0.0010746749,0.0017938386,0.022791566,0.15400454,0.20640907,0.028767098,0.570726],"study_design_scores_gemma":[0.00048805046,0.0014530884,0.0062098294,0.00032322895,0.0003847282,0.0041300473,0.0010208783,0.337835,0.19340695,0.19328257,0.26108092,0.0003847414],"about_ca_topic_score_codex":0.00067919143,"about_ca_topic_score_gemma":0.0009852768,"teacher_disagreement_score":0.004900127,"about_ca_system_score_codex":0.00071341486,"about_ca_system_score_gemma":0.0018507545,"threshold_uncertainty_score":0.025914669},"labels":[],"label_agreement":null},{"id":"W4312989454","doi":"10.14778/3554821.3554864","title":"CERTEM","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Counterfactual thinking; Computer science; Debugging; Order (exchange); Matching (statistics); State (computer science); Artificial intelligence; Programming language; Epistemology; Mathematics; Philosophy","score_opus":0.15111650697305168,"score_gpt":0.36334862928504613,"score_spread":0.21223212231199445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312989454","genre_codex":"software","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012177982,0.0022617765,0.2782329,0.0034081156,0.00087933923,0.00083597074,0.11865297,0.523381,0.06016994],"genre_scores_gemma":[0.09316325,0.0014297562,0.36437622,0.0023793583,0.00025980631,0.00083809457,0.46933094,0.03258862,0.035633955],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99774325,0.0003675203,0.00020631627,0.0007727333,0.00071601465,0.00019419621],"domain_scores_gemma":[0.9949058,0.001396849,0.00029951747,0.002214908,0.0010252248,0.0001577395],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0028172347,0.0015493998,0.00086932804,0.0029349239,0.00082093594,0.0031938353,0.004591544,0.00193911,0.062384203],"category_scores_gemma":[0.015980268,0.0006948407,0.0017017907,0.0023861318,0.0007594683,0.005256001,0.0043635503,0.0020845989,0.032430563],"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":[0.00056020875,0.00019193601,0.0046607507,0.0017814105,0.00020134586,0.00042047305,0.0003416282,0.0104452185,0.0030405608,0.043619007,0.72656506,0.20817235],"study_design_scores_gemma":[0.0002302288,0.00012386474,0.0024836378,0.00024653043,0.000076137985,0.0007067747,0.00019868344,0.11907602,0.012604092,0.045551736,0.8186055,0.00009685053],"about_ca_topic_score_codex":0.005388842,"about_ca_topic_score_gemma":0.008337027,"teacher_disagreement_score":0.9376158,"about_ca_system_score_codex":0.0014439869,"about_ca_system_score_gemma":0.0026848135,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4312990730","doi":"10.14778/3554821.3554858","title":"POEM","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"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; Artificial intelligence; Poetry; Image (mathematics); Convolutional neural network; Inference; Modular design; Bedroom; Deep learning; Pattern recognition (psychology); Machine learning; Art; Geography","score_opus":0.012971825591411225,"score_gpt":0.22596645889845804,"score_spread":0.21299463330704682,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312990730","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023224518,0.0022721959,0.34113228,0.01782939,0.0043514716,0.0005109404,0.0060404376,0.0119862985,0.59265244],"genre_scores_gemma":[0.3186084,0.0028747672,0.17500448,0.0065776245,0.0011094466,0.000585074,0.00913579,0.0031001342,0.48300436],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993686,0.000220704,0.000028046523,0.0001746675,0.00015525577,0.000052843323],"domain_scores_gemma":[0.9986577,0.00057876983,0.00006916021,0.00036390568,0.00022023235,0.00011016518],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0010806035,0.0007730209,0.00030287434,0.0008334689,0.0013239591,0.0024582387,0.0010543704,0.0012742116,0.11612214],"category_scores_gemma":[0.008820917,0.00030339716,0.00052119396,0.0006964416,0.0015473644,0.0061483425,0.002097152,0.0018202454,0.032133132],"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":[0.00019772771,0.000095435906,0.0021963974,0.00035569593,0.000026969708,0.00030616258,0.001437873,0.001858008,0.0012969217,0.3855493,0.32131442,0.2853652],"study_design_scores_gemma":[0.000021507973,0.00006507182,0.00072524225,0.00013224602,0.00001271096,0.00062967755,0.00049068505,0.010958939,0.0015812394,0.13006847,0.85528386,0.000030363832],"about_ca_topic_score_codex":0.0006774537,"about_ca_topic_score_gemma":0.0016164121,"teacher_disagreement_score":0.8838779,"about_ca_system_score_codex":0.00063881354,"about_ca_system_score_gemma":0.0005772669,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4313065965","doi":"10.14778/3551793.3551853","title":"Spooky","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Merge (version control); Computer science; Granularity; Associative array; Data structure; Parallel computing; Operating system; Programming language","score_opus":0.01181195605712601,"score_gpt":0.21574701719651604,"score_spread":0.20393506113939003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313065965","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":[0.08187572,0.0043512573,0.5716651,0.005845852,0.0037247161,0.0010664115,0.0056352345,0.03580888,0.29002687],"genre_scores_gemma":[0.52671874,0.0033952117,0.2036652,0.004577193,0.0007738487,0.0011992407,0.008612829,0.009382167,0.2416757],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961308,0.00040772549,0.00027735712,0.00081418897,0.0016764543,0.0006934116],"domain_scores_gemma":[0.9950051,0.0010558396,0.00040144727,0.0022643495,0.0010119514,0.00026149268],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018333703,0.0010521836,0.0009827989,0.0019575097,0.0032064694,0.0033077898,0.0022666135,0.0016643822,0.048282336],"category_scores_gemma":[0.009566142,0.0008145366,0.0010384866,0.0019951104,0.002738542,0.009653626,0.008285323,0.0021898649,0.016848594],"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":[0.0017418918,0.00020468081,0.0055445763,0.0009403037,0.00011218042,0.0010770148,0.002194074,0.004990342,0.017693464,0.21651947,0.15625066,0.59273124],"study_design_scores_gemma":[0.00012016891,0.0002506619,0.0018371508,0.00034477699,0.00011757568,0.0020074593,0.0009205551,0.022452239,0.03532056,0.09813274,0.8383226,0.00017362543],"about_ca_topic_score_codex":0.0016347147,"about_ca_topic_score_gemma":0.0023212866,"teacher_disagreement_score":0.048282336,"about_ca_system_score_codex":0.0013717285,"about_ca_system_score_gemma":0.0023236144,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4313152108","doi":"10.14778/3561261.3561263","title":"The case for distributed shared-memory databases with RDMA-enabled memory disaggregation","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Remote direct memory access; Scalability; Computer science; Shared memory; Database; Distributed memory; Scaling; Memory management; Operating system; Semiconductor memory","score_opus":0.01966093503013673,"score_gpt":0.2441219501968502,"score_spread":0.22446101516671346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313152108","genre_codex":"commentary","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":[0.040862363,0.02629316,0.36034402,0.43578207,0.0049729743,0.00029049208,0.0005884814,0.0022655297,0.12860093],"genre_scores_gemma":[0.7004373,0.010780916,0.21910839,0.029329067,0.0052109687,0.0006252169,0.00046496443,0.00055217894,0.033490982],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9903875,0.0025511298,0.00050258375,0.002035548,0.0032402552,0.001282995],"domain_scores_gemma":[0.9826972,0.005135534,0.00067448214,0.006654504,0.0027148095,0.0021235093],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01629987,0.0006323472,0.0015308886,0.0007435782,0.0032351022,0.016685713,0.0053349906,0.008966845,0.008421821],"category_scores_gemma":[0.025609609,0.0011934009,0.0011314141,0.0016781728,0.0060246154,0.040984746,0.008085343,0.012114695,0.0028438128],"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":[0.00020016811,0.00008031912,0.0011763709,0.00022681609,0.00003363879,0.0004454913,0.00033942913,0.004524721,0.0010093553,0.92682993,0.03042474,0.034709014],"study_design_scores_gemma":[0.00016474676,0.00012654685,0.00058145745,0.00036541477,0.000055218043,0.0009433603,0.0008368948,0.044178274,0.0020633289,0.6593111,0.29128915,0.00008445206],"about_ca_topic_score_codex":0.0040673786,"about_ca_topic_score_gemma":0.002842707,"teacher_disagreement_score":0.016685713,"about_ca_system_score_codex":0.003473966,"about_ca_system_score_gemma":0.0035012835,"threshold_uncertainty_score":0.08620298},"labels":[],"label_agreement":null},{"id":"W4313172799","doi":"10.14778/3551793.3551826","title":"Finding locally densest subgraphs","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Graph; Graph factorization; Computer science; Combinatorics; Regular polygon; Pruning; Theoretical computer science; Mathematics; Graph power; Biology; Line graph","score_opus":0.009705402500551377,"score_gpt":0.20046701625810603,"score_spread":0.19076161375755465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313172799","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":[0.109198056,0.001268806,0.87756866,0.0008722232,0.00006646331,0.0003762359,0.0028307182,0.0032951562,0.004523651],"genre_scores_gemma":[0.34851038,0.0005967922,0.6324998,0.0003551975,0.00010267775,0.00031208456,0.011108029,0.0007493508,0.0057656635],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99887973,0.0002051044,0.00006108952,0.0004404192,0.00027680447,0.00013685472],"domain_scores_gemma":[0.99692875,0.0014883977,0.00036658937,0.0006039793,0.0004123631,0.00019990596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008482772,0.00125586,0.0018267019,0.0033440161,0.0010439075,0.0015886765,0.0017805963,0.0013901315,0.0040504388],"category_scores_gemma":[0.0065092016,0.0008901958,0.0014556886,0.002899232,0.0009628981,0.00255263,0.0019464059,0.0010708475,0.0012117686],"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":[0.0005797053,0.00038405886,0.01795869,0.0012077701,0.00037689062,0.000906969,0.0010070488,0.24027906,0.024201006,0.046220917,0.043137643,0.6237403],"study_design_scores_gemma":[0.00008365513,0.00011589315,0.003048462,0.000090019385,0.000118728305,0.0006564417,0.00053050875,0.8785716,0.0072766985,0.09777113,0.011704789,0.00003194502],"about_ca_topic_score_codex":0.007617219,"about_ca_topic_score_gemma":0.016005235,"teacher_disagreement_score":0.007617219,"about_ca_system_score_codex":0.00096695736,"about_ca_system_score_gemma":0.0020518377,"threshold_uncertainty_score":0.015145719},"labels":[],"label_agreement":null},{"id":"W4317767745","doi":"10.14778/3570690.3570700","title":"FirmTruss Community Search in Multilayer Networks","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Community structure; Homophily; Theoretical computer science; Approximation algorithm; Mathematical optimization; Algorithm; Mathematics","score_opus":0.01712011195363708,"score_gpt":0.25401919604076983,"score_spread":0.23689908408713276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317767745","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":[0.13300669,0.0006839217,0.8610336,0.0006746625,0.000035937876,0.00014652275,0.00037217315,0.00042932812,0.0036171277],"genre_scores_gemma":[0.7643756,0.0005967716,0.22839874,0.00023672733,0.00004533018,0.0002042039,0.00074602634,0.00011337296,0.0052833157],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991418,0.00026534687,0.000038818536,0.00027255097,0.00015150214,0.0001299816],"domain_scores_gemma":[0.996567,0.0020309286,0.00043022895,0.00038251345,0.00030366814,0.0002857218],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013308687,0.0006673065,0.0009854621,0.0014330032,0.0013029472,0.00160327,0.0017016489,0.001637185,0.0023401491],"category_scores_gemma":[0.007960055,0.0006609392,0.0010714373,0.0015744428,0.0011112489,0.004088699,0.0024452645,0.001415159,0.00035515308],"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":[0.00019233952,0.00009220713,0.0025842316,0.00018775393,0.000079099584,0.00019875867,0.00032506287,0.82882464,0.0024618716,0.114613466,0.0035478738,0.046892628],"study_design_scores_gemma":[0.00001538003,0.00001891352,0.00017988583,0.00001245754,0.000009792154,0.000044978275,0.00005512062,0.9421188,0.00041637296,0.05626442,0.00085648906,0.0000074352893],"about_ca_topic_score_codex":0.00901135,"about_ca_topic_score_gemma":0.012435439,"teacher_disagreement_score":0.00901135,"about_ca_system_score_codex":0.0021171074,"about_ca_system_score_gemma":0.0011152995,"threshold_uncertainty_score":0.017917812},"labels":[],"label_agreement":null},{"id":"W4317767825","doi":"10.14778/3570690.3570704","title":"FILM","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Computer science; Search engine indexing; Overhead (engineering); Data structure; Range query (database); Parallel computing; Auxiliary memory; Database; Distributed computing; Information retrieval; Operating system; Sargable; Search engine; Web search query","score_opus":0.012163401002412973,"score_gpt":0.2179784930603923,"score_spread":0.20581509205797932,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317767825","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03436228,0.014706939,0.3937466,0.0064974544,0.0040806327,0.0016217052,0.047298286,0.098731905,0.39895418],"genre_scores_gemma":[0.17923622,0.012261562,0.38169256,0.004400195,0.0013818502,0.0011059927,0.11641067,0.008973623,0.29453734],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99886835,0.00008724805,0.00008037061,0.0002370802,0.0005822114,0.00014476645],"domain_scores_gemma":[0.9980044,0.00022946818,0.000113013215,0.0007999013,0.0006868435,0.00016637154],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0008191632,0.0008534225,0.0007448505,0.0015729639,0.0011646271,0.003514401,0.0025326973,0.0012632897,0.050325505],"category_scores_gemma":[0.0035212748,0.00058485306,0.0007418515,0.0027404518,0.00049346365,0.0071110046,0.0034192803,0.0015935656,0.035215348],"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":[0.00049588794,0.00015491135,0.0019284324,0.00070892234,0.00005506141,0.0002469355,0.00028836337,0.0024270453,0.012933366,0.03625882,0.38006073,0.56444144],"study_design_scores_gemma":[0.0000547076,0.00015389477,0.0016122491,0.00013487562,0.00003117068,0.0005408264,0.00018042278,0.016124867,0.011068986,0.013160996,0.9568851,0.000051798288],"about_ca_topic_score_codex":0.004847252,"about_ca_topic_score_gemma":0.006252232,"teacher_disagreement_score":0.9496745,"about_ca_system_score_codex":0.00089797395,"about_ca_system_score_gemma":0.0013313125,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4321448317","doi":"10.14778/3574245.3574246","title":"Cache Me If You Can","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"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; Differential privacy; Cache; Workload; Private information retrieval; CPU cache; Process (computing); Differential (mechanical device); Query optimization; Information retrieval; Data mining; Computer network; Computer security; Operating system","score_opus":0.023773982768973453,"score_gpt":0.23969471304651568,"score_spread":0.21592073027754222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321448317","genre_codex":"other","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":[0.024053384,0.009611816,0.072355315,0.09607065,0.009461508,0.0007370734,0.017702721,0.034850754,0.7351568],"genre_scores_gemma":[0.12217555,0.005264511,0.024094174,0.041662496,0.0032631624,0.0004950238,0.009733134,0.0075984257,0.78571343],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989876,0.00019312704,0.000050472194,0.00017830238,0.00036096782,0.00022945409],"domain_scores_gemma":[0.99594975,0.0007934025,0.00029062876,0.0011054513,0.0010827972,0.0007778972],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0011354105,0.0009990138,0.00078151206,0.0011282149,0.0018185913,0.0043278616,0.0015216534,0.0023757163,0.33342937],"category_scores_gemma":[0.012859061,0.00048314393,0.00061064184,0.0012860837,0.0007890896,0.0083687315,0.0042118686,0.0021007503,0.21300589],"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":[0.00032344618,0.00007210387,0.0018172623,0.0001709762,0.000024962312,0.00036560904,0.00063722074,0.00023979422,0.0011208456,0.00999298,0.83708036,0.14815448],"study_design_scores_gemma":[0.000023762928,0.000038901297,0.00051663566,0.000090940885,0.000017223356,0.000520905,0.00046115086,0.00067723636,0.0009094761,0.008229038,0.98847497,0.00003969179],"about_ca_topic_score_codex":0.0034788523,"about_ca_topic_score_gemma":0.004960708,"teacher_disagreement_score":0.33342937,"about_ca_system_score_codex":0.00059912825,"about_ca_system_score_gemma":0.0010955386,"threshold_uncertainty_score":0.9507821},"labels":[],"label_agreement":null},{"id":"W4321448342","doi":"10.14778/3574245.3574273","title":"On Efficient Approximate Queries over Machine Learning Models","year":2022,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Agence Nationale de la Recherche","keywords":"Oracle; Computer science; Heuristic; Quality (philosophy); Random oracle; Machine learning; Artificial intelligence; Data mining; Theoretical computer science; Information retrieval; Programming language","score_opus":0.011615232348690043,"score_gpt":0.2104266141063533,"score_spread":0.19881138175766325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321448342","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":[0.029270358,0.0023143203,0.95765233,0.0025569948,0.00011318053,0.00023835585,0.0008361059,0.0039966553,0.003021801],"genre_scores_gemma":[0.424165,0.001155703,0.5612234,0.0019805522,0.00075582485,0.0006720955,0.0044392627,0.0009271111,0.004681053],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.9835022,0.0077671865,0.0013238045,0.0025087758,0.0038514752,0.0010466704],"domain_scores_gemma":[0.91929406,0.065856285,0.0022001343,0.008974548,0.0027292462,0.00094573357],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014166757,0.0032483465,0.005197742,0.0031498382,0.0014625577,0.006029329,0.0055269944,0.0048785093,0.00760768],"category_scores_gemma":[0.087430686,0.0013773928,0.0020794927,0.0064577847,0.0034432262,0.0138972625,0.006789324,0.004908165,0.002594885],"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":[0.0018764415,0.0007223448,0.005279224,0.0007802875,0.00026761187,0.0003497601,0.0007157454,0.55246025,0.0025199524,0.11561373,0.028126653,0.291288],"study_design_scores_gemma":[0.000085578795,0.00008180733,0.00020875019,0.000029840821,0.000026228412,0.00010662923,0.00010512086,0.8998637,0.0008774767,0.097194865,0.0014020901,0.00001778478],"about_ca_topic_score_codex":0.005233924,"about_ca_topic_score_gemma":0.0065595927,"teacher_disagreement_score":0.014166757,"about_ca_system_score_codex":0.0029966985,"about_ca_system_score_gemma":0.0039095744,"threshold_uncertainty_score":0.07492185},"labels":[],"label_agreement":null},{"id":"W4323343708","doi":"10.14778/3579075.3579080","title":"Change Propagation Without Joins","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"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; Query plan; Latency (audio); Query optimization; Constant (computer programming); Operator (biology); Theoretical computer science; Plan (archaeology); Space (punctuation); Distributed computing; Information retrieval; Sargable; Search engine; Web search query; Telecommunications","score_opus":0.040322326480228064,"score_gpt":0.2605863433371262,"score_spread":0.22026401685689814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323343708","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":[0.0077371295,0.00043968653,0.9821978,0.00045981433,0.00011831453,0.000226772,0.0003277047,0.00523567,0.0032570306],"genre_scores_gemma":[0.22339979,0.000566147,0.7663136,0.0005835231,0.0003354818,0.0002907382,0.0010734772,0.001248972,0.006188314],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934189,0.00074705057,0.000473275,0.0013741947,0.0033867725,0.00059985707],"domain_scores_gemma":[0.99220335,0.002151669,0.00049775746,0.003632263,0.0012742887,0.0002406231],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041432814,0.0011198876,0.0012057648,0.002069676,0.0011246963,0.003889123,0.0047351983,0.0013571455,0.005205759],"category_scores_gemma":[0.010759964,0.00086176535,0.0015128814,0.0027608015,0.0018748199,0.010351041,0.0042137248,0.0033338098,0.0012700568],"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":[0.0005826576,0.00040700307,0.003581031,0.00059119676,0.00016364937,0.00048067668,0.00076023757,0.06489197,0.029851321,0.19518769,0.022112759,0.6813898],"study_design_scores_gemma":[0.00013249334,0.00046584583,0.0013298292,0.00008263711,0.00018949296,0.00071587384,0.0002397372,0.7195623,0.050522406,0.14312704,0.083493315,0.00013904463],"about_ca_topic_score_codex":0.006749473,"about_ca_topic_score_gemma":0.0054418244,"teacher_disagreement_score":0.006749473,"about_ca_system_score_codex":0.0013686123,"about_ca_system_score_gemma":0.0019848733,"threshold_uncertainty_score":0.021912038},"labels":[],"label_agreement":null},{"id":"W4323343774","doi":"10.14778/3579075.3579091","title":"A Hierarchical Grouping Algorithm for the Multi-Vehicle Dial-a-Ride Problem","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Transportation and Mobility Innovations","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Computer science; Latency (audio); Set (abstract data type); State (computer science); Destinations; Algorithm; Point (geometry); Mathematical