{"meta":{"query_hash":"323cac591371","filters":{"venue":"ACM SIGMOD Record"},"cohort_total":67,"direct_labels_cover":0,"predictions_cover":67,"exported":67,"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/323cac591371","api":"