optimization; Mathematics; Telecommunications","score_opus":0.021500584599856626,"score_gpt":0.24308542913708225,"score_spread":0.22158484453722563,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323343774","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":[0.017246494,0.00021633519,0.9783044,0.00017383302,0.00004299412,0.00021357798,0.00015543938,0.0010319278,0.0026150048],"genre_scores_gemma":[0.14016749,0.00015523002,0.85539657,0.000092218244,0.000028052062,0.0002179329,0.0008051477,0.00020002386,0.0029373479],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991885,0.00018167951,0.000045051806,0.00026984853,0.00016371963,0.00015129596],"domain_scores_gemma":[0.99909854,0.00028144405,0.00011856261,0.000264993,0.00015121685,0.00008530913],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009054457,0.0010748033,0.0012105862,0.0010815442,0.0011611372,0.0008146379,0.0021787393,0.0013070368,0.0043500126],"category_scores_gemma":[0.002348824,0.0005434979,0.0011143087,0.0016660294,0.00054102193,0.0022623828,0.0018440965,0.0013988147,0.0015568954],"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":[0.0002182667,0.0003115666,0.0012560338,0.0002604791,0.000075477394,0.00013430724,0.00039497463,0.61491114,0.008378772,0.023223326,0.0128179705,0.33801764],"study_design_scores_gemma":[0.000055638844,0.00013439219,0.00034546535,0.000016554874,0.000023702183,0.0000905873,0.00013685672,0.974573,0.0019653654,0.017218279,0.005420569,0.000019541076],"about_ca_topic_score_codex":0.0063790055,"about_ca_topic_score_gemma":0.0064499234,"teacher_disagreement_score":0.0063790055,"about_ca_system_score_codex":0.0010312196,"about_ca_system_score_gemma":0.0017158035,"threshold_uncertainty_score":0.014552236},"labels":[],"label_agreement":null},{"id":"W4323343874","doi":"10.14778/3579075.3579086","title":"On the Risks of Collecting Multidimensional Data Under Local Differential Privacy","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Agence Nationale de la Recherche","keywords":"Differential privacy; Computer science; Robustness (evolution); Inference; Data mining; Population; Identification (biology); Hash function; Computer security; Artificial intelligence","score_opus":0.15583339640858118,"score_gpt":0.33671182075033285,"score_spread":0.18087842434175166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323343874","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":[0.14957584,0.0019088056,0.8315223,0.007574876,0.00014809538,0.0003572165,0.0002646481,0.00051075657,0.008137463],"genre_scores_gemma":[0.9401808,0.0005098827,0.0570823,0.0006741636,0.00012053407,0.00023803372,0.00008672673,0.00006773197,0.001039833],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9464346,0.03259893,0.0019926096,0.004322966,0.012563739,0.002087158],"domain_scores_gemma":[0.71168095,0.20601282,0.016680928,0.05513789,0.00864074,0.0018466427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03969971,0.0011222,0.0022587143,0.0018795647,0.0025151896,0.0050372058,0.0032201754,0.0037689053,0.0015393354],"category_scores_gemma":[0.13654743,0.0009813518,0.0013292239,0.002619959,0.0075980844,0.014371957,0.011222697,0.0069281715,0.00041089536],"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":[0.0021920283,0.0003618639,0.024479967,0.0007042072,0.00059802434,0.001339071,0.0033063402,0.2226126,0.009735531,0.6116661,0.004418452,0.11858571],"study_design_scores_gemma":[0.00014318783,0.0006218485,0.0030222286,0.0002710748,0.00019813624,0.0027560948,0.0013070722,0.58465624,0.015557612,0.38639244,0.0049041533,0.00016993974],"about_ca_topic_score_codex":0.0008097729,"about_ca_topic_score_gemma":0.00042765978,"teacher_disagreement_score":0.03969971,"about_ca_system_score_codex":0.003199787,"about_ca_system_score_gemma":0.0024631699,"threshold_uncertainty_score":0.20995468},"labels":[],"label_agreement":null},{"id":"W4366660614","doi":"10.14778/3583140.3583159","title":"NV-SQL: Boosting OLTP Performance with Non-Volatile DIMMs","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"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; SQL; Operating system; Commit; Online transaction processing; Cache; In-Memory Processing; Database; Overhead (engineering); Parallel computing; Database transaction; Transaction processing; Query by Example; World Wide Web","score_opus":0.008316156622858105,"score_gpt":0.189260400132968,"score_spread":0.18094424351010988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366660614","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":[0.763661,0.0031415345,0.12303693,0.00090748264,0.0006097965,0.00033251286,0.0015923659,0.09124927,0.015469127],"genre_scores_gemma":[0.9518063,0.00038329046,0.04149002,0.00036861596,0.00008900452,0.000084050116,0.0016996129,0.0007137361,0.0033653008],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987196,0.00014054573,0.00011329197,0.00023915023,0.0006114754,0.00017592846],"domain_scores_gemma":[0.99759835,0.00041859015,0.00013280059,0.0008075318,0.0007337306,0.00030890672],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010962292,0.0007755691,0.000499923,0.00088422047,0.00052785286,0.0018183828,0.0039431793,0.00050641864,0.0026408639],"category_scores_gemma":[0.003394865,0.00044411488,0.00029413574,0.00093785056,0.00067737146,0.0028267012,0.0025887487,0.00084581546,0.00085775764],"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":[0.0060435403,0.002059147,0.040145967,0.0010521376,0.00038191542,0.0014377588,0.00088093436,0.05221456,0.35452363,0.011331271,0.063578166,0.46635088],"study_design_scores_gemma":[0.0009047096,0.0021430298,0.013120408,0.00005897732,0.00022464641,0.001003448,0.00057296193,0.69634247,0.24083789,0.0079493895,0.036644183,0.00019787689],"about_ca_topic_score_codex":0.0027690742,"about_ca_topic_score_gemma":0.0026988331,"teacher_disagreement_score":0.0039431793,"about_ca_system_score_codex":0.0007994316,"about_ca_system_score_gemma":0.000882878,"threshold_uncertainty_score":0.0088346},"labels":[],"label_agreement":null},{"id":"W4366660902","doi":"10.14778/3583140.3583143","title":"<i> B <sup>link</sup> </i> -hash: An Adaptive Hybrid Index for In-Memory Time-Series Databases","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"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; Hash function; B-tree; Scalability; Hash table; Node (physics); Merkle tree; Parallel computing; Timestamp; Throughput; Tree (set theory); Data structure; Computer network; Cryptographic hash function; Database; Mathematics; Operating system","score_opus":0.025771553727448302,"score_gpt":0.2545432789060886,"score_spread":0.22877172517864028,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366660902","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":[0.25138208,0.0020726186,0.71083635,0.00060103496,0.0003647099,0.0003368383,0.002057378,0.02287848,0.009470521],"genre_scores_gemma":[0.6748664,0.00062024605,0.31525373,0.0003011264,0.00017601959,0.00015373871,0.003114666,0.0006862202,0.0048279203],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996101,0.000038898226,0.000045534085,0.000056522196,0.00020455537,0.000044336317],"domain_scores_gemma":[0.99901175,0.00016560315,0.00009737666,0.00032718238,0.00028277765,0.000115260984],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045648013,0.0004238583,0.0003883107,0.0009197372,0.00045327732,0.0011009906,0.0016385629,0.00038787306,0.001945168],"category_scores_gemma":[0.001605009,0.00023668332,0.00021776586,0.0019769883,0.0003609868,0.002258882,0.0010204979,0.00039039316,0.00095291005],"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":[0.002380871,0.0005708164,0.01632721,0.0004899832,0.0001968541,0.00056555547,0.00041972278,0.04880072,0.26339003,0.015265246,0.045324787,0.60626817],"study_design_scores_gemma":[0.00020598274,0.0012662745,0.005279397,0.000030524272,0.00013873584,0.0010809911,0.000292908,0.72109604,0.21134323,0.008453148,0.05068952,0.00012329621],"about_ca_topic_score_codex":0.0017572072,"about_ca_topic_score_gemma":0.0025458934,"teacher_disagreement_score":0.001945168,"about_ca_system_score_codex":0.00055964076,"about_ca_system_score_gemma":0.0006368631,"threshold_uncertainty_score":0.006507218},"labels":[],"label_agreement":null},{"id":"W4385653228","doi":"10.14778/3603581.3603585","title":"Autonomously Computable Information Extraction","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"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; Information extraction; Perspective (graphical); Mechanism (biology); Data mining; Relation (database); Information retrieval; Artificial intelligence","score_opus":0.01020545676619336,"score_gpt":0.2254807868001845,"score_spread":0.21527533003399113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385653228","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":[0.022067048,0.00018989149,0.9630607,0.0005309041,0.000046322155,0.00012506983,0.0003238888,0.009787342,0.0038688686],"genre_scores_gemma":[0.4180244,0.00025230797,0.5742159,0.0002919889,0.0000878605,0.00028546675,0.0010850374,0.0011086226,0.004648423],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949886,0.0010824332,0.00041174606,0.0013559299,0.0017831115,0.00037823588],"domain_scores_gemma":[0.9858596,0.006676912,0.0009419799,0.005231976,0.0011421874,0.00014738696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024192065,0.0009100305,0.0010972396,0.0013451248,0.0010912364,0.0035933005,0.0024823877,0.0013044438,0.00465021],"category_scores_gemma":[0.020023579,0.0011004866,0.0012502951,0.0015832018,0.0026284943,0.0076741492,0.004436542,0.0019783436,0.0016779604],"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":[0.0007784635,0.00023832354,0.0057284064,0.0008185203,0.00019296411,0.00043410802,0.0008626973,0.12629344,0.05275756,0.2993072,0.012342173,0.5002461],"study_design_scores_gemma":[0.00008694655,0.000097591015,0.0009903444,0.00006422737,0.00009916662,0.0002519953,0.00011475718,0.643613,0.060714673,0.2737583,0.0201339,0.00007520321],"about_ca_topic_score_codex":0.0023150225,"about_ca_topic_score_gemma":0.003073125,"teacher_disagreement_score":0.00465021,"about_ca_system_score_codex":0.0013805572,"about_ca_system_score_gemma":0.0028322695,"threshold_uncertainty_score":0.015556514},"labels":[],"label_agreement":null},{"id":"W4386123456","doi":"10.14778/3611479.3611516","title":"Scaling Up Structural Clustering to Large Probabilistic Graphs Using Lyapunov Central Limit Theorem","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Cluster analysis; Probabilistic logic; Theoretical computer science; Correlation clustering; Computer science; Mathematics; Algorithm; Artificial intelligence","score_opus":0.02037009250056965,"score_gpt":0.27360625782669246,"score_spread":0.2532361653261228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386123456","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":[0.009423119,0.00011350982,0.98850787,0.00016038393,0.000026314829,0.000037079786,0.00003936867,0.00077026803,0.0009220752],"genre_scores_gemma":[0.49698448,0.00038133038,0.49756047,0.0002690245,0.00013122654,0.00029832916,0.00052775117,0.00068254373,0.003164705],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99885285,0.00032068227,0.0000527953,0.00025916297,0.0004084059,0.00010608665],"domain_scores_gemma":[0.9944252,0.0034166088,0.00041944007,0.0006819689,0.00083018607,0.00022667687],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018700092,0.0010926974,0.0011613789,0.0015619135,0.000841764,0.001355565,0.0018625719,0.00088852725,0.0025329504],"category_scores_gemma":[0.013583268,0.00053248025,0.0010269788,0.0014778305,0.0010484557,0.0027600545,0.0026240803,0.00183572,0.0010961446],"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":[0.000045848596,0.00007292816,0.0010009805,0.0001358721,0.000058580703,0.00013363606,0.00012089325,0.8658429,0.003602583,0.059066888,0.002855279,0.0670637],"study_design_scores_gemma":[0.0000021847757,0.0000081822545,0.000051508046,0.0000019081301,0.000002135399,0.000010528259,0.0000063905486,0.981187,0.0002869843,0.018212456,0.00022745236,0.0000033615886],"about_ca_topic_score_codex":0.004093787,"about_ca_topic_score_gemma":0.0050393017,"teacher_disagreement_score":0.004093787,"about_ca_system_score_codex":0.0015772856,"about_ca_system_score_gemma":0.00148746,"threshold_uncertainty_score":0.011444092},"labels":[],"label_agreement":null},{"id":"W4386128170","doi":"10.14778/3611479.3611493","title":"Semi-Oblivious Chase Termination for Linear Existential Rules: An Experimental Study","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Chase; Set (abstract data type); Class (philosophy); Focus (optics); Existential quantification; Existentialism; Theoretical computer science; Implementation; Key (lock); Algorithm; Programming language; Database; Artificial intelligence; Mathematics; Computer security","score_opus":0.0371439558842468,"score_gpt":0.3083710072932785,"score_spread":0.2712270514090317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386128170","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":[0.81725764,0.0024542569,0.13639645,0.0020198624,0.000573833,0.0011311778,0.0018467783,0.0069176564,0.031402305],"genre_scores_gemma":[0.8870313,0.000586687,0.102992356,0.00051745574,0.00013162073,0.00067446096,0.0022343716,0.0011646285,0.004667012],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98861593,0.0042502605,0.0010172743,0.0017230875,0.003072477,0.0013209129],"domain_scores_gemma":[0.88000536,0.08790834,0.0032926807,0.0217573,0.005120391,0.0019159376],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009835738,0.0010469257,0.0015096702,0.0008746893,0.0017294695,0.0023146635,0.0032579605,0.0021574055,0.010541339],"category_scores_gemma":[0.061588284,0.0005583208,0.0010080665,0.0013823793,0.0026327844,0.0067519415,0.0032751192,0.0052437144,0.0021240604],"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":[0.02057628,0.018929454,0.023455346,0.0048454357,0.00087517337,0.0015058967,0.0039244494,0.20894453,0.09081427,0.16079956,0.054522093,0.41080746],"study_design_scores_gemma":[0.0017977697,0.004595242,0.0036427355,0.00027228985,0.00023303337,0.0006556158,0.0012084472,0.81693745,0.061174665,0.09219864,0.017131655,0.00015240959],"about_ca_topic_score_codex":0.002349041,"about_ca_topic_score_gemma":0.0024575547,"teacher_disagreement_score":0.010541339,"about_ca_system_score_codex":0.001847658,"about_ca_system_score_gemma":0.003383743,"threshold_uncertainty_score":0.052016973},"labels":[],"label_agreement":null},{"id":"W4386128184","doi":"10.14778/3611479.3611518","title":"POEM: Pattern-Oriented Explanations of Convolutional Neural Networks","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Convolutional neural network; Poetry; Computer science; Artificial intelligence; Modular design; Classifier (UML); Image (mathematics); Meaning (existential); Pattern recognition (psychology); Deep learning; Artificial neural network; Machine learning; Psychology; Art; Literature","score_opus":0.023780839036322252,"score_gpt":0.24839148151546825,"score_spread":0.224610642479146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386128184","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":[0.04251303,0.00092111883,0.92355424,0.0046829605,0.00018585288,0.0001933569,0.0019689405,0.0026278975,0.023352552],"genre_scores_gemma":[0.65061074,0.00093734084,0.3363829,0.00063305395,0.000088383684,0.00027703147,0.0026242363,0.0002950453,0.008151202],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99958223,0.0001984103,0.000016655727,0.00009741777,0.00008219217,0.00002307471],"domain_scores_gemma":[0.9985506,0.00093296106,0.00012830252,0.00022943378,0.000108855915,0.00004991864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007842574,0.00074355584,0.00017037056,0.0006088597,0.00053201575,0.0010958789,0.0009667916,0.00093343924,0.0106989555],"category_scores_gemma":[0.00670975,0.00029062293,0.000831976,0.00044539903,0.0012791211,0.0037347763,0.0013933975,0.0012676173,0.0007528361],"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":[0.00021965281,0.00012830441,0.0050849607,0.00065655744,0.00014167036,0.0006277597,0.00247655,0.06298498,0.007635436,0.663764,0.027485028,0.22879517],"study_design_scores_gemma":[0.00004531255,0.00007338614,0.0021096882,0.00013130365,0.00004739313,0.00036835592,0.00039824404,0.28705275,0.0051220693,0.65888274,0.045733336,0.000035409426],"about_ca_topic_score_codex":0.0026207243,"about_ca_topic_score_gemma":0.0038698819,"teacher_disagreement_score":0.0106989555,"about_ca_system_score_codex":0.00077729375,"about_ca_system_score_gemma":0.00042238203,"threshold_uncertainty_score":0.035791516},"labels":[],"label_agreement":null},{"id":"W4386528681","doi":"10.14778/3611540.3611548","title":"Towards General and Efficient Online Tuning for Spark","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute","funders":"Tencent","keywords":"Computer science; Overhead (engineering); SPARK (programming language); Generality; Bayesian optimization; Process (computing); Distributed computing; Cloud computing; Computer engineering; Machine learning","score_opus":0.020878018701399263,"score_gpt":0.2481364875198195,"score_spread":0.22725846881842024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386528681","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":[0.003962625,0.00026921034,0.99084395,0.00015380203,0.000030422883,0.00007351678,0.000041023468,0.003416879,0.0012085929],"genre_scores_gemma":[0.22856434,0.00038903882,0.76672816,0.00039205948,0.000096762255,0.00038067772,0.00034445603,0.001709403,0.001395027],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99594826,0.0011264995,0.00018784877,0.0007677485,0.0014682623,0.0005014348],"domain_scores_gemma":[0.9977259,0.00087871443,0.0002102336,0.0006022303,0.00041285128,0.00017007969],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046752035,0.0016084542,0.001607573,0.00082927383,0.00086387614,0.001851661,0.0029133165,0.0013311234,0.0018941523],"category_scores_gemma":[0.011127138,0.00097446016,0.0013039235,0.0009689032,0.0015084593,0.0022914035,0.0031906385,0.0028710642,0.0010164081],"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":[0.00028251176,0.00020157326,0.0014316797,0.00032154046,0.000110422916,0.00013532679,0.00027218636,0.7647256,0.0155110555,0.043985702,0.00925799,0.16376442],"study_design_scores_gemma":[0.00003614592,0.000029057968,0.00009662528,0.000012329948,0.000006330344,0.000026090229,0.0000197569,0.979984,0.0015841384,0.015706418,0.002487338,0.000011804317],"about_ca_topic_score_codex":0.0049365703,"about_ca_topic_score_gemma":0.0046619023,"teacher_disagreement_score":0.0049365703,"about_ca_system_score_codex":0.0011940877,"about_ca_system_score_gemma":0.003878468,"threshold_uncertainty_score":0.02472514},"labels":[],"label_agreement":null},{"id":"W4386768533","doi":"10.14778/3611540.3611599","title":"Demonstration of SPARQL <sup> <i>ML</i> </sup> : An Interfacing Language for Supporting Graph Machine Learning for RDF Graphs","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"SPARQL; Computer science; Named graph; RDF; RDF query language; RDF Schema; Graph; Scripting language; Information retrieval; Query language; Programming language; Web search query; Search engine; Theoretical computer science; Semantic Web; Web query classification","score_opus":0.018095438311099787,"score_gpt":0.27273469682791746,"score_spread":0.2546392585168177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386768533","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":[0.010165342,0.00021988046,0.6596635,0.0014446073,0.0003718983,0.00028371104,0.016869487,0.27353296,0.037448596],"genre_scores_gemma":[0.16038035,0.0008003755,0.67587143,0.0026176688,0.00014117514,0.0008123445,0.069389984,0.057296176,0.032690503],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99934846,0.000114066475,0.000053811043,0.000102390215,0.00031898607,0.00006213823],"domain_scores_gemma":[0.99860495,0.00063180365,0.000046288656,0.0003499652,0.0002576196,0.000109421824],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016882739,0.00087472063,0.00039019153,0.00073785597,0.00046282125,0.0016634316,0.0019696606,0.000932059,0.042289156],"category_scores_gemma":[0.0034812232,0.0006601448,0.00089845585,0.000781912,0.00075541605,0.0032628465,0.002536673,0.0021781786,0.012878311],"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":[0.001236699,0.00062548794,0.00484874,0.001031689,0.00012514638,0.0020006108,0.0012557958,0.028501851,0.039457116,0.09750682,0.6059897,0.21742034],"study_design_scores_gemma":[0.00034274603,0.0001271202,0.0017596667,0.00022094235,0.000033508884,0.0008998725,0.0002779907,0.33898678,0.05547245,0.055553064,0.54620314,0.00012280348],"about_ca_topic_score_codex":0.007743081,"about_ca_topic_score_gemma":0.0094871335,"teacher_disagreement_score":0.042289156,"about_ca_system_score_codex":0.00078651274,"about_ca_system_score_gemma":0.00089803606,"threshold_uncertainty_score":0.14147139},"labels":[],"label_agreement":null},{"id":"W4386768554","doi":"10.14778/3611540.3611616","title":"Web Connector: A Unified API Wrapper to Simplify Web Data Collection","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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; Application programming interface; World Wide Web; Web API; Web application; Web service; Database; Web development; Operating system","score_opus":0.04619606448523445,"score_gpt":0.2756170296515338,"score_spread":0.22942096516629934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386768554","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036689525,0.0001414673,0.6587129,0.00035904185,0.00029935208,0.00079108245,0.001881705,0.327244,0.006901391],"genre_scores_gemma":[0.05508042,0.00066712376,0.7397883,0.0028576101,0.0004815099,0.0030608934,0.023047075,0.1416182,0.03339893],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946274,0.0009091798,0.0007839623,0.00078288716,0.0023412784,0.00055531424],"domain_scores_gemma":[0.9868023,0.0033224868,0.0007123409,0.005117121,0.003040759,0.0010050375],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0068174587,0.0026232954,0.0012840482,0.0037967132,0.0011092902,0.004641606,0.0036728212,0.0019763845,0.016288437],"category_scores_gemma":[0.016736893,0.0022544144,0.0024194815,0.0025198562,0.0011514988,0.007701629,0.007230097,0.0047175293,0.021009117],"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":[0.001504418,0.0010153679,0.006792056,0.0010728298,0.00033697023,0.0031944867,0.0018796978,0.005473885,0.06112909,0.039491802,0.49177197,0.3863374],"study_design_scores_gemma":[0.00039103904,0.00033837865,0.0047713104,0.0003673952,0.00019052596,0.0024603314,0.000278532,0.09699093,0.11648856,0.022537645,0.7546852,0.0005001139],"about_ca_topic_score_codex":0.0023852808,"about_ca_topic_score_gemma":0.0021342074,"teacher_disagreement_score":0.016288437,"about_ca_system_score_codex":0.00080503966,"about_ca_system_score_gemma":0.002595412,"threshold_uncertainty_score":0.05449021},"labels":[],"label_agreement":null},{"id":"W4386768673","doi":"10.14778/3611540.3611542","title":"Taurus MM: Bringing Multi-Master to the Cloud","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Huawei Technologies (Canada)","funders":"","keywords":"Computer science; Cloud computing; Distributed computing; Computer network; Protocol (science); Node (physics); Operating system; Engineering","score_opus":0.02720882880582502,"score_gpt":0.2355862427393772,"score_spread":0.2083774139335522,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386768673","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":[0.12899089,0.0016207244,0.8018354,0.0008439635,0.00084238587,0.000489509,0.00056748313,0.053042572,0.011767045],"genre_scores_gemma":[0.5986301,0.0004932124,0.38879564,0.0003973206,0.00030796442,0.00018026875,0.00091676717,0.0015042495,0.008774516],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983558,0.00022176963,0.00011333281,0.00036241242,0.00067715795,0.00026948735],"domain_scores_gemma":[0.996965,0.00029921083,0.00024620202,0.0016153527,0.00041136882,0.00046286717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014972115,0.00066505204,0.0007994503,0.0008928887,0.00096877635,0.00212323,0.0032982156,0.0005757682,0.0033459838],"category_scores_gemma":[0.0037148125,0.00063574535,0.00048520832,0.0011930054,0.00069566694,0.0036971802,0.0036447227,0.0015234605,0.001047979],"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":[0.0028479344,0.0005541693,0.012414035,0.00029640135,0.00026963325,0.00054543314,0.0007439983,0.04208339,0.07657509,0.04254168,0.06503579,0.7560925],"study_design_scores_gemma":[0.0009783981,0.0013465144,0.00568509,0.00006683479,0.00021289983,0.0009860436,0.00062181696,0.71354973,0.12885273,0.027342413,0.12015297,0.00020454392],"about_ca_topic_score_codex":0.0043251356,"about_ca_topic_score_gemma":0.0045520454,"teacher_disagreement_score":0.0043251356,"about_ca_system_score_codex":0.0008722321,"about_ca_system_score_gemma":0.0022526958,"threshold_uncertainty_score":0.011193454},"labels":[],"label_agreement":null},{"id":"W4386768927","doi":"10.14778/3611540.3611573","title":"Data and AI Model Markets: Opportunities for Data and Model Sharing, Discovery, and Integration","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Data science; Data sharing; Data integration; Big data; Computer science; Emerging markets; Data modeling; Realm; Artificial intelligence; Economics; Data mining; Database; Finance","score_opus":0.44129423908556503,"score_gpt":0.411258332634344,"score_spread":0.030035906451221006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386768927","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":[0.019288922,0.0154571,0.7518568,0.1337261,0.001831677,0.00040114645,0.0009774058,0.001228756,0.07523213],"genre_scores_gemma":[0.32282323,0.017766913,0.6211827,0.012673545,0.004424864,0.0013772374,0.002066453,0.0010076223,0.016677499],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9721957,0.013623825,0.0016561641,0.0027860138,0.008169879,0.0015684711],"domain_scores_gemma":[0.9179285,0.046761114,0.004157921,0.020815734,0.006022468,0.0043142475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04718155,0.0009825259,0.0021615967,0.0037192872,0.0043019424,0.02668073,0.0059297085,0.0073133563,0.017540863],"category_scores_gemma":[0.09272844,0.0016429988,0.002842126,0.00646082,0.009597783,0.07058637,0.022715926,0.011954788,0.0024252846],"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":[0.000034308097,0.00004070055,0.00037728113,0.000076315715,0.000024559611,0.00008620507,0.0003096174,0.0015611694,0.00016204444,0.96692085,0.006225715,0.024181196],"study_design_scores_gemma":[0.000022276015,0.000026221163,0.0001243191,0.00012993069,0.000013590951,0.00010605756,0.00035985932,0.013488653,0.00028928445,0.9274985,0.057907153,0.00003411815],"about_ca_topic_score_codex":0.0018671998,"about_ca_topic_score_gemma":0.0019500933,"teacher_disagreement_score":0.04718155,"about_ca_system_score_codex":0.005066422,"about_ca_system_score_gemma":0.0076423506,"threshold_uncertainty_score":0.24952286},"labels":[],"label_agreement":null},{"id":"W4389315073","doi":"10.14778/3617838.3617842","title":"FedGTA: Topology-Aware Averaging for Federated Graph Learning","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute; HEC Montréal","funders":"","keywords":"Scalability; Computer science; Graph; Distributed computing; Machine learning; Robustness (evolution); Artificial intelligence; Software deployment; Theoretical computer science; Database","score_opus":0.016568994672844187,"score_gpt":0.2504238959387287,"score_spread":0.23385490126588454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389315073","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":[0.0140505545,0.00026428638,0.97261244,0.00024580455,0.00007663649,0.00009324468,0.00031842952,0.01109676,0.0012419446],"genre_scores_gemma":[0.43085974,0.00020937703,0.5613198,0.00047039776,0.00008516292,0.0002979704,0.002549544,0.001323537,0.002884498],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99858165,0.00035828885,0.000062735584,0.0004885463,0.0003671777,0.00014163146],"domain_scores_gemma":[0.99732035,0.0008482336,0.00021136233,0.000985255,0.00043273057,0.00020202351],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018368143,0.0016973467,0.0018674787,0.0013598185,0.0010998312,0.0015922422,0.0043529235,0.0016039265,0.0029042957],"category_scores_gemma":[0.0069673476,0.00068639807,0.0012061099,0.0017977392,0.0009756037,0.003363624,0.0025325986,0.0020990896,0.001187052],"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":[0.00012182556,0.00021338275,0.0018981317,0.00012437417,0.00012453843,0.00009290177,0.00010915438,0.74593884,0.003926016,0.007298883,0.013804492,0.22634755],"study_design_scores_gemma":[0.00000825592,0.000014839866,0.00008597552,0.0000025966567,0.000004823082,0.000015073015,0.000011196144,0.9947803,0.0005954961,0.0038906045,0.0005864962,0.000004441172],"about_ca_topic_score_codex":0.010352937,"about_ca_topic_score_gemma":0.01914595,"teacher_disagreement_score":0.010352937,"about_ca_system_score_codex":0.0018606439,"about_ca_system_score_gemma":0.0021677227,"threshold_uncertainty_score":0.020585358},"labels":[],"label_agreement":null},{"id":"W4389539753","doi":"10.14778/3626292.3626293","title":"Cryptographically Secure Private Record Linkage using Locality-Sensitive Hashing","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Hash function; Locality-sensitive hashing; Locality; Plaintext; Cryptography; Theoretical computer science; Encryption; Differential privacy; Hash table; Computer security; Data mining","score_opus":0.02040080730661333,"score_gpt":0.24468185188577368,"score_spread":0.22428104457916034,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389539753","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":[0.026684362,0.00045532902,0.9643377,0.00081233983,0.00010315861,0.0003399684,0.00027684082,0.003071852,0.003918459],"genre_scores_gemma":[0.6854567,0.0005216208,0.304816,0.00075667,0.00016529785,0.000595152,0.00074343075,0.00035792604,0.0065872897],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.986845,0.0028514252,0.0014269701,0.002015477,0.005568951,0.001292246],"domain_scores_gemma":[0.98166454,0.0028993103,0.0019013941,0.011585645,0.0015026885,0.00044639598],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006301199,0.0007764983,0.0014847316,0.0012653321,0.0019149209,0.004179796,0.005790414,0.002645276,0.004114626],"category_scores_gemma":[0.013944587,0.0010008634,0.0019036952,0.0028007007,0.003448681,0.014553937,0.0110467365,0.0037251154,0.0033908996],"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":[0.0019982425,0.000935132,0.004920434,0.0009175034,0.00038286057,0.0006990973,0.0018918149,0.07274734,0.050617345,0.6046746,0.0128906425,0.24732506],"study_design_scores_gemma":[0.00068439724,0.0011850154,0.0010957855,0.000160794,0.00029552245,0.0021902495,0.0005340605,0.40519163,0.1417673,0.38361746,0.06291392,0.00036381942],"about_ca_topic_score_codex":0.0005433534,"about_ca_topic_score_gemma":0.0003812325,"teacher_disagreement_score":0.006301199,"about_ca_system_score_codex":0.0021597985,"about_ca_system_score_gemma":0.003357368,"threshold_uncertainty_score":0.0333243},"labels":[],"label_agreement":null},{"id":"W4389576335","doi":"10.14778/3626292.3626305","title":"VeLP: Vehicle Loading Plan Learning from Human Behavior in Nationwide Logistics System","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Urban and Freight Transport Logistics","field":"Engineering","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Plan (archaeology); Analytics; Sorting; Process (computing); Container (type theory); Task (project management); Transport engineering; Distribution center; Logistics center; Operations research; Engineering; Data mining; Systems engineering; Business","score_opus":0.041658636348730654,"score_gpt":0.21151845366066302,"score_spread":0.16985981731193237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389576335","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":[0.38721767,0.0015719193,0.58585423,0.0016657589,0.000261514,0.00039470117,0.0057141026,0.011806369,0.005513793],"genre_scores_gemma":[0.9197274,0.00032609073,0.06602873,0.00047962,0.00006338566,0.00032660255,0.008597944,0.000118898344,0.0043313284],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99969316,0.00005659162,0.00001443263,0.00014725776,0.00003873723,0.00004983208],"domain_scores_gemma":[0.9996146,0.00017393679,0.00004592338,0.000053953958,0.00006878871,0.000042825508],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005614025,0.0012239622,0.0007142309,0.0006048497,0.00025434888,0.0004547347,0.0020693138,0.0009158122,0.0018110262],"category_scores_gemma":[0.0015376352,0.00050510064,0.000668179,0.0006129117,0.00032307682,0.000940153,0.0010333217,0.0015470925,0.0005501298],"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":[0.00027757487,0.0006698163,0.018691504,0.00017956678,0.00018260324,0.00021307841,0.00010760098,0.71404,0.0016238288,0.0010630166,0.01364776,0.24930364],"study_design_scores_gemma":[0.0000077364475,0.000038601644,0.0007594185,0.0000036810084,0.000006694265,0.000009744381,0.000011775071,0.99799323,0.0002462685,0.00064667664,0.00027218525,0.0000040583996],"about_ca_topic_score_codex":0.014489062,"about_ca_topic_score_gemma":0.014033187,"teacher_disagreement_score":0.014489062,"about_ca_system_score_codex":0.0007359988,"about_ca_system_score_gemma":0.0009968645,"threshold_uncertainty_score":0.028809428},"labels":[],"label_agreement":null},{"id":"W4391054874","doi":"10.14778/3632093.3632117","title":"The Art of Latency Hiding in Modern Database Engines","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"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; Latency (audio); Interleaving; CAS latency; Speedup; Parallel computing; Operating system; Scheduling (production processes); CPU cache; Cache; Semiconductor memory; Memory controller","score_opus":0.020143839000601394,"score_gpt":0.24488364605150129,"score_spread":0.22473980705089988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391054874","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":[0.18074588,0.06319427,0.6609125,0.009531217,0.001825237,0.00039770332,0.0010488083,0.04328354,0.03906087],"genre_scores_gemma":[0.7192002,0.015291633,0.23440813,0.0040940857,0.0009560934,0.00022134591,0.0010503017,0.004116586,0.020661583],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9942766,0.0006843054,0.0004232004,0.0008348006,0.0030801864,0.0007010364],"domain_scores_gemma":[0.9907206,0.0019580342,0.0005866531,0.0053187013,0.0010855045,0.00033050668],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037980967,0.0013683197,0.00093559304,0.0011981438,0.0016354043,0.0050387066,0.005459676,0.0014649503,0.0049415524],"category_scores_gemma":[0.010634068,0.0015401307,0.00088059955,0.0024467723,0.0025343264,0.015484684,0.0039695674,0.004025926,0.0028394838],"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":[0.0029747211,0.00066623854,0.0076261256,0.0019116319,0.00032561645,0.00036314823,0.0016032885,0.030111415,0.1468446,0.12623152,0.05075032,0.63059145],"study_design_scores_gemma":[0.0005915252,0.0019847676,0.0047594593,0.00068136194,0.0005784296,0.0018249439,0.0010364269,0.17336322,0.29107955,0.13897638,0.38444752,0.0006763746],"about_ca_topic_score_codex":0.005567097,"about_ca_topic_score_gemma":0.0036702221,"teacher_disagreement_score":0.005567097,"about_ca_system_score_codex":0.0018169475,"about_ca_system_score_gemma":0.0035308914,"threshold_uncertainty_score":0.020086527},"labels":[],"label_agreement":null},{"id":"W4391054882","doi":"10.14778/3632093.3632109","title":"Caerus: Low-Latency Distributed Transactions for Geo-Replicated Systems","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"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; Latency (audio); Distributed transaction; Database transaction; Transaction processing; Workload; Online transaction processing; Distributed computing; Protocol (science); Distributed database; Database; Compensating transaction; Computer network; Operating system","score_opus":0.014421848795280147,"score_gpt":0.23148567763324981,"score_spread":0.21706382883796968,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391054882","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":[0.07564382,0.0032954072,0.8151418,0.00090835727,0.0004744689,0.0011296746,0.001404785,0.086374514,0.01562705],"genre_scores_gemma":[0.63734776,0.0009839163,0.34596255,0.00033719288,0.00016952456,0.0006724569,0.0025920272,0.0019111532,0.010023436],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99609166,0.0008501893,0.00034656673,0.00041580948,0.0019409353,0.0003547596],"domain_scores_gemma":[0.9942572,0.001311474,0.0004612371,0.0025965665,0.0009819415,0.0003915986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026329109,0.00095255015,0.0008535782,0.001072289,0.0011927487,0.0025126417,0.0041627563,0.0010908279,0.004661451],"category_scores_gemma":[0.010616519,0.00064537395,0.00038451247,0.0013927408,0.0010591238,0.003305456,0.0021963448,0.0014853687,0.001522304],"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":[0.003661236,0.0005732158,0.007282654,0.001357412,0.00031650037,0.0010335793,0.00077551877,0.14034066,0.11998829,0.10110574,0.09426791,0.5292972],"study_design_scores_gemma":[0.000584564,0.00069435843,0.0021244856,0.0000794836,0.000089359164,0.00061246654,0.00020601384,0.8169493,0.0665776,0.024042588,0.0878953,0.00014455401],"about_ca_topic_score_codex":0.006364225,"about_ca_topic_score_gemma":0.0060472125,"teacher_disagreement_score":0.006364225,"about_ca_system_score_codex":0.0012277594,"about_ca_system_score_gemma":0.0028112598,"threshold_uncertainty_score":0.015594065},"labels":[],"label_agreement":null},{"id":"W4391054883","doi":"10.14778/3632093.3632118","title":"MOSER: Scalable Network Motif Discovery Using Serial Test","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Motif (music); Computer science; Scalability; Graph; Theoretical computer science; Cluster analysis; Network motif; Data mining; Artificial intelligence; Complex network; Database","score_opus":0.015416777603827437,"score_gpt":0.24724935576746154,"score_spread":0.23183257816363412,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391054883","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":[0.042083286,0.00043389326,0.93145394,0.00057975407,0.00014442964,0.00029329525,0.0019789701,0.020956026,0.0020764484],"genre_scores_gemma":[0.38059583,0.00021542986,0.60704786,0.0005195547,0.00016912153,0.0006604471,0.0064263456,0.001053737,0.0033116448],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9975708,0.00081931986,0.00013272134,0.0006328662,0.0006542281,0.00019010293],"domain_scores_gemma":[0.989096,0.0070822784,0.0008135099,0.0018294653,0.0007678951,0.00041090391],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026527327,0.0014658167,0.0019462457,0.00294775,0.0008204588,0.0014773823,0.0039007508,0.0016312809,0.00584463],"category_scores_gemma":[0.019207597,0.00069606316,0.001560083,0.0021299466,0.00095783314,0.0029495864,0.002527155,0.0018345236,0.001841133],"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":[0.0020000956,0.000733017,0.020660212,0.00077100436,0.00062371895,0.0013268832,0.00027702792,0.24859229,0.021986106,0.046191186,0.04397929,0.61285925],"study_design_scores_gemma":[0.00009545242,0.00008624394,0.0005250935,0.000008270157,0.000020497238,0.00016294677,0.000025542211,0.9725008,0.0030108872,0.02177145,0.0017761253,0.000016696398],"about_ca_topic_score_codex":0.0029700554,"about_ca_topic_score_gemma":0.0067532975,"teacher_disagreement_score":0.00584463,"about_ca_system_score_codex":0.00087329856,"about_ca_system_score_gemma":0.0018943802,"threshold_uncertainty_score":0.01955223},"labels":[],"label_agreement":null},{"id":"W4391054937","doi":"10.14778/3632093.3632096","title":"Blocker and Matcher Can Mutually Benefit: A Co-Learning Framework for Low-Resource Entity Resolution","year":2023,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; University of Toronto","funders":"","keywords":"Benchmark (surveying); Computer science; Resource (disambiguation); Noise (video); Selection (genetic algorithm); Machine learning; Artificial intelligence; Resolution (logic); Data mining","score_opus":0.07733140174241149,"score_gpt":0.3638194075327134,"score_spread":0.2864880057903019,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391054937","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":[0.013516396,0.0004946823,0.9809168,0.00055108406,0.00003366843,0.00009979177,0.00007368843,0.0022963623,0.0020175646],"genre_scores_gemma":[0.5547828,0.00038778156,0.43276218,0.0008577046,0.000121366735,0.0002491354,0.00060998026,0.00034678783,0.009882308],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966414,0.0012700416,0.00012334183,0.0010025357,0.00061055634,0.00035216153],"domain_scores_gemma":[0.9957962,0.0018860537,0.00038072816,0.0009946006,0.00065556896,0.0002868274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006412088,0.001559198,0.0016711736,0.0012831985,0.0012527114,0.0019623104,0.0054007596,0.0034559353,0.003134022],"category_scores_gemma":[0.0105550205,0.0009331186,0.00092440227,0.0013747398,0.0018281848,0.006155999,0.005761973,0.004021928,0.0018645943],"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":[0.0006967189,0.0008197749,0.006063588,0.00018873381,0.00025956865,0.00045000936,0.00092825416,0.41153347,0.007687685,0.038313862,0.011719022,0.52133924],"study_design_scores_gemma":[0.000023457274,0.00007619999,0.00023786884,0.000013077759,0.000034395227,0.00007893422,0.000055702494,0.9795758,0.003080169,0.014192465,0.0026124795,0.00001954139],"about_ca_topic_score_codex":0.004745875,"about_ca_topic_score_gemma":0.007418619,"teacher_disagreement_score":0.006412088,"about_ca_system_score_codex":0.0011113496,"about_ca_system_score_gemma":0.0021587845,"threshold_uncertainty_score":0.03391075},"labels":[],"label_agreement":null},{"id":"W4396601617","doi":"10.14778/3648160.3648187","title":"AeonG: An Efficient Built-in Temporal Support in Graph Databases","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"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 database; Database; Graph; Theoretical computer science","score_opus":0.020735606360406958,"score_gpt":0.26751256505131277,"score_spread":0.24677695869090582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396601617","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":[0.049040258,0.006372926,0.856475,0.0014118971,0.00048380237,0.00066193816,0.014455715,0.06573944,0.005359046],"genre_scores_gemma":[0.26986894,0.0029315997,0.6911725,0.00088505534,0.00022862882,0.00047628756,0.028453406,0.0015624928,0.0044210884],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.997528,0.00030532063,0.00035800776,0.00058778055,0.0010796668,0.00014116206],"domain_scores_gemma":[0.99363744,0.0008581755,0.0003552745,0.0038859497,0.00091772986,0.00034545298],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019651467,0.0011129805,0.0013472763,0.0036003839,0.0008500204,0.0035628239,0.0056851846,0.0009373467,0.0028486173],"category_scores_gemma":[0.009798744,0.00089421,0.001211187,0.0064705117,0.00056147954,0.011152656,0.004389247,0.0013038459,0.001543176],"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":[0.0019157314,0.0006551513,0.0111011835,0.0012501772,0.00045672382,0.00050082995,0.0007077223,0.037339002,0.034789864,0.046406377,0.09848747,0.76638985],"study_design_scores_gemma":[0.00043519546,0.00069366646,0.004773089,0.00013639122,0.00032473286,0.0015780157,0.0008156126,0.73991454,0.042334918,0.06726614,0.14144833,0.00027943283],"about_ca_topic_score_codex":0.009742966,"about_ca_topic_score_gemma":0.010355107,"teacher_disagreement_score":0.009742966,"about_ca_system_score_codex":0.0009225484,"about_ca_system_score_gemma":0.001979907,"threshold_uncertainty_score":0.019372523},"labels":[],"label_agreement":null},{"id":"W4396628099","doi":"10.14778/3648160.3648165","title":"How Can We Train Deep Learning Models Across Clouds and Continents? An Experimental Study","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Deep learning; Geology; Data science; Computer science; Artificial intelligence","score_opus":0.03507132555401607,"score_gpt":0.2938870684634521,"score_spread":0.258815742909436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396628099","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":[0.95716083,0.0014529733,0.026252968,0.0033342158,0.0004015319,0.00017798068,0.0019285106,0.001107528,0.008183528],"genre_scores_gemma":[0.9748503,0.0002697666,0.019135093,0.00048950507,0.00008788016,0.00009988973,0.003103646,0.00015998178,0.001803927],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99724907,0.0012904503,0.0001703787,0.0006281903,0.00030719713,0.00035470977],"domain_scores_gemma":[0.9903803,0.005287206,0.00050633907,0.0023898499,0.0010047064,0.0004316557],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007725714,0.0014945762,0.0009617405,0.00078934495,0.0009336985,0.0016643916,0.0016594209,0.0020769956,0.0027590822],"category_scores_gemma":[0.01859002,0.00046305754,0.0009936558,0.0012277237,0.0013665993,0.0046841716,0.0017649108,0.0028672703,0.0007882764],"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":[0.0034072157,0.0033433572,0.047769386,0.00041085834,0.00075543486,0.00034897705,0.00025291834,0.764809,0.0038268955,0.005374302,0.019033365,0.1506683],"study_design_scores_gemma":[0.00032248252,0.0011192287,0.008329098,0.000057266425,0.0001358674,0.00012152704,0.0005190059,0.97277856,0.005779208,0.0066006035,0.0041951765,0.00004201593],"about_ca_topic_score_codex":0.01440319,"about_ca_topic_score_gemma":0.014811338,"teacher_disagreement_score":0.01440319,"about_ca_system_score_codex":0.0015335988,"about_ca_system_score_gemma":0.001179978,"threshold_uncertainty_score":0.04085797},"labels":[],"label_agreement":null},{"id":"W4399208560","doi":"10.14778/3659437.3659455","title":"Differentially Private Data Generation with Missing Data","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Missing data; Synthetic data; Differential privacy; Computer science; Data mining; Data quality; Ground truth; Process (computing); Data modeling; Algorithm; Machine learning; Engineering; Database","score_opus":0.08668987133323494,"score_gpt":0.2909003034966938,"score_spread":0.20421043216345888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399208560","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":[0.035379637,0.00020557597,0.96177113,0.0007793351,0.000046745747,0.000115222916,0.00034991888,0.00040142753,0.00095104607],"genre_scores_gemma":[0.7573719,0.00017262499,0.23973514,0.0004903528,0.00006685988,0.0002565108,0.00094566843,0.00009933224,0.0008616163],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98920935,0.006329082,0.0005372912,0.0013900863,0.0020693406,0.00046495005],"domain_scores_gemma":[0.93386316,0.04156884,0.0033276863,0.017650798,0.0028785903,0.0007109247],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015990086,0.00072553655,0.0011067126,0.0008321441,0.0008811821,0.0020007493,0.0026433293,0.0017312453,0.0013394061],"category_scores_gemma":[0.058223873,0.0005080013,0.0009941933,0.0014622464,0.0023538487,0.003975042,0.0041299555,0.0029257745,0.0003943981],"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":[0.0012737806,0.0002872397,0.012221035,0.00036034905,0.00020121393,0.0005091346,0.0008768094,0.64823174,0.012362559,0.15747395,0.0055604526,0.16064174],"study_design_scores_gemma":[0.00007191369,0.00012981019,0.00057564233,0.000032606727,0.000021473914,0.0003244028,0.0001246822,0.87564176,0.010341519,0.110465765,0.0022428827,0.000027524604],"about_ca_topic_score_codex":0.0005482364,"about_ca_topic_score_gemma":0.00053335086,"teacher_disagreement_score":0.015990086,"about_ca_system_score_codex":0.0013375819,"about_ca_system_score_gemma":0.0015413197,"threshold_uncertainty_score":0.084564686},"labels":[],"label_agreement":null},{"id":"W4399208580","doi":"10.14778/3659437.3659439","title":"Accelerating String-Key Learned Index Structures via Memoization-Based Incremental Training","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Computer science; Memoization; Bottleneck; Leverage (statistics); Key (lock); Matrix decomposition; Theoretical computer science; Parallel computing; Artificial intelligence; Machine learning; Embedded system; Operating system; Eigenvalues and eigenvectors","score_opus":0.03452865552063113,"score_gpt":0.26440748166450906,"score_spread":0.22987882614387795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399208580","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":[0.1029036,0.001456555,0.8429489,0.0004096483,0.00036589146,0.00023300856,0.00096267817,0.042843983,0.007875793],"genre_scores_gemma":[0.5372015,0.0004879992,0.44994304,0.00046958835,0.00010976449,0.00031305125,0.0029093295,0.0007243096,0.00784138],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996847,0.000025414458,0.000025070483,0.00008812884,0.00012060452,0.000056097753],"domain_scores_gemma":[0.9992269,0.0001972367,0.000063965264,0.00027065116,0.0001905655,0.00005067806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032660546,0.00076261285,0.0007396293,0.00051569316,0.000403486,0.00076652976,0.0030117263,0.0005729964,0.0039442806],"category_scores_gemma":[0.002373518,0.0003546165,0.0004394201,0.00084811874,0.0004690791,0.003050634,0.0012659288,0.0013099225,0.0017165486],"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":[0.0004977796,0.00037402782,0.0038037905,0.00034044578,0.00010169661,0.00026323646,0.00024340715,0.14010434,0.035968766,0.011834785,0.030139772,0.7763279],"study_design_scores_gemma":[0.000036848112,0.00014259756,0.00035756163,0.0000101941505,0.00002394099,0.000101804966,0.00003203854,0.96684664,0.02276483,0.004541242,0.0051272125,0.000015131798],"about_ca_topic_score_codex":0.004285426,"about_ca_topic_score_gemma":0.006922049,"teacher_disagreement_score":0.004285426,"about_ca_system_score_codex":0.0007485799,"about_ca_system_score_gemma":0.0016199542,"threshold_uncertainty_score":0.013194919},"labels":[],"label_agreement":null},{"id":"W4401353058","doi":"10.14778/3675034.3675040","title":"Incremental Sliding Window Connectivity over Streaming Graphs","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"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; Search engine indexing; Sliding window protocol; Computation; Leverage (statistics); Latency (audio); Window (computing); Stream processing; Throughput; Graph; Theoretical computer science; Data mining; Parallel computing; Algorithm; Artificial intelligence","score_opus":0.012101642950259544,"score_gpt":0.2308971289352785,"score_spread":0.21879548598501897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401353058","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":[0.32322684,0.0011937782,0.6634788,0.00049291604,0.00008711869,0.0002193847,0.0011805715,0.0057401247,0.0043804506],"genre_scores_gemma":[0.8412493,0.00044173113,0.1550122,0.00006485287,0.000092366005,0.0001215347,0.0013569843,0.00022225767,0.0014387438],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99916613,0.00011981921,0.000054626275,0.00020033932,0.0003184732,0.00014056233],"domain_scores_gemma":[0.9967542,0.0016434401,0.000398868,0.00064591353,0.00037977347,0.00017777029],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006495852,0.0005765972,0.0007281264,0.0015236093,0.00056549313,0.0013399522,0.0016402011,0.00047338323,0.0014216942],"category_scores_gemma":[0.0061121644,0.00034466997,0.00040306462,0.0028108815,0.00064165756,0.004054977,0.0012560982,0.0005781664,0.00030648874],"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":[0.0011951225,0.0002736353,0.010858359,0.0004699334,0.0001309005,0.00094506587,0.0009596936,0.53214216,0.06272421,0.07174402,0.011480427,0.30707648],"study_design_scores_gemma":[0.000025802014,0.000105107545,0.0009405907,0.000008316797,0.000019654182,0.00013992925,0.0001122151,0.96759105,0.00933685,0.01950081,0.0022064277,0.000013253801],"about_ca_topic_score_codex":0.007530583,"about_ca_topic_score_gemma":0.0056459415,"teacher_disagreement_score":0.007530583,"about_ca_system_score_codex":0.0009417891,"about_ca_system_score_gemma":0.0007323533,"threshold_uncertainty_score":0.014973462},"labels":[],"label_agreement":null},{"id":"W4401353529","doi":"10.14778/3675034.3675050","title":"DEX: Scalable Range Indexing on Disaggregated Memory","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Search engine indexing; Scalability; Range (aeronautics); Computer science; Parallel computing; Information retrieval; Database; Engineering","score_opus":0.01133561931545859,"score_gpt":0.2271684346161253,"score_spread":0.2158328153006667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401353529","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":[0.14323203,0.006406331,0.78670186,0.0006783484,0.00038609977,0.00037965953,0.0033328237,0.036859874,0.022023024],"genre_scores_gemma":[0.54559034,0.0015738178,0.43568414,0.00044273635,0.00015162976,0.0003088389,0.006426105,0.0009898838,0.008832618],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992951,0.00007350729,0.00007202533,0.00009037232,0.00037329085,0.000095558215],"domain_scores_gemma":[0.9984446,0.0003087462,0.0001046241,0.0007325537,0.00030852103,0.00010101617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045826047,0.0005680234,0.0007851438,0.0013009643,0.00056551874,0.0012698089,0.0015411262,0.00044299712,0.003611249],"category_scores_gemma":[0.002787534,0.0002967041,0.00030482362,0.0032278725,0.0003854373,0.0034964038,0.002389834,0.00067727844,0.0017513726],"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":[0.0011656752,0.00025963536,0.003939004,0.0004378997,0.00009754077,0.00033031098,0.00038188373,0.04408745,0.06987282,0.025080595,0.055286624,0.7990605],"study_design_scores_gemma":[0.00032555076,0.0005777157,0.0037030745,0.000107738015,0.00007455744,0.0010326682,0.0005399312,0.77231133,0.08203579,0.047272433,0.091913275,0.000105961626],"about_ca_topic_score_codex":0.0030842875,"about_ca_topic_score_gemma":0.004327853,"teacher_disagreement_score":0.003611249,"about_ca_system_score_codex":0.00050009985,"about_ca_system_score_gemma":0.00081278745,"threshold_uncertainty_score":0.012080789},"labels":[],"label_agreement":null},{"id":"W4402042375","doi":"10.14778/3681954.3681998","title":"Optimizing Video Queries with Declarative Clues","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; University of Toronto","funders":"","keywords":"Computer science; Leverage (statistics); Query optimization; Limiting; Query expansion; Query language; Information retrieval; Domain (mathematical analysis); Web query classification; Extensibility; Data mining; Web search query; Machine learning; Search engine","score_opus":0.011642662368546322,"score_gpt":0.2542588442781897,"score_spread":0.24261618190964335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402042375","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":[0.0829329,0.0014630905,0.8941281,0.0006346617,0.00007174181,0.0002631419,0.0012066031,0.014857715,0.0044420823],"genre_scores_gemma":[0.45125255,0.000585342,0.5398177,0.00031764654,0.00007025651,0.0001949006,0.0028605943,0.001225289,0.003675718],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9980877,0.0004805023,0.00020473996,0.00042309836,0.0006284059,0.00017548483],"domain_scores_gemma":[0.99772507,0.0012193322,0.00012302998,0.0004985728,0.00033756156,0.00009647786],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017368607,0.0012471142,0.001168267,0.00085865764,0.0005158003,0.002334113,0.0020284595,0.0010984177,0.0028521246],"category_scores_gemma":[0.008250245,0.00053910847,0.0007263768,0.0013596851,0.000707818,0.0036891687,0.0024647452,0.0013479424,0.00088132476],"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":[0.0021783663,0.00054558733,0.0071852747,0.0007039881,0.00014248786,0.00046738243,0.00081344193,0.19584805,0.07227458,0.043990318,0.030788582,0.64506197],"study_design_scores_gemma":[0.000086674285,0.00017713672,0.00059505383,0.000025007752,0.00003668316,0.00019194829,0.00033565005,0.9368278,0.0296747,0.022207296,0.009803741,0.00003831073],"about_ca_topic_score_codex":0.00554331,"about_ca_topic_score_gemma":0.009290199,"teacher_disagreement_score":0.00554331,"about_ca_system_score_codex":0.00094167877,"about_ca_system_score_gemma":0.0014831729,"threshold_uncertainty_score":0.011022091},"labels":[],"label_agreement":null},{"id":"W4402042395","doi":"10.14778/3681954.3682002","title":"DDS: DPU-Optimized Disaggregated Storage","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"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; Operating system; Latency (audio); Throughput; Server; Embedded system; Overhead (engineering)","score_opus":0.008606006174710003,"score_gpt":0.21326284760540754,"score_spread":0.20465684143069754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402042395","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":[0.30728397,0.006580437,0.58607584,0.0011678789,0.0012547162,0.00032944305,0.0044857548,0.04986611,0.04295586],"genre_scores_gemma":[0.8497251,0.000893882,0.13325153,0.00044386572,0.00013664928,0.00013211436,0.0031244212,0.0006564622,0.011635977],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994967,0.000038823582,0.00003485431,0.00008948883,0.00024625356,0.00009381166],"domain_scores_gemma":[0.99955446,0.000043779208,0.00004126509,0.0001860658,0.00012672259,0.000047699596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023214746,0.0004660786,0.0005463595,0.00069029606,0.0006029196,0.0012479839,0.0014868177,0.0002711324,0.0042190985],"category_scores_gemma":[0.000659339,0.0002910455,0.00025055892,0.001298557,0.00041442786,0.0016596391,0.0016834105,0.0005453109,0.0011002346],"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":[0.0017034853,0.00037161066,0.008895034,0.0007582867,0.00015917246,0.000566868,0.00051364984,0.09826458,0.20203657,0.037541408,0.10847658,0.54071283],"study_design_scores_gemma":[0.00020271697,0.0004448273,0.004494353,0.00005970717,0.00009142861,0.00086325494,0.00023834527,0.6352869,0.19699655,0.01564962,0.1455315,0.00014072399],"about_ca_topic_score_codex":0.002962922,"about_ca_topic_score_gemma":0.0036611874,"teacher_disagreement_score":0.0042190985,"about_ca_system_score_codex":0.0010825976,"about_ca_system_score_gemma":0.001157993,"threshold_uncertainty_score":0.014114261},"labels":[],"label_agreement":null},{"id":"W4402043623","doi":"10.14778/3681954.3682026","title":"Blueprinting the Cloud: Unifying and Automatically Optimizing Cloud Data Infrastructures with BRAD","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","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":"Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Cloud computing; Computer science; Blueprint; Operating system; Engineering","score_opus":0.016380022218180846,"score_gpt":0.24502677028409897,"score_spread":0.22864674806591812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402043623","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":[0.27481413,0.0024407476,0.6706065,0.00090731133,0.00013451588,0.0006403705,0.0006931291,0.04036662,0.009396693],"genre_scores_gemma":[0.37406224,0.00035008392,0.6209478,0.00025646907,0.000014186257,0.00015349427,0.0009749384,0.0011705098,0.002070254],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982132,0.00056204986,0.00010288159,0.00030019463,0.00057567656,0.00024601264],"domain_scores_gemma":[0.99770725,0.0010612179,0.00025280134,0.00060887635,0.00025482493,0.00011488445],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002198945,0.0020217588,0.0009332817,0.0013532727,0.00050482893,0.0016765696,0.0018812796,0.0010176764,0.0016517336],"category_scores_gemma":[0.0048048226,0.0012457086,0.0010958961,0.0010928217,0.0010657244,0.0029202867,0.0016689777,0.0016616724,0.0006851096],"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":[0.00034552568,0.00041712762,0.009309129,0.00038098585,0.00008555908,0.00020753624,0.00027687443,0.6316691,0.021874035,0.009335623,0.007139113,0.31895947],"study_design_scores_gemma":[0.00003021579,0.00010101067,0.00034680244,0.000011872009,0.00001928282,0.000029488412,0.0000604187,0.9886878,0.005307218,0.003081123,0.0023102267,0.000014572771],"about_ca_topic_score_codex":0.0093350755,"about_ca_topic_score_gemma":0.018294215,"teacher_disagreement_score":0.0093350755,"about_ca_system_score_codex":0.0017956131,"about_ca_system_score_gemma":0.0038966748,"threshold_uncertainty_score":0.018561482},"labels":[],"label_agreement":null},{"id":"W4402043624","doi":"10.14778/3681954.3682027","title":"Aleph Filter: To Infinity in Constant Time","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Caching and Content Delivery","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Infinity; Constant (computer programming); Aleph; Mathematics; Filter (signal processing); Mathematical analysis; Physics; Computer science; Programming language; Particle physics","score_opus":0.010547553289931642,"score_gpt":0.21081831526496186,"score_spread":0.20027076197503021,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402043624","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":[0.0102986945,0.00095031934,0.9454395,0.0011817096,0.00031935915,0.00018273488,0.0009181191,0.027127512,0.013581971],"genre_scores_gemma":[0.12048478,0.0007460403,0.84959877,0.0016289908,0.00047105155,0.0006185319,0.00198211,0.0047289855,0.01974076],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9943897,0.0008672194,0.00040478702,0.0011859178,0.0024179365,0.0007343113],"domain_scores_gemma":[0.9857496,0.0052835597,0.00057417597,0.006078322,0.0019383698,0.0003760042],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005165143,0.0021294907,0.0025341157,0.0026985263,0.0021574066,0.0059351,0.004413158,0.0025104687,0.018893803],"category_scores_gemma":[0.028731473,0.0011145454,0.0019040178,0.0033367404,0.0028460745,0.014321688,0.0069827293,0.0033371972,0.010597553],"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":[0.00189218,0.00027242763,0.0030347796,0.00059661665,0.00012206751,0.000326798,0.00072776806,0.029902248,0.013052123,0.23638134,0.08615798,0.6275336],"study_design_scores_gemma":[0.0002733208,0.00023349958,0.0004496086,0.00018106936,0.00010880933,0.0006341659,0.00024750843,0.45355374,0.022272974,0.43730554,0.08461997,0.00011971766],"about_ca_topic_score_codex":0.0055117565,"about_ca_topic_score_gemma":0.006422203,"teacher_disagreement_score":0.018893803,"about_ca_system_score_codex":0.0030945044,"about_ca_system_score_gemma":0.0044766665,"threshold_uncertainty_score":0.06320608},"labels":[],"label_agreement":null},{"id":"W4404181113","doi":"10.14778/3685800.3685876","title":"Demonstration of DB-GPT: Next Generation Data Interaction System Empowered by Large Language Models","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"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","score_opus":0.21580710100271508,"score_gpt":0.38459209005308537,"score_spread":0.1687849890503703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404181113","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035727136,0.00030416393,0.46749005,0.001866289,0.00055322476,0.000517428,0.016619492,0.45354816,0.023374036],"genre_scores_gemma":[0.2567695,0.0004809885,0.6205059,0.0026413377,0.00014370315,0.0013379568,0.04766537,0.0453557,0.02509949],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99869686,0.0003151153,0.000126911,0.0002400231,0.00049759704,0.0001235274],"domain_scores_gemma":[0.99802846,0.00085532444,0.0000815705,0.00044868057,0.00030045238,0.0002854605],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025554954,0.0010060037,0.0005441627,0.0006689114,0.0005470565,0.0017314844,0.002585617,0.0010258602,0.020150563],"category_scores_gemma":[0.004960103,0.0006276479,0.0012415786,0.00062481227,0.0007587886,0.0035544795,0.003795193,0.0024749788,0.00992982],"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":[0.0049560266,0.00140434,0.010442886,0.0019857946,0.00042292388,0.0046953936,0.004040794,0.016188266,0.07589249,0.048879463,0.5554882,0.27560353],"study_design_scores_gemma":[0.0009837793,0.0005820134,0.004304231,0.00023123181,0.00015445828,0.002873347,0.0004442308,0.36982146,0.09150041,0.032522466,0.4961833,0.0003991268],"about_ca_topic_score_codex":0.0047968538,"about_ca_topic_score_gemma":0.0031425252,"teacher_disagreement_score":0.020150563,"about_ca_system_score_codex":0.00069015037,"about_ca_system_score_gemma":0.0017825535,"threshold_uncertainty_score":0.06741035},"labels":[],"label_agreement":null},{"id":"W4404181204","doi":"10.14778/3685800.3685811","title":"Db2une: Tuning Under Pressure via Deep Learning","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Ferroelectric and Negative Capacitance Devices","field":"Engineering","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"Computer science; Artificial intelligence; Deep learning; Psychology","score_opus":0.006234418694740766,"score_gpt":0.19458057697407258,"score_spread":0.18834615827933182,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404181204","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":[0.14100634,0.0012007807,0.8151998,0.00089732313,0.00016558019,0.00014029376,0.0007154555,0.033508502,0.0071659684],"genre_scores_gemma":[0.81833446,0.00025885907,0.17553711,0.0006811695,0.00004934507,0.00014615346,0.0009797276,0.0007867177,0.0032264958],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994671,0.00009265481,0.000024172434,0.00016734505,0.00016927971,0.00007948718],"domain_scores_gemma":[0.99913126,0.0003959658,0.00006471018,0.00018975533,0.00015271995,0.00006561409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00092454674,0.000871386,0.0006130052,0.00034871462,0.0003195437,0.0009488333,0.0022022254,0.0008012345,0.0024400346],"category_scores_gemma":[0.004082021,0.0004518078,0.00032402072,0.00041570555,0.0005455969,0.001707544,0.00107374,0.0017609986,0.0007814815],"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":[0.0004463411,0.00048728057,0.006532404,0.00017755602,0.00009364534,0.000121951605,0.00013135519,0.59352237,0.0227143,0.005519175,0.019977586,0.35027608],"study_design_scores_gemma":[0.000014782477,0.000028634235,0.00013739726,0.0000032285973,0.0000038022042,0.000010896901,0.0000076189017,0.99508494,0.002344238,0.0016311811,0.0007285262,0.0000047095946],"about_ca_topic_score_codex":0.007825398,"about_ca_topic_score_gemma":0.014320403,"teacher_disagreement_score":0.007825398,"about_ca_system_score_codex":0.0010208888,"about_ca_system_score_gemma":0.0015621998,"threshold_uncertainty_score":0.015559733},"labels":[],"label_agreement":null},{"id":"W4404181435","doi":"10.14778/3685800.3685894","title":"Demonstration of the VeriEQL Equivalence Checker for Complex SQL Queries","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Programming language; Computer science; Equivalence (formal languages); SQL; SQL/PSM; Database; Stored procedure; Information retrieval; Query by Example; Mathematics; Discrete mathematics; Web search query","score_opus":0.02881900024655679,"score_gpt":0.2677268580327518,"score_spread":0.23890785778619503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404181435","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04982857,0.0002558432,0.88954806,0.0020705117,0.00040429094,0.00040246444,0.0013601447,0.048801776,0.0073284646],"genre_scores_gemma":[0.52368224,0.00025897112,0.46136978,0.00222539,0.00016994508,0.00044267447,0.0028094486,0.0047568753,0.0042845826],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9787724,0.0063833673,0.0018746763,0.0026707589,0.008314859,0.0019839485],"domain_scores_gemma":[0.9415306,0.03611302,0.0017466496,0.013439334,0.0065159965,0.0006543318],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012074768,0.0011602057,0.00095624424,0.001375407,0.0015239959,0.0038099685,0.004953372,0.0023933358,0.008913528],"category_scores_gemma":[0.048192494,0.0014127292,0.0023106313,0.0010978978,0.0055511408,0.009109594,0.009176444,0.005701418,0.0022834092],"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":[0.0038143133,0.0022180942,0.014934069,0.0015476377,0.0004337449,0.003075618,0.002696302,0.084019184,0.16613698,0.3663023,0.06386595,0.2909558],"study_design_scores_gemma":[0.00091275835,0.0006307596,0.0011784786,0.00020536319,0.00017783849,0.0006168956,0.00028106655,0.56772244,0.262283,0.13266282,0.033113807,0.0002147417],"about_ca_topic_score_codex":0.0069583734,"about_ca_topic_score_gemma":0.0055294014,"teacher_disagreement_score":0.012074768,"about_ca_system_score_codex":0.002280068,"about_ca_system_score_gemma":0.005235538,"threshold_uncertainty_score":0.06385827},"labels":[],"label_agreement":null},{"id":"W4404988426","doi":"10.14778/3749646.3749657","title":"TabulaX: Leveraging Large Language Models for Multi-Class Table Transformations","year":2025,"lang":"en","type":"preprint","venue":"Proceedings of the VLDB Endowment","topic":"Big Data Technologies and Applications","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Table (database); Class (philosophy); Computer science; Programming language; Natural language processing; Theoretical computer science; Artificial intelligence; Database","score_opus":0.20772441500396746,"score_gpt":0.38232798555009745,"score_spread":0.17460357054613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404988426","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":[0.0049084346,0.0003395845,0.92024267,0.00045548557,0.0001106509,0.00022456964,0.0039519486,0.06835503,0.0014115607],"genre_scores_gemma":[0.073786914,0.00045947958,0.8981178,0.0005523128,0.0000835646,0.00060599385,0.01520021,0.008956855,0.002236887],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9959084,0.0014606578,0.00047289397,0.0008760854,0.0010970971,0.0001849562],"domain_scores_gemma":[0.986997,0.0070384345,0.0007308892,0.00407854,0.0009604462,0.00019465586],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051582847,0.0017073164,0.0009623235,0.0022298405,0.0009113707,0.0049305586,0.0030206293,0.0010471591,0.0072131604],"category_scores_gemma":[0.02480926,0.0010820435,0.0032004106,0.0024966386,0.0013493468,0.0071246964,0.004568659,0.0033073544,0.004424698],"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":[0.0007702241,0.0005031665,0.008139218,0.0015075776,0.0004142469,0.00057014445,0.0020181458,0.11126847,0.014927553,0.12804537,0.09813705,0.63369876],"study_design_scores_gemma":[0.00013051665,0.00010790357,0.00056353817,0.0001908525,0.00007644325,0.00022340684,0.0002718546,0.76642114,0.018205106,0.13084206,0.082875684,0.00009149112],"about_ca_topic_score_codex":0.0060846354,"about_ca_topic_score_gemma":0.010521096,"teacher_disagreement_score":0.0072131604,"about_ca_system_score_codex":0.001471956,"about_ca_system_score_gemma":0.0035493725,"threshold_uncertainty_score":0.027279973},"labels":[],"label_agreement":null},{"id":"W4407385487","doi":"10.14778/3696435.3696437","title":"CUTTANA: Scalable Graph Partitioning for Faster Distributed Graph Databases and Analytics","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Scalability; Graph database; Analytics; Graph; Database; Theoretical computer science","score_opus":0.022006714325906684,"score_gpt":0.24875675482104148,"score_spread":0.2267500404951348,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407385487","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":[0.03453922,0.0013460517,0.7483149,0.00084075896,0.00042943325,0.0005752716,0.00473387,0.20057999,0.008640471],"genre_scores_gemma":[0.17021865,0.00057844137,0.79673684,0.00044853063,0.00008958737,0.0006699946,0.017725358,0.008500356,0.0050322064],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990451,0.00013155224,0.00006784372,0.00029268366,0.00036880243,0.00009395528],"domain_scores_gemma":[0.99746644,0.0007032862,0.00013605147,0.0010104183,0.00042267793,0.00026116584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010431956,0.0021654323,0.0009335494,0.0018565942,0.0012049588,0.002099722,0.004115294,0.0011168518,0.0091362195],"category_scores_gemma":[0.005790655,0.0011061508,0.0013273471,0.0023294075,0.00078754063,0.003984512,0.0031799434,0.0023094874,0.0035196904],"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":[0.0014076408,0.00072190724,0.004611598,0.0013564584,0.00052030757,0.00047991754,0.0010014429,0.14299291,0.042845577,0.030369049,0.24087282,0.5328204],"study_design_scores_gemma":[0.0004622455,0.00015505937,0.0013753276,0.00004504548,0.00005652254,0.00022793705,0.00025470473,0.9152912,0.018148351,0.027857568,0.036058865,0.00006714407],"about_ca_topic_score_codex":0.010100594,"about_ca_topic_score_gemma":0.019944701,"teacher_disagreement_score":0.010100594,"about_ca_system_score_codex":0.0013104441,"about_ca_system_score_gemma":0.0021019452,"threshold_uncertainty_score":0.030563712},"labels":[],"label_agreement":null},{"id":"W4407694628","doi":"10.14778/3704965.3704979","title":"Eventual Durability","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Durability; Materials science; Composite material","score_opus":0.006778615071000851,"score_gpt":0.2150393170096381,"score_spread":0.20826070193863724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407694628","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":[0.028829275,0.0055379528,0.67213345,0.003910351,0.0025238495,0.00088705274,0.0043098493,0.019504836,0.26236334],"genre_scores_gemma":[0.47825027,0.0065786405,0.2273194,0.0037453144,0.001745601,0.0012151061,0.009982706,0.0053782295,0.26578477],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99482036,0.0006935298,0.0006658228,0.0009052729,0.0021962894,0.0007187517],"domain_scores_gemma":[0.9865827,0.0018150694,0.00082894927,0.0065116966,0.0035893028,0.0006723001],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037694334,0.0009838364,0.0009972198,0.0012571687,0.0024898993,0.0059283404,0.0044557345,0.0021988812,0.037769437],"category_scores_gemma":[0.013647975,0.00094443286,0.0015123484,0.0013845995,0.0021824143,0.012975083,0.0071521997,0.0030386078,0.017005114],"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":[0.00047799604,0.00016378907,0.004089482,0.0011765448,0.000115265175,0.0010104297,0.0013523778,0.009171531,0.009406742,0.64713746,0.0801681,0.24573027],"study_design_scores_gemma":[0.000072276554,0.00024314671,0.00067276024,0.00030958749,0.00009831287,0.00168837,0.00040844607,0.01583264,0.009978369,0.22326304,0.74734473,0.00008835094],"about_ca_topic_score_codex":0.0017192258,"about_ca_topic_score_gemma":0.0018647921,"teacher_disagreement_score":0.037769437,"about_ca_system_score_codex":0.0014897394,"about_ca_system_score_gemma":0.0027290105,"threshold_uncertainty_score":0.12635136},"labels":[],"label_agreement":null},{"id":"W4408061126","doi":"10.14778/3705829.3705850","title":"Making CRDTs Not So Eventual","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Blockchain Technology Applications and Security","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; University of Toronto","funders":"","keywords":"Computer science","score_opus":0.02123369697432146,"score_gpt":0.2670379170096228,"score_spread":0.24580422003530136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408061126","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":[0.097972274,0.0005968872,0.865031,0.004002684,0.0010819483,0.00037975502,0.000319459,0.01320083,0.017415164],"genre_scores_gemma":[0.69771296,0.000646929,0.2790518,0.0016963682,0.0004422726,0.000450699,0.0007261527,0.0029705612,0.016302258],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98631334,0.0037603686,0.0012670389,0.0019504023,0.00513071,0.00157811],"domain_scores_gemma":[0.95328796,0.00773169,0.0032793172,0.02731214,0.0070850844,0.0013038227],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012265138,0.0007273434,0.0008257043,0.0007705527,0.0018920746,0.005287783,0.003207027,0.0020520955,0.0037713214],"category_scores_gemma":[0.036534127,0.0010687419,0.0010107297,0.00095970614,0.004194192,0.011229478,0.0085620135,0.004514335,0.0022239105],"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":[0.0015274812,0.00041888305,0.012133078,0.0009626742,0.00021970499,0.0011938371,0.0033122965,0.06174547,0.088611335,0.5276653,0.0316144,0.2705956],"study_design_scores_gemma":[0.00047593765,0.00081427017,0.0014895842,0.00028461005,0.00025546245,0.0011294882,0.0012932947,0.25699785,0.14127065,0.3278641,0.26786673,0.00025807548],"about_ca_topic_score_codex":0.0017376009,"about_ca_topic_score_gemma":0.0022230707,"teacher_disagreement_score":0.012265138,"about_ca_system_score_codex":0.0010481647,"about_ca_system_score_gemma":0.0034452805,"threshold_uncertainty_score":0.06486499},"labels":[],"label_agreement":null},{"id":"W4408061162","doi":"10.14778/3705829.3705830","title":"RED: Effective Trajectory Representation Learning with Comprehensive Information","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Trajectory; Representation (politics); Computer science; Artificial intelligence; Human–computer interaction; Political science; Physics; Politics","score_opus":0.00837476942388125,"score_gpt":0.22392301270177595,"score_spread":0.2155482432778947,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408061162","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":[0.01829392,0.00042676297,0.97269404,0.00021496268,0.0000711058,0.000079365636,0.0007094361,0.0063918736,0.0011184961],"genre_scores_gemma":[0.3952116,0.00052220625,0.5858401,0.0005150308,0.00014774693,0.00028238105,0.008052205,0.00060849526,0.008820211],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992385,0.00017251792,0.000040782063,0.00031126136,0.00015358937,0.00008341483],"domain_scores_gemma":[0.9988966,0.00034529544,0.00012628076,0.00036131893,0.00020222453,0.0000683046],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010718345,0.0014545122,0.0010368549,0.001101882,0.00040122066,0.00088960945,0.0026163468,0.001373565,0.0027596767],"category_scores_gemma":[0.0032245037,0.00057033694,0.0011618636,0.0012478068,0.0005636004,0.0028650232,0.0020221833,0.0021564818,0.0015542575],"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":[0.00028191396,0.0003037952,0.0023371533,0.00020210604,0.00016686149,0.00014033,0.00014521822,0.36384773,0.009310083,0.010468743,0.017701352,0.59509474],"study_design_scores_gemma":[0.000011628548,0.00005247787,0.00021710874,0.000008486413,0.000009483869,0.000031006442,0.000013052004,0.9931664,0.0018812963,0.0033767666,0.0012228381,0.000009482051],"about_ca_topic_score_codex":0.0072143804,"about_ca_topic_score_gemma":0.008909013,"teacher_disagreement_score":0.0072143804,"about_ca_system_score_codex":0.0009063856,"about_ca_system_score_gemma":0.0015155156,"threshold_uncertainty_score":0.014344752},"labels":[],"label_agreement":null},{"id":"W4408061173","doi":"10.14778/3705829.3705846","title":"Can Graph Reordering Speed Up Graph Neural Network Training? An Experimental Study","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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; Graph; Artificial neural network; Artificial intelligence; Machine learning; Theoretical computer science","score_opus":0.02788617619331354,"score_gpt":0.26391288883681896,"score_spread":0.23602671264350542,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408061173","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":[0.8935978,0.0045379396,0.068616904,0.0013678655,0.00093742274,0.0002964783,0.0022544526,0.017253002,0.011138149],"genre_scores_gemma":[0.8946356,0.0010360088,0.09365695,0.00055138924,0.00007195051,0.00013654213,0.005676776,0.001057773,0.003176962],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992907,0.00015727675,0.000059890768,0.00022196313,0.00013336669,0.0001367483],"domain_scores_gemma":[0.99640924,0.0019824768,0.00022491501,0.00077280897,0.00044011534,0.00017038886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012439573,0.0015207531,0.00069795526,0.00084330735,0.0005502077,0.00068782596,0.0017889513,0.0010952151,0.0033008626],"category_scores_gemma":[0.008673406,0.000460078,0.0006317204,0.0009800018,0.00057836535,0.0027643105,0.00074320444,0.001660815,0.001146267],"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":[0.002327074,0.0014831186,0.011209928,0.0014676511,0.0003002814,0.00042342828,0.00026723606,0.51455504,0.022043088,0.0031712104,0.030213507,0.41253847],"study_design_scores_gemma":[0.00013805515,0.0006147325,0.0024953636,0.00005625209,0.000079821824,0.00010927791,0.00013756072,0.9662321,0.022618959,0.0032360025,0.004255134,0.000026623986],"about_ca_topic_score_codex":0.011123602,"about_ca_topic_score_gemma":0.019799577,"teacher_disagreement_score":0.011123602,"about_ca_system_score_codex":0.0009887872,"about_ca_system_score_gemma":0.0010194412,"threshold_uncertainty_score":0.022117674},"labels":[],"label_agreement":null},{"id":"W4408061184","doi":"10.14778/3705829.3705833","title":"Efficient and Effective Algorithms for A Family of Influence Maximization Problems with A Matroid Constraint","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Matroid; Constraint (computer-aided design); Matroid partitioning; Weighted matroid; Maximization; Oriented matroid; Combinatorics; Algorithm; Computer science; Mathematics; Mathematical optimization; Graphic matroid; Geometry","score_opus":0.006558692355830441,"score_gpt":0.23206969491545273,"score_spread":0.22551100255962228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408061184","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":[0.007581735,0.0008023354,0.9829487,0.000759504,0.00010164977,0.00028963128,0.00029731003,0.0013169603,0.0059022265],"genre_scores_gemma":[0.10397521,0.0006693524,0.8882559,0.00052402366,0.0002752062,0.00077257527,0.0010981342,0.0005280558,0.0039014842],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99710447,0.0009299465,0.00016202484,0.0007806127,0.00063186715,0.00039105653],"domain_scores_gemma":[0.99132603,0.006210982,0.0005634218,0.0008673078,0.0006736343,0.0003586455],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031657908,0.0033445263,0.0028983157,0.0018516188,0.0013521777,0.003349632,0.004536448,0.0035379063,0.0077600586],"category_scores_gemma":[0.015320706,0.001370984,0.0028566455,0.0027222782,0.0014991623,0.00467786,0.0036998931,0.0044697654,0.002813869],"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":[0.0003982326,0.00048664797,0.0019575243,0.0007880044,0.00021190563,0.00025005284,0.00036307258,0.6015408,0.0033517012,0.06902322,0.030699033,0.29092985],"study_design_scores_gemma":[0.000073692616,0.00004717888,0.00010841993,0.00002152394,0.000020210597,0.000084158084,0.000043486314,0.95953035,0.00050447945,0.03736673,0.0021872332,0.000012612058],"about_ca_topic_score_codex":0.004750364,"about_ca_topic_score_gemma":0.007962108,"teacher_disagreement_score":0.0077600586,"about_ca_system_score_codex":0.0026999088,"about_ca_system_score_gemma":0.0031891055,"threshold_uncertainty_score":0.025959969},"labels":[],"label_agreement":null},{"id":"W4409231750","doi":"10.14778/3712221.3712227","title":"How Reliable are Streams? End-to-End Processing-Guarantee Validation and Performance Benchmarking of Stream Processing Systems","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Benchmarking; Stream processing; STREAMS; Computer science; End-to-end principle; Real-time computing; Distributed computing; Computer network; Business","score_opus":0.012008114402951214,"score_gpt":0.220075315563754,"score_spread":0.2080672011608028,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409231750","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":[0.60751384,0.0009570543,0.36804658,0.0008865803,0.0001383136,0.00024178931,0.00089474727,0.01796612,0.0033549136],"genre_scores_gemma":[0.9570993,0.00015820943,0.04127083,0.00007609788,0.000022403672,0.00008180212,0.00064787845,0.0003751644,0.0002683703],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9884597,0.0044911015,0.0010711683,0.0009960883,0.0043222588,0.0006595989],"domain_scores_gemma":[0.9510816,0.026080018,0.00468151,0.008435803,0.008516825,0.0012042486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011372523,0.000976522,0.0007921665,0.0014601266,0.000619518,0.0024358402,0.0018856218,0.0009825006,0.0009136312],"category_scores_gemma":[0.072497666,0.00047308323,0.00036960797,0.001279207,0.0012511523,0.0039187637,0.0013914733,0.0013493232,0.00029956864],"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":[0.0029628584,0.00078598654,0.12621269,0.0009874929,0.00042079817,0.0007647815,0.0018752818,0.5200634,0.04826692,0.020650737,0.0104570575,0.26655194],"study_design_scores_gemma":[0.00004796888,0.00034586937,0.008716609,0.000059874365,0.00004862916,0.00016057046,0.00026182292,0.9369318,0.041283928,0.00948051,0.002615806,0.000046549958],"about_ca_topic_score_codex":0.0029913667,"about_ca_topic_score_gemma":0.0018798761,"teacher_disagreement_score":0.011372523,"about_ca_system_score_codex":0.0011597312,"about_ca_system_score_gemma":0.0015770152,"threshold_uncertainty_score":0.060144365},"labels":[],"label_agreement":null},{"id":"W4410543982","doi":"10.14778/3717755.3717766","title":"WeShap: Weak Supervision Source Evaluation with Shapley Values","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Shapley value; Mathematical economics; Mathematics; Game theory","score_opus":0.024034614301506736,"score_gpt":0.2549961208786427,"score_spread":0.23096150657713596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410543982","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":[0.047082994,0.00075012574,0.9351886,0.0011284319,0.00019938493,0.00035102994,0.0009115106,0.0069814003,0.0074065896],"genre_scores_gemma":[0.48160687,0.00030227096,0.5088531,0.00066363555,0.00013837783,0.0005412253,0.0024981296,0.0014511903,0.003945245],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9942194,0.0025120126,0.00033497094,0.0010508416,0.0015224309,0.000360385],"domain_scores_gemma":[0.98173565,0.011745287,0.0007999783,0.0030530875,0.0020879046,0.0005780731],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011865559,0.002104456,0.0016053612,0.002353991,0.0016765097,0.003421476,0.0036145824,0.0026567224,0.007847093],"category_scores_gemma":[0.03918682,0.00091187406,0.001352123,0.0018880274,0.002541406,0.006261868,0.0053492985,0.0043949,0.0017183601],"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":[0.00096948654,0.0004139285,0.007216698,0.00068436365,0.00026391353,0.00023934181,0.00041873867,0.48785755,0.0055123325,0.0876551,0.023661265,0.3851073],"study_design_scores_gemma":[0.000055692955,0.00010963033,0.00031248873,0.000050085026,0.000021273734,0.000042201853,0.000048944552,0.9111509,0.0039689024,0.08223721,0.0019825285,0.000020133331],"about_ca_topic_score_codex":0.0033069174,"about_ca_topic_score_gemma":0.00541851,"teacher_disagreement_score":0.011865559,"about_ca_system_score_codex":0.0030443277,"about_ca_system_score_gemma":0.0052707456,"threshold_uncertainty_score":0.06275183},"labels":[],"label_agreement":null},{"id":"W4410544001","doi":"10.14778/3717755.3717771","title":"In-Depth Analysis of Densest Subgraph Discovery in a Unified Framework","year":2024,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Computational biology; Biology","score_opus":0.00938181969614827,"score_gpt":0.2451218865393913,"score_spread":0.23574006684324303,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410544001","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":[0.049430836,0.005306741,0.93952864,0.0010763137,0.00007672143,0.00012250103,0.00073599775,0.0011479032,0.0025743558],"genre_scores_gemma":[0.374308,0.0034815473,0.61642236,0.00041047324,0.00026593226,0.00012619367,0.0029621976,0.00041790341,0.0016053498],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9928308,0.0023191164,0.00035602102,0.0016686297,0.00224763,0.0005777375],"domain_scores_gemma":[0.9802712,0.01187954,0.0013219985,0.0043724817,0.0015594236,0.0005952858],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007226887,0.0013427953,0.0030603497,0.0060115173,0.0014232851,0.0044362033,0.0035344535,0.0020352043,0.0024056723],"category_scores_gemma":[0.030223886,0.000913986,0.0019680387,0.0063355854,0.002042448,0.011052832,0.004302676,0.0024707867,0.00057679944],"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":[0.00048325886,0.00063098903,0.017227981,0.001474205,0.0006034713,0.00036465112,0.00096811884,0.46138957,0.012488763,0.23020929,0.012820217,0.26133943],"study_design_scores_gemma":[0.000021961863,0.00006602425,0.0012190549,0.000046710633,0.000092736744,0.00025407432,0.00015500978,0.900622,0.0028892206,0.08992828,0.0046855505,0.00001933171],"about_ca_topic_score_codex":0.005197784,"about_ca_topic_score_gemma":0.009713241,"teacher_disagreement_score":0.007226887,"about_ca_system_score_codex":0.0030155012,"about_ca_system_score_gemma":0.0037645285,"threshold_uncertainty_score":0.03821987},"labels":[],"label_agreement":null},{"id":"W4413755882","doi":"10.14778/3718057.3718059","title":"Esc: An Early-Stopping Checker for Budget-Aware Index Tuning","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Index (typography); Computer science; Model checking; Programming language","score_opus":0.012348128414331278,"score_gpt":0.2604888289779635,"score_spread":0.24814070056363222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413755882","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":[0.036647473,0.00095936557,0.814797,0.0006746485,0.00047024767,0.0006062952,0.0008159691,0.13952267,0.005506308],"genre_scores_gemma":[0.4110975,0.00035084446,0.5666103,0.0014201789,0.00024197936,0.00062825694,0.001977284,0.012243309,0.0054303664],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9873346,0.0032466992,0.0014453285,0.0020580387,0.0048595313,0.0010557793],"domain_scores_gemma":[0.96046424,0.02015448,0.0029209065,0.010937893,0.0042704944,0.0012520466],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010851969,0.0025709574,0.0018344189,0.0021824194,0.001272061,0.0036020246,0.004858888,0.0021700314,0.005727821],"category_scores_gemma":[0.04916378,0.001762623,0.0016063956,0.0011142121,0.0020391964,0.0057983524,0.0048013697,0.0037943888,0.0027932434],"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":[0.004822856,0.0013500348,0.030307353,0.0014256157,0.0005556486,0.0012316897,0.0012997763,0.13935988,0.074634865,0.042417638,0.076159716,0.626435],"study_design_scores_gemma":[0.000377676,0.00035977576,0.0018411516,0.00011742317,0.00014444605,0.00028554862,0.0001237082,0.9223467,0.03544196,0.019661762,0.019150874,0.00014886123],"about_ca_topic_score_codex":0.0029524066,"about_ca_topic_score_gemma":0.004335444,"teacher_disagreement_score":0.010851969,"about_ca_system_score_codex":0.001251427,"about_ca_system_score_gemma":0.003908501,"threshold_uncertainty_score":0.057391346},"labels":[],"label_agreement":null},{"id":"W4413755894","doi":"10.14778/3718057.3718071","title":"Cabinet: Dynamically Weighted Consensus Made Fast","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Concordia University","funders":"","keywords":"Cabinet (room); Consensus conference; Computer science; History; Library science; Archaeology","score_opus":0.004561944860713993,"score_gpt":0.21195633322434074,"score_spread":0.20739438836362675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413755894","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":[0.03144537,0.0004970291,0.94618183,0.00041417396,0.00029234585,0.00028387143,0.00027555198,0.014596258,0.006013528],"genre_scores_gemma":[0.52953845,0.00028752073,0.45824626,0.00033163163,0.000095632764,0.00054377125,0.0011128881,0.00077947037,0.009064437],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9983,0.00040141074,0.00008588541,0.00035962142,0.0005954641,0.0002575996],"domain_scores_gemma":[0.99716115,0.0007723808,0.00024081052,0.0007943797,0.00079772377,0.00023356278],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021153057,0.00084296643,0.0010253178,0.0010403348,0.0014250415,0.0012458283,0.0030602089,0.0011829191,0.004163448],"category_scores_gemma":[0.0050348523,0.00052211905,0.0004887401,0.0010834298,0.0009226926,0.002272297,0.0025451027,0.001673021,0.0013032152],"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":[0.0016010138,0.00027097296,0.0016118274,0.0004788527,0.00018866659,0.00040005785,0.0005654362,0.4751628,0.043594755,0.048776038,0.04006705,0.3872825],"study_design_scores_gemma":[0.00020306311,0.00023494553,0.00022438657,0.000020067104,0.000025054736,0.00013556016,0.00008576727,0.9560271,0.012833698,0.016583396,0.013583568,0.00004340563],"about_ca_topic_score_codex":0.0042654285,"about_ca_topic_score_gemma":0.0046256306,"teacher_disagreement_score":0.0042654285,"about_ca_system_score_codex":0.0008459431,"about_ca_system_score_gemma":0.0021695732,"threshold_uncertainty_score":0.013928115},"labels":[],"label_agreement":null},{"id":"W4413755925","doi":"10.14778/3718057.3718077","title":"FLEET: High-Performance Durable Replicated State Machines Using Scattered and Coordinated Log Entries","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"State (computer science); Computer science; Environmental science; Algorithm","score_opus":0.007847050474531114,"score_gpt":0.2207120930686482,"score_spread":0.2128650425941171,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413755925","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":[0.1569668,0.0005704962,0.76969856,0.0003889366,0.00014035127,0.0006611323,0.0010783212,0.06338233,0.007113121],"genre_scores_gemma":[0.8154177,0.0001995316,0.17649391,0.00013573072,0.000038285594,0.00035505753,0.002199793,0.00067853194,0.004481468],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990939,0.00018763654,0.0000801448,0.00016107086,0.00037148813,0.0001057277],"domain_scores_gemma":[0.9980915,0.0004549113,0.00019832651,0.0009243314,0.00018829951,0.00014250253],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012942886,0.0005920398,0.0005068426,0.0006760973,0.0004699178,0.0012612742,0.0025268316,0.00053762575,0.0028653187],"category_scores_gemma":[0.0036212313,0.0004452216,0.0003364872,0.00076924526,0.000642402,0.0027880126,0.0022468423,0.00083810516,0.0007142499],"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":[0.0032258942,0.0012559111,0.00686027,0.0006412799,0.0003267023,0.001067479,0.00052780984,0.32390246,0.11652715,0.032897227,0.04871726,0.46405062],"study_design_scores_gemma":[0.00041039113,0.0006834933,0.0014688468,0.000023704719,0.00004037634,0.00034015547,0.00012184857,0.93634176,0.033017162,0.0096444655,0.017823536,0.000084187326],"about_ca_topic_score_codex":0.0024172694,"about_ca_topic_score_gemma":0.003042713,"teacher_disagreement_score":0.0028653187,"about_ca_system_score_codex":0.00054931943,"about_ca_system_score_gemma":0.0009815056,"threshold_uncertainty_score":0.00958544},"labels":[],"label_agreement":null},{"id":"W4413812270","doi":"10.14778/3725688.3725720","title":"A Flexible Framework for Query-Oriented Interactive Community Search","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Peer-to-Peer Network Technologies","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"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; Query optimization; Information retrieval; Web search query; Query expansion; Query language; Theoretical computer science; Search engine","score_opus":0.024246797064614276,"score_gpt":0.3154671397279443,"score_spread":0.29122034266333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413812270","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":[0.0038723215,0.00022511774,0.99244857,0.00023480445,0.000022517064,0.00016159864,0.00024708526,0.0017976284,0.0009903986],"genre_scores_gemma":[0.17179774,0.00025070555,0.8237951,0.00024653572,0.00007432961,0.00046768115,0.0012732195,0.00033246476,0.0017623071],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9964588,0.0012896468,0.0001407512,0.0009480595,0.0008451292,0.00031747704],"domain_scores_gemma":[0.99580574,0.0016165632,0.00027194657,0.0011391076,0.0007614618,0.00040508716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036210523,0.0012665789,0.0013257331,0.0032465744,0.0018101989,0.0020822017,0.0047187787,0.0021775558,0.0037027716],"category_scores_gemma":[0.012046853,0.0006881321,0.0017982988,0.0038736258,0.0017517462,0.0053567044,0.004995357,0.002723594,0.0013739374],"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":[0.00047914224,0.00044025632,0.0038131503,0.0005287713,0.00019328806,0.0005490575,0.0014432681,0.32391146,0.015731249,0.29068127,0.032084372,0.33014473],"study_design_scores_gemma":[0.00005633057,0.000054906886,0.00015997609,0.000018599603,0.0000201406,0.00013615946,0.0000924614,0.914081,0.0011919742,0.07764256,0.006516917,0.000028925566],"about_ca_topic_score_codex":0.013190043,"about_ca_topic_score_gemma":0.017884934,"teacher_disagreement_score":0.013190043,"about_ca_system_score_codex":0.001840446,"about_ca_system_score_gemma":0.0029281196,"threshold_uncertainty_score":0.02622652},"labels":[],"label_agreement":null},{"id":"W4413812283","doi":"10.14778/3725688.3725721","title":"Tabular: Efficiently Building Efficient Indexes","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"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","score_opus":0.003819544727282654,"score_gpt":0.2172714366275403,"score_spread":0.21345189190025765,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413812283","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":[0.027522806,0.001261699,0.8673923,0.00020364024,0.000116197036,0.00024134219,0.0025507777,0.09371444,0.006996787],"genre_scores_gemma":[0.1636599,0.0012169206,0.80970025,0.00027798308,0.00008844715,0.00036252438,0.00807539,0.008585123,0.008033492],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982262,0.00030527744,0.00025513212,0.0002321857,0.0007827686,0.00019843181],"domain_scores_gemma":[0.99451005,0.0015073742,0.0004283475,0.0020285014,0.0012315848,0.00029421397],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022362373,0.0010263489,0.0010001984,0.001954665,0.0010217689,0.0045398516,0.0028578872,0.00082781725,0.0072703036],"category_scores_gemma":[0.009422118,0.0011444036,0.0010368036,0.0034642909,0.00087198574,0.007167565,0.0036006065,0.00097611174,0.0049474505],"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":[0.0014929365,0.00034269737,0.007792738,0.001908659,0.0002062755,0.00055218034,0.0009875657,0.053662516,0.043654848,0.055855814,0.07505081,0.75849295],"study_design_scores_gemma":[0.00036607136,0.0006867914,0.0023834908,0.00034925708,0.00022785555,0.0010266897,0.00053903606,0.53016907,0.16917793,0.07303827,0.22179194,0.00024360018],"about_ca_topic_score_codex":0.002326653,"about_ca_topic_score_gemma":0.0024478037,"teacher_disagreement_score":0.0072703036,"about_ca_system_score_codex":0.0008671998,"about_ca_system_score_gemma":0.0022142185,"threshold_uncertainty_score":0.024321616},"labels":[],"label_agreement":null},{"id":"W4413825196","doi":"10.14778/3734839.3734849","title":"Accio: Bolt-on Query Federation","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Russian federation; Computer science; Information retrieval; Geography; Regional science","score_opus":0.005557855411297835,"score_gpt":0.22850257909836866,"score_spread":0.22294472368707083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413825196","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":[0.05730551,0.0013946624,0.6682981,0.0013443186,0.0008509272,0.0007828317,0.0017771561,0.25133365,0.016912818],"genre_scores_gemma":[0.44472638,0.0008120393,0.503323,0.002480467,0.0005184979,0.0006457047,0.014135332,0.020002188,0.013356431],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9879736,0.001866802,0.0012577946,0.001974084,0.005442995,0.0014846796],"domain_scores_gemma":[0.98047054,0.0023233488,0.0007606718,0.012701196,0.002629614,0.0011145874],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008981935,0.0021707045,0.0014732827,0.0020218925,0.002007473,0.0057345545,0.0071431478,0.0018272758,0.004903167],"category_scores_gemma":[0.016406188,0.0017041408,0.0024825353,0.0023686502,0.0019281555,0.008869699,0.011173242,0.0038999913,0.0026012708],"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":[0.0066275317,0.0019931062,0.038925882,0.0019040265,0.0016029689,0.0017666542,0.0053013754,0.042513434,0.114623606,0.07070595,0.24891013,0.4651253],"study_design_scores_gemma":[0.00076314696,0.0011264768,0.009831591,0.00020293635,0.0005944353,0.0019586745,0.0016232625,0.54405665,0.10960177,0.036879253,0.29274583,0.0006160698],"about_ca_topic_score_codex":0.012532781,"about_ca_topic_score_gemma":0.008611553,"teacher_disagreement_score":0.012532781,"about_ca_system_score_codex":0.0019733207,"about_ca_system_score_gemma":0.004500308,"threshold_uncertainty_score":0.047501564},"labels":[],"label_agreement":null},{"id":"W4413825311","doi":"10.14778/3725688.3725693","title":"Efficient Historical Butterfly Counting in Large Temporal Bipartite Networks via Graph Structure-Aware Index","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Bipartite graph; Computer science; Butterfly; Graph; Index (typography); Theoretical computer science; Artificial intelligence; World Wide Web; Biology; Ecology","score_opus":0.0045790634591464825,"score_gpt":0.22311424491938264,"score_spread":0.21853518146023615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413825311","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":[0.10213051,0.0012073336,0.88197064,0.0006020857,0.00012559426,0.00022335899,0.0017606779,0.007489246,0.0044905446],"genre_scores_gemma":[0.44694263,0.0007588165,0.54179853,0.0002965247,0.00013702057,0.00029782322,0.005520823,0.00069086277,0.0035569135],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987154,0.00017336516,0.000103034705,0.00028875805,0.00056211255,0.00015746827],"domain_scores_gemma":[0.9955124,0.0017521617,0.000456025,0.0014601459,0.00057008624,0.00024909567],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008778203,0.0010479101,0.0016319201,0.0022451743,0.0009901272,0.002358202,0.0030449748,0.0011289519,0.0026962417],"category_scores_gemma":[0.008786016,0.0005545624,0.00093327113,0.004788386,0.000808236,0.005264517,0.0021750003,0.0012500874,0.00144205],"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":[0.00069763587,0.00046428715,0.010334327,0.00058569585,0.00015124228,0.00042779275,0.00096657104,0.22107129,0.032256898,0.057691008,0.033566475,0.6417869],"study_design_scores_gemma":[0.0000368939,0.00006379249,0.0006505439,0.00001612258,0.000023881657,0.00021998642,0.00015005478,0.9632051,0.004777897,0.027784696,0.0030481203,0.0000228781],"about_ca_topic_score_codex":0.006575123,"about_ca_topic_score_gemma":0.011001424,"teacher_disagreement_score":0.006575123,"about_ca_system_score_codex":0.0015139934,"about_ca_system_score_gemma":0.0021350475,"threshold_uncertainty_score":0.013073742},"labels":[],"label_agreement":null},{"id":"W4413827533","doi":"10.14778/3725688.3725702","title":"G-View: View Management for Graph Databases","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Graph database; Database; Graph; Information retrieval; Theoretical computer science","score_opus":0.019736695528153752,"score_gpt":0.26537238348359177,"score_spread":0.24563568795543803,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413827533","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":[0.006827001,0.00086167443,0.8948014,0.00063178915,0.00015241881,0.0003469056,0.005846601,0.08622141,0.004310789],"genre_scores_gemma":[0.14266354,0.0016977894,0.8086345,0.001150756,0.00017461488,0.00091704284,0.02659115,0.01345263,0.00471796],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99665713,0.00079924933,0.00044444198,0.0006033123,0.0012223576,0.000273498],"domain_scores_gemma":[0.9933344,0.0017040499,0.00033135686,0.003426098,0.0008339308,0.00037023317],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036425018,0.0014405916,0.0012890889,0.0027176284,0.0009117641,0.006097507,0.0051711495,0.0016217143,0.0065693264],"category_scores_gemma":[0.013348754,0.0011842512,0.0019633202,0.003485774,0.0011323323,0.008940582,0.0056981593,0.0026392192,0.0026705144],"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":[0.0018147187,0.00031592953,0.008628903,0.0014919868,0.0005388198,0.0005597016,0.0014012979,0.03415604,0.025077496,0.24489208,0.20332867,0.47779432],"study_design_scores_gemma":[0.0004705528,0.0004364266,0.0020532808,0.00035403224,0.00028867862,0.0011567933,0.00062412414,0.3929759,0.055340067,0.23099525,0.3149927,0.00031215686],"about_ca_topic_score_codex":0.005855026,"about_ca_topic_score_gemma":0.006012671,"teacher_disagreement_score":0.0065693264,"about_ca_system_score_codex":0.0013601844,"about_ca_system_score_gemma":0.0021688628,"threshold_uncertainty_score":0.02197659},"labels":[],"label_agreement":null},{"id":"W4413943427","doi":"10.14778/3742728.3742754","title":"CatDB: Data-Catalog-Guided, LLM-Based Generation of Data-Centric ML Pipelines","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Pipeline transport; Computer science; Environmental science","score_opus":0.10422959380219059,"score_gpt":0.32988666442145576,"score_spread":0.22565707061926515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413943427","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":[0.005317667,0.00046066198,0.37114197,0.00044055254,0.00021616917,0.0003155965,0.01120617,0.6073793,0.0035218885],"genre_scores_gemma":[0.06842161,0.00040631066,0.7577076,0.0012848913,0.00009736888,0.0012409762,0.09384237,0.07128597,0.005712932],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9946148,0.00077943475,0.00064275996,0.0017959718,0.0018613443,0.00030568347],"domain_scores_gemma":[0.9855899,0.0038341316,0.00070090254,0.0063567376,0.0029660666,0.00055224914],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051434916,0.003028408,0.001745546,0.0037514658,0.0015727077,0.005097552,0.0077342615,0.0020805052,0.013268007],"category_scores_gemma":[0.028405065,0.0023795257,0.0029072934,0.0033724383,0.0023448754,0.006832651,0.009485458,0.003963036,0.014043957],"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":[0.0015940849,0.0004784999,0.008973233,0.0020304318,0.00045622344,0.0006196785,0.0016800538,0.035564266,0.02783486,0.023019707,0.46404263,0.43370646],"study_design_scores_gemma":[0.00063555734,0.00034090775,0.0028836261,0.00026965272,0.0001494141,0.00051318586,0.00046331892,0.56284535,0.12029338,0.04704397,0.2642062,0.00035546208],"about_ca_topic_score_codex":0.008105272,"about_ca_topic_score_gemma":0.007960838,"teacher_disagreement_score":0.013268007,"about_ca_system_score_codex":0.0025282563,"about_ca_system_score_gemma":0.0054875384,"threshold_uncertainty_score":0.04438585},"labels":[],"label_agreement":null},{"id":"W4413943450","doi":"10.14778/3742728.3742738","title":"Asymmetric Linearizable Local Reads","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"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; Programming language; Mathematics; Process management; Business","score_opus":0.005162037269071246,"score_gpt":0.21707470301876206,"score_spread":0.2119126657496908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413943450","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":[0.0717325,0.0006012358,0.90712464,0.00039807986,0.00013981303,0.00021242145,0.00028812166,0.011525746,0.007977566],"genre_scores_gemma":[0.741711,0.00025027755,0.24163628,0.00035595943,0.00015256905,0.00028070257,0.00068791053,0.0005930544,0.014332164],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99753064,0.00041061794,0.00017013286,0.0006718776,0.0007534492,0.00046330737],"domain_scores_gemma":[0.9917385,0.0023068888,0.001005221,0.0033155852,0.0013036248,0.00033011945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012763701,0.0010450961,0.0009993846,0.0006548383,0.0009905966,0.0018656417,0.0036738317,0.0010091763,0.0072881323],"category_scores_gemma":[0.004889504,0.0005224745,0.00047544896,0.0013463591,0.0010353087,0.003512857,0.0028633862,0.0015645064,0.0019946233],"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":[0.0030939293,0.00096720323,0.004561057,0.00059218047,0.00010641798,0.00040246043,0.0008681981,0.26167348,0.08751024,0.03715958,0.02022116,0.582844],"study_design_scores_gemma":[0.00019811219,0.00061555556,0.00065770216,0.000031384097,0.0000649455,0.00031717587,0.00040209282,0.889522,0.07519285,0.02417095,0.008763982,0.0000633082],"about_ca_topic_score_codex":0.0020647743,"about_ca_topic_score_gemma":0.0040981052,"teacher_disagreement_score":0.0072881323,"about_ca_system_score_codex":0.0014195585,"about_ca_system_score_gemma":0.0017891498,"threshold_uncertainty_score":0.02438122},"labels":[],"label_agreement":null},{"id":"W4413948809","doi":"10.14778/3746405.3746414","title":"KEIGO: Co-Designing Log-Structured Merge Key-Value Stores with a Non-Volatile, Concurrency-Aware Storage Hierarchy","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Merge (version control); Computer science; Concurrency; Hierarchy; Memory hierarchy; Database; Key (lock); Parallel computing; Distributed computing; Operating system","score_opus":0.007952278613537941,"score_gpt":0.24574192578378504,"score_spread":0.2377896471702471,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413948809","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":[0.09822407,0.0013304051,0.7926864,0.0006039933,0.00025238143,0.0005712601,0.0007770271,0.097226724,0.008327741],"genre_scores_gemma":[0.52480733,0.00051409093,0.4584493,0.00046357294,0.00008221318,0.0003844267,0.0019607996,0.0041629514,0.009175358],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986761,0.00015201887,0.00014376736,0.00028370594,0.0005245088,0.00021982116],"domain_scores_gemma":[0.9973695,0.00042904972,0.00019111169,0.0011934483,0.0004987187,0.00031810222],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013971912,0.0009665218,0.0006807217,0.00082047255,0.00067701965,0.002312008,0.00450884,0.00062338915,0.0029740287],"category_scores_gemma":[0.0052439906,0.001090094,0.0006833789,0.0009031847,0.0011227583,0.004605883,0.0051531876,0.0015536471,0.0018168119],"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":[0.0033825047,0.0011578568,0.021875778,0.0018821788,0.00046628286,0.0010361861,0.0024481234,0.076489754,0.25180256,0.0357189,0.06347966,0.5402602],"study_design_scores_gemma":[0.00051113975,0.0010272573,0.0034213266,0.000115778574,0.00025352286,0.0011030383,0.0007468803,0.67467916,0.2102069,0.023446199,0.08415773,0.00033096183],"about_ca_topic_score_codex":0.0025169023,"about_ca_topic_score_gemma":0.0039790673,"teacher_disagreement_score":0.00450884,"about_ca_system_score_codex":0.00079270493,"about_ca_system_score_gemma":0.0022135675,"threshold_uncertainty_score":0.009949148},"labels":[],"label_agreement":null},{"id":"W4413953194","doi":"10.14778/3742728.3742730","title":"Oze: Decentralized Graph-Based Concurrency Control for Long-Running Update Transactions","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nautilus Environmental","funders":"","keywords":"Concurrency control; Computer science; Concurrency; Graph; Control (management); Distributed computing; Programming language; Theoretical computer science; Database transaction; Artificial intelligence","score_opus":0.007703296319875869,"score_gpt":0.24421487510674494,"score_spread":0.23651157878686907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413953194","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":[0.023282355,0.00026469713,0.96918696,0.00021619283,0.00009239565,0.00023644646,0.00006910155,0.005016216,0.0016355503],"genre_scores_gemma":[0.6898145,0.00023275574,0.30525944,0.00028730757,0.0000920814,0.0003525758,0.00032598292,0.0005130182,0.0031223008],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9964218,0.0009010216,0.00027128024,0.00061931356,0.00144604,0.0003404666],"domain_scores_gemma":[0.9932539,0.0024504082,0.00072032347,0.0022956533,0.0008803681,0.00039938994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035672956,0.000895809,0.00088390533,0.0011828785,0.0010237572,0.0016858918,0.0030065433,0.00071365014,0.0020582685],"category_scores_gemma":[0.007062805,0.0005921907,0.00056234235,0.0008377359,0.0019083776,0.0036925776,0.0030465354,0.0022699353,0.0003444742],"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":[0.0017958034,0.00075261306,0.0052462174,0.00047177338,0.00027130113,0.00040840032,0.00086520123,0.35163122,0.09793555,0.12615454,0.008171965,0.40629545],"study_design_scores_gemma":[0.00022184363,0.0002461787,0.0005591421,0.000018779734,0.000050186434,0.00013758172,0.00010166558,0.92262644,0.031681176,0.03571251,0.008568472,0.00007599224],"about_ca_topic_score_codex":0.0029867964,"about_ca_topic_score_gemma":0.0044730487,"teacher_disagreement_score":0.0035672956,"about_ca_system_score_codex":0.0010177196,"about_ca_system_score_gemma":0.002085221,"threshold_uncertainty_score":0.018865883},"labels":[],"label_agreement":null},{"id":"W4413953529","doi":"10.14778/3742728.3742753","title":"Robust Plan Evaluation Based on Approximate Probabilistic Machine Learning","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Rough Sets and Fuzzy Logic","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada); University of Ottawa","funders":"","keywords":"Plan (archaeology); Probabilistic logic; Computer science; Machine learning; Artificial intelligence","score_opus":0.03655889814260427,"score_gpt":0.24015690820941218,"score_spread":0.20359801006680792,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413953529","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":[0.026791979,0.00063152995,0.96965194,0.0003764669,0.000019198984,0.00011667534,0.00015965097,0.0012760238,0.0009765781],"genre_scores_gemma":[0.672756,0.00033555785,0.32449883,0.00022803163,0.00007129122,0.00023460669,0.0005174961,0.00025118008,0.0011069794],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99293643,0.0024769183,0.00043581708,0.0011081452,0.0026007353,0.00044187624],"domain_scores_gemma":[0.98658586,0.009171231,0.0012628777,0.001532779,0.0011598958,0.00028733345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006307388,0.0011835841,0.0021377911,0.0016038318,0.0005285119,0.002485813,0.0023135187,0.0012440189,0.0018506339],"category_scores_gemma":[0.022985283,0.00071039615,0.001155116,0.0016662229,0.0016886451,0.0040801507,0.0024395995,0.002013884,0.00036144376],"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":[0.00030721398,0.000094383824,0.0018883321,0.00012893582,0.00009609753,0.00005837588,0.00011925045,0.865527,0.0023304748,0.022305403,0.0015773926,0.10556714],"study_design_scores_gemma":[0.000009477503,0.00003533204,0.00011619076,0.000006231203,0.000008492429,0.000011370748,0.00001048179,0.9904697,0.00074109377,0.00838417,0.00020091879,0.000006488992],"about_ca_topic_score_codex":0.006202673,"about_ca_topic_score_gemma":0.0053663137,"teacher_disagreement_score":0.006307388,"about_ca_system_score_codex":0.0030012853,"about_ca_system_score_gemma":0.0027646916,"threshold_uncertainty_score":0.033357084},"labels":[],"label_agreement":null},{"id":"W4413968449","doi":"10.14778/3746405.3746413","title":"Locality-Aware Cache Replacement Policy for Graph Traversals","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Locality; Cache; Computer science; Locality of reference; Graph; Parallel computing; Theoretical computer science","score_opus":0.008567600864450518,"score_gpt":0.2535545745113512,"score_spread":0.2449869736469007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413968449","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":[0.605959,0.0034737147,0.36106515,0.0008971165,0.00020431395,0.00037485504,0.00075317716,0.0205192,0.006753458],"genre_scores_gemma":[0.8870694,0.00029870967,0.10940731,0.00019304383,0.000030274006,0.00010365016,0.00081473385,0.00034171218,0.001741213],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989453,0.00024957737,0.00008244689,0.00017517524,0.00035694367,0.00019061264],"domain_scores_gemma":[0.995799,0.001083632,0.00034645037,0.0014218491,0.0011179333,0.00023126623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010312446,0.0005869676,0.00069608865,0.0012909048,0.0011935299,0.0010109195,0.0019419814,0.0006433118,0.0010523599],"category_scores_gemma":[0.0057125767,0.00030407534,0.00038789745,0.0019626056,0.00076169334,0.0018135158,0.00076524017,0.00064301543,0.00037558877],"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":[0.0019163396,0.0010235235,0.030714758,0.0005570562,0.00021874915,0.0005171202,0.00088345906,0.3152662,0.112830244,0.020526072,0.03916757,0.47637898],"study_design_scores_gemma":[0.00014272702,0.00044041532,0.0028711662,0.000025487765,0.00008929006,0.0003987653,0.00029179355,0.94062823,0.036754433,0.008833843,0.009476427,0.000047405312],"about_ca_topic_score_codex":0.011475162,"about_ca_topic_score_gemma":0.025860716,"teacher_disagreement_score":0.011475162,"about_ca_system_score_codex":0.0018718855,"about_ca_system_score_gemma":0.0032958263,"threshold_uncertainty_score":0.022816777},"labels":[],"label_agreement":null},{"id":"W4413978687","doi":"10.14778/3749646.3749676","title":"Environmental Footprints of Query Processing: A Vision for Sustainable Database Architectures","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"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; Database; Information retrieval","score_opus":0.005778839045125856,"score_gpt":0.230482807218248,"score_spread":0.22470396817312213,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413978687","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":[0.059641533,0.039412208,0.6677154,0.14217587,0.0026669488,0.00016236602,0.0007308224,0.0019239517,0.08557093],"genre_scores_gemma":[0.7109702,0.025651982,0.23826343,0.0064566983,0.0033382487,0.00034521392,0.00055730384,0.0011534687,0.013263569],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99455345,0.0015275754,0.00027201415,0.00050955836,0.0027775518,0.0003599391],"domain_scores_gemma":[0.9859899,0.0058113234,0.00084941834,0.002729907,0.0036147048,0.0010046578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00809813,0.0011776322,0.00092651014,0.0025003678,0.0016201825,0.011202629,0.0027310553,0.003105363,0.003063934],"category_scores_gemma":[0.019089,0.00068158965,0.0005336488,0.002744809,0.007522231,0.02436391,0.0052509373,0.0051030563,0.00079298485],"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":[0.00010978037,0.00013129607,0.0025476986,0.00027041417,0.00006803131,0.000074204014,0.0004468483,0.021986242,0.00333338,0.8170652,0.013214104,0.14075275],"study_design_scores_gemma":[0.000016380749,0.00009228045,0.0009779648,0.0001896913,0.00003488283,0.000116741,0.0006498529,0.039316837,0.002911453,0.8644362,0.0911999,0.000057774836],"about_ca_topic_score_codex":0.002338738,"about_ca_topic_score_gemma":0.0017058622,"teacher_disagreement_score":0.011202629,"about_ca_system_score_codex":0.0028083206,"about_ca_system_score_gemma":0.0034762993,"threshold_uncertainty_score":0.042827487},"labels":[],"label_agreement":null},{"id":"W4413978785","doi":"10.14778/3749646.3749727","title":"ParSEval: Plan-Aware Test Database Generation for SQL Equivalence Evaluation","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"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; Database; SQL; Parseval's theorem; Plan (archaeology); Equivalence (formal languages); Programming language; Mathematics; Discrete mathematics","score_opus":0.30194040095456554,"score_gpt":0.43839978790090695,"score_spread":0.1364593869463414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413978785","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":[0.012902081,0.0004184019,0.8915955,0.00025115345,0.000080364465,0.00057592214,0.0021767798,0.09002889,0.0019710443],"genre_scores_gemma":[0.2306956,0.00025245702,0.7520573,0.00037297604,0.000043724765,0.0007938365,0.009752013,0.004916668,0.0011155052],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949184,0.0016867226,0.00045218074,0.0008607434,0.001731855,0.0003499735],"domain_scores_gemma":[0.9900918,0.00586262,0.0005360389,0.002154443,0.0011606178,0.0001944573],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047211535,0.0018201952,0.001014382,0.002305634,0.0005254734,0.0020586986,0.003503745,0.0010864307,0.0062244474],"category_scores_gemma":[0.018443909,0.0008737901,0.0022017425,0.0012366145,0.001332199,0.0027481695,0.0027898692,0.0018274757,0.0015137728],"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":[0.0015105447,0.0007873001,0.013842195,0.0018108195,0.000502769,0.0008344909,0.0007127408,0.20981792,0.026804874,0.048039403,0.07347863,0.62185824],"study_design_scores_gemma":[0.00021203946,0.00024122892,0.0009923831,0.00008189465,0.00006813174,0.000272456,0.00011792941,0.9404092,0.021356378,0.023268932,0.01292199,0.000057435307],"about_ca_topic_score_codex":0.006306109,"about_ca_topic_score_gemma":0.007195491,"teacher_disagreement_score":0.006306109,"about_ca_system_score_codex":0.0013276082,"about_ca_system_score_gemma":0.0029766127,"threshold_uncertainty_score":0.024968147},"labels":[],"label_agreement":null},{"id":"W4413980865","doi":"10.14778/3749646.3749703","title":"Sphinx: A Succinct Perfect Hash Index for x86","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Hash function; Sphinx; Index (typography); Computer science; x86; Computer security; History; World Wide Web; Programming language; Archaeology","score_opus":0.00943147209938856,"score_gpt":0.24889599828362524,"score_spread":0.2394645261842367,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413980865","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":[0.12488463,0.0038632345,0.7545288,0.00088197395,0.00076143816,0.00085987267,0.0091039315,0.06789794,0.037218112],"genre_scores_gemma":[0.5798948,0.0016836962,0.35591602,0.00064382405,0.00027689326,0.00063672994,0.021552067,0.003925216,0.035470784],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991289,0.00008961856,0.00009850749,0.000086755485,0.0004778346,0.00011841671],"domain_scores_gemma":[0.9989497,0.00013474139,0.000113689784,0.00048657687,0.00023179933,0.00008338054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006399964,0.00058502477,0.0005606063,0.00080371555,0.0006832764,0.0014796687,0.0017673495,0.00043693915,0.009974441],"category_scores_gemma":[0.002378857,0.00048384906,0.00034840408,0.0014170908,0.0005952079,0.0033161715,0.002384161,0.000842432,0.004584399],"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":[0.0036721616,0.00029154998,0.006931183,0.0009200914,0.00012536527,0.0005342339,0.0007333775,0.020519108,0.07595912,0.09965957,0.15805934,0.63259494],"study_design_scores_gemma":[0.00076278544,0.0013735482,0.0048835548,0.00020722393,0.0001176021,0.0010166086,0.00041224508,0.18963823,0.20439127,0.07924139,0.5176688,0.00028685044],"about_ca_topic_score_codex":0.0017891763,"about_ca_topic_score_gemma":0.001937695,"teacher_disagreement_score":0.009974441,"about_ca_system_score_codex":0.0007302218,"about_ca_system_score_gemma":0.0016439761,"threshold_uncertainty_score":0.033367872},"labels":[],"label_agreement":null},{"id":"W4413985233","doi":"10.14778/3748191.3748215","title":"PS-MI: Accurate, Efficient, and Private Data Valuation in Vertical Federated Learning","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Valuation (finance); Federated learning; Computer science; Business; Artificial intelligence; Finance","score_opus":0.04816835999275128,"score_gpt":0.30465121280028323,"score_spread":0.25648285280753197,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413985233","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":[0.030055044,0.00018501548,0.96688163,0.00040988796,0.000028662655,0.00012213553,0.00014673173,0.0012230352,0.0009478305],"genre_scores_gemma":[0.7371707,0.00011972034,0.25945586,0.00031693053,0.00005042513,0.00027825675,0.0005359078,0.00013729095,0.0019348915],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99250364,0.0031381818,0.0004137007,0.0014385219,0.0017773238,0.00072857377],"domain_scores_gemma":[0.9898587,0.0036937005,0.00075767044,0.0042252373,0.0009485515,0.00051608187],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009123986,0.00097205763,0.0019573271,0.0008162899,0.0012281125,0.0023820288,0.004823263,0.0020283754,0.0016508971],"category_scores_gemma":[0.023237448,0.00074631657,0.0010706822,0.0015349772,0.001911612,0.006674054,0.007503983,0.0030458318,0.0005017615],"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":[0.0014321203,0.00059675344,0.008847932,0.00018236847,0.00019112161,0.0003680947,0.0006352293,0.5155469,0.005872188,0.098027185,0.005581958,0.3627182],"study_design_scores_gemma":[0.00003599866,0.00008830424,0.00026532673,0.000013149308,0.000010106202,0.00007926524,0.00004941086,0.94700146,0.0018601371,0.049859047,0.0007238472,0.000014040696],"about_ca_topic_score_codex":0.0022927271,"about_ca_topic_score_gemma":0.002524835,"teacher_disagreement_score":0.009123986,"about_ca_system_score_codex":0.0021119094,"about_ca_system_score_gemma":0.0036328007,"threshold_uncertainty_score":0.04825282},"labels":[],"label_agreement":null},{"id":"W4413986970","doi":"10.14778/3748191.3748195","title":"Déjà Vu: Efficient Video-Language Query Engine with Learning-Based Inter-Frame Computation Reuse","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Computer science; Reuse; Déjà vu; Frame (networking); Computation; Artificial intelligence; Natural language processing; Programming language; Computer network; Engineering; Psychology","score_opus":0.0037931050791894048,"score_gpt":0.24165399323101416,"score_spread":0.23786088815182477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413986970","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":[0.016738588,0.0006863781,0.9098299,0.00026856852,0.00017663414,0.00020975144,0.000969156,0.068568826,0.0025521358],"genre_scores_gemma":[0.2655531,0.00048520198,0.71235996,0.00061436347,0.00011419075,0.00044830196,0.006289948,0.0038264822,0.010308537],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991026,0.000108472224,0.000061059225,0.00029672464,0.0003348884,0.000096172764],"domain_scores_gemma":[0.9991986,0.00026715797,0.000054332053,0.00022245738,0.00019572704,0.000061798935],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00096841686,0.0014646769,0.00088507467,0.0007565432,0.00040791652,0.0016018403,0.0036867198,0.0010507066,0.005818575],"category_scores_gemma":[0.0044962894,0.0005685488,0.0008574761,0.00076104584,0.0006027001,0.0036540786,0.00277851,0.0015904299,0.003189101],"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":[0.0012546633,0.0004435887,0.0024144184,0.00047196314,0.00016468161,0.0004448416,0.00036402515,0.09586532,0.06782362,0.025041835,0.07743357,0.72827756],"study_design_scores_gemma":[0.00007018345,0.00011628122,0.0001778118,0.00000984751,0.000016736265,0.00011064248,0.000053829943,0.95374894,0.028600836,0.0070222253,0.010042507,0.00003021332],"about_ca_topic_score_codex":0.009119345,"about_ca_topic_score_gemma":0.013079966,"teacher_disagreement_score":0.009119345,"about_ca_system_score_codex":0.0010095306,"about_ca_system_score_gemma":0.0014366672,"threshold_uncertainty_score":0.019465089},"labels":[],"label_agreement":null},{"id":"W4414003971","doi":"10.14778/3749646.3749707","title":"OasisDB: An Oblivious and Scalable System for Relational Data","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Cryptography and Data Security","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Scalability; Computer science; Relational database; Database","score_opus":0.02953292981573211,"score_gpt":0.26023942873683936,"score_spread":0.23070649892110726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414003971","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":[0.039210696,0.0033772318,0.7296429,0.0018008624,0.00050722633,0.0011468995,0.0042346166,0.20314828,0.016931279],"genre_scores_gemma":[0.39330852,0.0022710015,0.55639577,0.0016191845,0.0004035536,0.0012587488,0.019346807,0.00776568,0.0176307],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99423367,0.0007195132,0.0005883259,0.0006833029,0.0032821458,0.0004931127],"domain_scores_gemma":[0.9909071,0.0009833274,0.00047289397,0.0062653525,0.0008738449,0.0004974756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038151336,0.0010955976,0.0010383461,0.0020994092,0.0015922535,0.004769771,0.00573646,0.0011516968,0.00863201],"category_scores_gemma":[0.0112034,0.0013222791,0.0010971292,0.0027513076,0.0022063702,0.0127484165,0.010866208,0.0029473605,0.005518394],"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":[0.002730276,0.00067828555,0.010984435,0.0017894107,0.000724992,0.0006802458,0.0016462822,0.027774105,0.07945528,0.2088656,0.24985205,0.41481897],"study_design_scores_gemma":[0.0009262885,0.0008332031,0.00388684,0.0002608258,0.00026154442,0.0014570286,0.00057805784,0.34193394,0.11366032,0.12681651,0.40894687,0.00043852231],"about_ca_topic_score_codex":0.0048283124,"about_ca_topic_score_gemma":0.003608294,"teacher_disagreement_score":0.00863201,"about_ca_system_score_codex":0.0018892364,"about_ca_system_score_gemma":0.0041403766,"threshold_uncertainty_score":0.02887696},"labels":[],"label_agreement":null},{"id":"W4414003972","doi":"10.14778/3749646.3749704","title":"NaviX: A Native Vector Index Design for Graph DBMSs With Robust Predicate-Agnostic Search Performance","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"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; Predicate (mathematical logic); Graph; Programming language; Theoretical computer science","score_opus":0.017912365227868456,"score_gpt":0.22755192029124968,"score_spread":0.20963955506338122,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414003972","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":[0.053276926,0.0014339159,0.7706548,0.00052614306,0.00031694045,0.000777078,0.0031691235,0.15859728,0.011247756],"genre_scores_gemma":[0.2676563,0.0005961309,0.7103581,0.0005919929,0.00012384054,0.00075302506,0.008035665,0.0047465144,0.007138384],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99831784,0.00020336217,0.00020006503,0.00028979217,0.0008576397,0.00013131401],"domain_scores_gemma":[0.9979875,0.00041020077,0.0001640073,0.0007511929,0.0005390656,0.0001479827],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016820993,0.0010189931,0.0007485824,0.0012945574,0.0007281184,0.0026590354,0.0037467817,0.00071649964,0.0049714404],"category_scores_gemma":[0.005775984,0.0006699076,0.000581257,0.0017862746,0.0008047378,0.0044999453,0.002781831,0.0013153151,0.001728324],"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":[0.0023814132,0.0007458909,0.009116405,0.0012570516,0.0003588266,0.0002970961,0.00089104316,0.055569995,0.08858375,0.067277364,0.14374311,0.62977815],"study_design_scores_gemma":[0.00057351287,0.0010100851,0.0025549536,0.0000745843,0.00010176058,0.000435163,0.00026988453,0.78613734,0.08360367,0.03267367,0.09236603,0.00019937231],"about_ca_topic_score_codex":0.0068057007,"about_ca_topic_score_gemma":0.008323879,"teacher_disagreement_score":0.0068057007,"about_ca_system_score_codex":0.0012710007,"about_ca_system_score_gemma":0.0018132405,"threshold_uncertainty_score":0.016631126},"labels":[],"label_agreement":null},{"id":"W4414004064","doi":"10.14778/3749646.3749702","title":"ThriftLLM: On Cost-Effective Selection of Large Language Models for Classification Queries","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Selection (genetic algorithm); Computer science; Natural language processing; Artificial intelligence; Information retrieval","score_opus":0.09834815259862227,"score_gpt":0.4145340017749151,"score_spread":0.31618584917629283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414004064","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":[0.061506424,0.0015982151,0.9186172,0.002612154,0.0001478318,0.0006784962,0.0010196805,0.010402101,0.0034179704],"genre_scores_gemma":[0.47299284,0.0005788163,0.5124458,0.0020082432,0.00037877177,0.0008683321,0.0048030014,0.0012832688,0.0046408447],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9942926,0.0018681596,0.00036383167,0.0010866482,0.0015834753,0.0008052532],"domain_scores_gemma":[0.9835699,0.012198033,0.00073612906,0.0018809317,0.0010425373,0.0005724457],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0065317727,0.0030163815,0.003507783,0.0019121425,0.0016166731,0.003047073,0.004863843,0.0030646943,0.004626693],"category_scores_gemma":[0.023223119,0.0010981101,0.0027608671,0.0028073138,0.0012318897,0.0053878557,0.0050296877,0.0042795637,0.0019465078],"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":[0.00090184726,0.0008895669,0.0063492768,0.00037846432,0.00025222727,0.00045609602,0.00052074966,0.54396915,0.0044403602,0.018705614,0.034252908,0.38888365],"study_design_scores_gemma":[0.00004192943,0.000049267932,0.00011997075,0.000006909867,0.00001790088,0.000055114666,0.000048889793,0.98972356,0.0004979025,0.008776002,0.00065439934,0.000008099775],"about_ca_topic_score_codex":0.012571327,"about_ca_topic_score_gemma":0.016893066,"teacher_disagreement_score":0.012571327,"about_ca_system_score_codex":0.0036158192,"about_ca_system_score_gemma":0.004265406,"threshold_uncertainty_score":0.034543753},"labels":[],"label_agreement":null},{"id":"W4414078184","doi":"10.14778/3749646.3749706","title":"Robust Recursive Query Parallelism in Graph Database Management Systems","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Graph Theory and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computation; Graph; Variety (cybernetics); Node (physics); Query optimization; Graph database; Parallelism (grammar); Directed acyclic graph; Materialized view","score_opus":0.013331584993892084,"score_gpt":0.21476247281281138,"score_spread":0.2014308878189193,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414078184","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":[0.5050397,0.0014558809,0.472182,0.0011221486,0.00009980549,0.00032408122,0.00041638996,0.013500871,0.0058591235],"genre_scores_gemma":[0.8541811,0.00017883112,0.14370163,0.00023113747,0.000028833512,0.00010473051,0.00037407866,0.00032216738,0.00087740395],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99674916,0.0009958649,0.00023012921,0.0007840532,0.00083057384,0.00041024774],"domain_scores_gemma":[0.996409,0.001333902,0.00032660022,0.0012765203,0.00046582578,0.000188106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030248268,0.00088881596,0.0009470288,0.0008402522,0.001020232,0.002106989,0.0029002123,0.0009856339,0.0011412557],"category_scores_gemma":[0.008354736,0.00061961095,0.0005976301,0.0016720297,0.001305049,0.0038712863,0.0022225694,0.0012299371,0.00040450387],"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":[0.00077930006,0.0004691369,0.005980574,0.00027338768,0.00013039605,0.00024644355,0.0005689008,0.7995254,0.032873668,0.021855006,0.00803037,0.12926747],"study_design_scores_gemma":[0.000057911482,0.000117669515,0.0008444988,0.0000069432863,0.000021608057,0.000059038703,0.00012010322,0.9751037,0.007334762,0.014725793,0.0015878774,0.000020184285],"about_ca_topic_score_codex":0.0077437116,"about_ca_topic_score_gemma":0.009581782,"teacher_disagreement_score":0.0077437116,"about_ca_system_score_codex":0.001783337,"about_ca_system_score_gemma":0.0025150054,"threshold_uncertainty_score":0.015997052},"labels":[],"label_agreement":null},{"id":"W4414266787","doi":"10.14778/3750601.3750638","title":"SQL:Trek Automated Index Design at Airbnb","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Air Canada","funders":"","keywords":"Index (typography); Scalability; Index selection; False positive paradox; Selection (genetic algorithm); Database index; Compiler; Relational database","score_opus":0.011183793592322706,"score_gpt":0.22989208488466692,"score_spread":0.2187082912923442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414266787","genre_codex":"software","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":[0.046322364,0.001137004,0.38391688,0.00058698544,0.00036399218,0.00052466366,0.0049405643,0.5482848,0.0139226485],"genre_scores_gemma":[0.36195397,0.0006819779,0.54159665,0.00097522576,0.00019384103,0.0006050255,0.027363054,0.052173797,0.014456521],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9940971,0.0010720184,0.00048511455,0.0011195025,0.0027132737,0.0005128666],"domain_scores_gemma":[0.9927532,0.0020706027,0.00039718527,0.0024915466,0.0019195465,0.00036792888],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005140516,0.0017124043,0.0012883162,0.0013989854,0.0007947845,0.004414225,0.0040711467,0.0010332183,0.015791016],"category_scores_gemma":[0.0140424995,0.001362544,0.001225554,0.0015495443,0.0011439616,0.00534754,0.0026248854,0.0022679176,0.011149406],"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":[0.005073494,0.00091845234,0.011456672,0.0013882045,0.00039930674,0.00075364596,0.00085049385,0.051671475,0.09659583,0.02289318,0.33268306,0.47531617],"study_design_scores_gemma":[0.00083133334,0.00069378776,0.003599538,0.00011628235,0.00008495302,0.0005101602,0.00020340356,0.7817953,0.09665499,0.01593696,0.099366456,0.00020684328],"about_ca_topic_score_codex":0.0046679066,"about_ca_topic_score_gemma":0.004255686,"teacher_disagreement_score":0.015791016,"about_ca_system_score_codex":0.0016543919,"about_ca_system_score_gemma":0.003537254,"threshold_uncertainty_score":0.052826226},"labels":[],"label_agreement":null},{"id":"W4414266922","doi":"10.14778/3750601.3750685","title":"Beyond Quacking: Deep Integration of Language Models and RAG into DuckDB","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Polytechnique Montréal","funders":"","keywords":"Schema (genetic algorithms); Data integration; Data modeling; Relational database; Context (archaeology); SQL; Language model; Context model; Rapid prototyping","score_opus":0.008653647840289753,"score_gpt":0.2656501601318978,"score_spread":0.256996512291608,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414266922","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":[0.012242266,0.0016308263,0.86595446,0.0017149124,0.00028729418,0.00027268205,0.005726524,0.1044629,0.0077081774],"genre_scores_gemma":[0.11741411,0.00089793996,0.8488277,0.0011892325,0.00008124131,0.00030233944,0.017661657,0.008379709,0.005245999],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99700254,0.00084503647,0.0003512854,0.0006520029,0.000999023,0.00015006254],"domain_scores_gemma":[0.9959798,0.0014994112,0.00012862557,0.0017060064,0.0005069432,0.00017920196],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0050998665,0.00091859174,0.0012209076,0.0020315826,0.0009506996,0.0055923136,0.0038577374,0.0010955678,0.0058353585],"category_scores_gemma":[0.014781449,0.0012374253,0.0015436639,0.0018593735,0.0013597911,0.007818766,0.005409986,0.0028982467,0.0035204447],"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":[0.00061044964,0.00044937877,0.007800912,0.0012094226,0.00046225687,0.00047319685,0.0015381898,0.13683805,0.005464096,0.17954634,0.10210148,0.5635063],"study_design_scores_gemma":[0.00008479683,0.000059870406,0.00064593536,0.00018117872,0.000064971915,0.00019672618,0.0002470612,0.69953233,0.007195883,0.14579868,0.14587241,0.00012014109],"about_ca_topic_score_codex":0.017485162,"about_ca_topic_score_gemma":0.029461414,"teacher_disagreement_score":0.017485162,"about_ca_system_score_codex":0.0019914764,"about_ca_system_score_gemma":0.0031319982,"threshold_uncertainty_score":0.034766793},"labels":[],"label_agreement":null},{"id":"W4414267806","doi":"10.14778/3750601.3750656","title":"Analytics Are Heavy. The DBMS Is Busy. When Will My Mission-Critical Transaction Start Running?","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Scheduling (production processes); Preemption; Context switch; Database transaction; Transaction processing; Concurrency; Concurrency control; Analytics","score_opus":0.04457798281609549,"score_gpt":0.2847344295664768,"score_spread":0.24015644675038134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414267806","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":[0.3417118,0.014781289,0.25987467,0.116222106,0.009402181,0.00062731747,0.0025052328,0.039408255,0.21546723],"genre_scores_gemma":[0.87349844,0.0033454942,0.059140008,0.012340424,0.001827803,0.00014718743,0.0011620903,0.0025021636,0.04603638],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99785334,0.0002928957,0.00009894015,0.00030945425,0.0009875742,0.000457729],"domain_scores_gemma":[0.993676,0.0015535627,0.00059496047,0.0014382132,0.001173029,0.0015640999],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032681543,0.00077376526,0.00056767074,0.0005237228,0.0030416097,0.0059681423,0.0012453379,0.0013912441,0.011234768],"category_scores_gemma":[0.010479435,0.0008208628,0.00042254597,0.0007886239,0.0015292147,0.00742852,0.0028109015,0.0041655875,0.006958924],"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":[0.0039437534,0.0006387805,0.038483504,0.0007835634,0.00023305119,0.002544948,0.0073794713,0.006175033,0.085939504,0.11283607,0.36478862,0.3762537],"study_design_scores_gemma":[0.0003106653,0.00081259664,0.022358822,0.0005098886,0.00020117128,0.0036164848,0.007389039,0.055088732,0.041075278,0.20583071,0.66247463,0.0003321017],"about_ca_topic_score_codex":0.0037486712,"about_ca_topic_score_gemma":0.0042813746,"teacher_disagreement_score":0.011234768,"about_ca_system_score_codex":0.0010276184,"about_ca_system_score_gemma":0.002062196,"threshold_uncertainty_score":0.037584007},"labels":[],"label_agreement":null},{"id":"W7080011846","doi":"10.14778/3749646.3749664","title":"Diva: Dynamic Range Filter for Var-Length Keys and Queries","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Range query (database); Range (aeronautics); Trie; Probabilistic logic; Filter (signal processing); Data structure; Query optimization; State (computer science)","score_opus":0.0073251686661112755,"score_gpt":0.21609304688765688,"score_spread":0.2087678782215456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7080011846","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":[0.04168691,0.0038270196,0.89249873,0.0011614755,0.00033033648,0.0006283654,0.0075528035,0.046853833,0.005460489],"genre_scores_gemma":[0.3155295,0.0014008738,0.6572483,0.0012567234,0.00037585694,0.0010891781,0.015335482,0.0019305341,0.005833591],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9937018,0.00086553994,0.00066778006,0.0012631668,0.0030355188,0.00046630146],"domain_scores_gemma":[0.98068696,0.0078095095,0.0012763888,0.007634833,0.0022372967,0.0003550268],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046893074,0.0014868907,0.0024109397,0.0034073456,0.0015103999,0.00405153,0.004415493,0.0023077202,0.0048118504],"category_scores_gemma":[0.029874502,0.0010683215,0.0016433077,0.004066776,0.0017164088,0.012130644,0.0053290906,0.0025262823,0.003937636],"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":[0.001979976,0.00041174173,0.018329876,0.0010260377,0.00023924123,0.00046035773,0.0008160055,0.05175481,0.028929116,0.054288514,0.07769878,0.7640655],"study_design_scores_gemma":[0.00022888642,0.0005222809,0.0028841428,0.00016985355,0.00008914669,0.001402784,0.00052298716,0.7946814,0.04259734,0.0935015,0.06323047,0.00016930692],"about_ca_topic_score_codex":0.004088266,"about_ca_topic_score_gemma":0.0057776985,"teacher_disagreement_score":0.0048118504,"about_ca_system_score_codex":0.0018189291,"about_ca_system_score_gemma":0.0034343433,"threshold_uncertainty_score":0.024799705},"labels":[],"label_agreement":null},{"id":"W7118911591","doi":"10.14778/3773731.3773734","title":"Sampling-Based Predictive Database Buffer Management","year":2025,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Workload; Buffer (optical fiber); Sample (material); Volume (thermodynamics); Set (abstract data type); Data access; Data set","score_opus":0.011922581415300254,"score_gpt":0.25314774292857595,"score_spread":0.2412251615132757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7118911591","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":[0.0561664,0.00080797856,0.9361778,0.00024318635,0.0000679452,0.00011999775,0.00022918588,0.0051614773,0.0010259446],"genre_scores_gemma":[0.811875,0.00039915755,0.18593371,0.00016624008,0.00007657797,0.00014501695,0.00041457536,0.00016723853,0.0008225798],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99833316,0.00033625044,0.000113882365,0.00042741274,0.000635663,0.0001536907],"domain_scores_gemma":[0.9936219,0.002890103,0.0006984888,0.0015268025,0.0010330498,0.00022972531],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022302936,0.000932616,0.0010646497,0.0011170055,0.00074629387,0.0015019322,0.0035878804,0.0005593653,0.00087274547],"category_scores_gemma":[0.010212601,0.00063342357,0.00042499823,0.0015037553,0.0006811727,0.0024954134,0.0012122084,0.0010553066,0.000274573],"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":[0.0007493925,0.00041613824,0.021423591,0.00021765233,0.00018496762,0.00016446153,0.00063037267,0.55010074,0.029962884,0.009059271,0.0059006494,0.38118982],"study_design_scores_gemma":[0.000016182517,0.00006211455,0.00066717155,0.0000078735175,0.00002574723,0.000060290477,0.00006205949,0.9892351,0.0068400507,0.0021517673,0.00085353333,0.000018065923],"about_ca_topic_score_codex":0.008015106,"about_ca_topic_score_gemma":0.012139035,"teacher_disagreement_score":0.008015106,"about_ca_system_score_codex":0.0012624977,"about_ca_system_score_gemma":0.002369787,"threshold_uncertainty_score":0.015936911},"labels":[],"label_agreement":null},{"id":"W763614111","doi":"10.14778/2735479.2735491","title":"Mining revenue-maximizing bundling configuration","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Consumer Market Behavior and Pricing","field":"Business, Management and Accounting","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Singapore Telecommunications Limited","keywords":"Bundle; Revenue; Perspective (graphical); Set (abstract data type); Heuristic; Computer science; Marketing; Data set; Business; Artificial intelligence","score_opus":0.04354427650522499,"score_gpt":0.240351671876838,"score_spread":0.19680739537161301,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W763614111","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":[0.6737212,0.001958013,0.3047351,0.0016825624,0.00012160135,0.00055476,0.0050157886,0.00271908,0.009491988],"genre_scores_gemma":[0.7769,0.0004930401,0.20893967,0.00019679997,0.000088925844,0.00024370343,0.009862424,0.00035653816,0.0029189033],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.997607,0.0006586522,0.00020692294,0.0007376214,0.00045522675,0.00033455613],"domain_scores_gemma":[0.9944371,0.002609757,0.0008172338,0.0007453662,0.0008302365,0.00056036096],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023459196,0.0020841288,0.0024732547,0.0041017616,0.0015432347,0.0025575126,0.0023330585,0.0017639289,0.0039625186],"category_scores_gemma":[0.012607777,0.0018238123,0.0016565102,0.0049254196,0.0010271119,0.004876174,0.0015671245,0.0014327291,0.0013254307],"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":[0.0010103734,0.0014724012,0.05029054,0.00071546686,0.0002957006,0.0015322652,0.00049652683,0.61033475,0.0069659147,0.023815498,0.0364954,0.26657516],"study_design_scores_gemma":[0.00006969328,0.000113662,0.002153252,0.000031875636,0.000040913634,0.00029624967,0.00023924727,0.9592996,0.0013994726,0.03461512,0.0017145004,0.00002629606],"about_ca_topic_score_codex":0.0040901825,"about_ca_topic_score_gemma":0.0059769102,"teacher_disagreement_score":0.0041017616,"about_ca_system_score_codex":0.0015090333,"about_ca_system_score_gemma":0.0023616562,"threshold_uncertainty_score":0.013255954},"labels":[],"label_agreement":null},{"id":"W79208629","doi":"10.14778/2732269.2732275","title":"Computing k-regret minimizing sets","year":2014,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Regret; Dimension (graph theory); Relaxation (psychology); Set (abstract data type); Greedy algorithm; Linear programming relaxation; Dynamic programming; Linear programming; Duality (order theory); Mathematics; Combinatorics; Computer science; Mathematical optimization; Algorithm; Statistics","score_opus":0.011085699081155623,"score_gpt":0.21721503497013323,"score_spread":0.2061293358889776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W79208629","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":[0.11136877,0.0010736702,0.87516546,0.0016216907,0.000108389075,0.00030686203,0.0016570407,0.002060795,0.0066373283],"genre_scores_gemma":[0.45373535,0.00051808846,0.53486,0.00077576726,0.00022168714,0.0006087493,0.004102672,0.0007210879,0.004456599],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99643505,0.0014025902,0.00021974534,0.00085058867,0.0006826647,0.00040939002],"domain_scores_gemma":[0.98898727,0.007886954,0.0008022841,0.0012414826,0.0006963024,0.00038564045],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039756577,0.0017362292,0.003138189,0.0019181216,0.001086796,0.0032677318,0.0026688501,0.0020753082,0.0041907504],"category_scores_gemma":[0.021193119,0.0011706755,0.0017472368,0.0028300998,0.0017332474,0.0037163552,0.0022444702,0.0025355022,0.0010292243],"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":[0.0005436638,0.00027807776,0.0022509035,0.0003601787,0.00019812428,0.00012787474,0.000323603,0.8541072,0.002071622,0.039221726,0.011694138,0.08882293],"study_design_scores_gemma":[0.000037623682,0.000083495506,0.00032851045,0.00002984834,0.000022813172,0.000046376666,0.00007783878,0.93958235,0.0014273288,0.05702335,0.0013212548,0.00001924856],"about_ca_topic_score_codex":0.0023215548,"about_ca_topic_score_gemma":0.002960521,"teacher_disagreement_score":0.0041907504,"about_ca_system_score_codex":0.002865898,"about_ca_system_score_gemma":0.0019207625,"threshold_uncertainty_score":0.021025538},"labels":[],"label_agreement":null},{"id":"W808055529","doi":"10.14778/2850578.2850581","title":"From competition to complementarity","year":2015,"lang":"en","type":"article","venue":"Proceedings of the VLDB Endowment","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":216,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Complementarity (molecular biology); Computer science; Maximization; Competition (biology); Mathematical optimization; Cellular automaton; Set (abstract data type); Margin (machine learning); Theoretical computer science; Artificial intelligence; Machine learning; Mathematics","score_opus":0.027410872042148204,"score_gpt":0.27390038399454764,"score_spread":0.24648951195239943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W808055529","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":[0.0673905,0.0006615175,0.9090972,0.0017887207,0.000055638164,0.00015977806,0.00036957493,0.0003597548,0.020117365],"genre_scores_gemma":[0.85888934,0.0007544078,0.13008921,0.0006618105,0.00017622959,0.00037623514,0.0005057031,0.0002129608,0.008334192],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984126,0.0005992377,0.000049519356,0.00047858345,0.00028990177,0.00017011173],"domain_scores_gemma":[0.99477273,0.004141596,0.0003329836,0.00027590178,0.00022155781,0.0002552828],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016520732,0.0014368363,0.0015955004,0.00092815974,0.001009876,0.0019789387,0.0016155206,0.0017907702,0.0058360393],"category_scores_gemma":[0.008870943,0.00081123604,0.00158204,0.0012089493,0.0021179426,0.0036222877,0.0025144683,0.0022588496,0.00048580853],"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":[0.00013213586,0.00012683167,0.0019142384,0.00026211803,0.00011974801,0.00035439717,0.00023374414,0.61669403,0.0016686558,0.33747733,0.007088011,0.03392882],"study_design_scores_gemma":[0.000023995808,0.00003894291,0.0003045998,0.00001812491,0.000028580509,0.0000916241,0.000044695003,0.79916584,0.00064364495,0.19761762,0.0020070581,0.000015333579],"about_ca_topic_score_codex":0.005331788,"about_ca_topic_score_gemma":0.004615034,"teacher_disagreement_score":0.0058360393,"about_ca_system_score_codex":0.0020694535,"about_ca_system_score_gemma":0.0013846469,"threshold_uncertainty_score":0.019523501},"labels":[],"label_agreement":null}]}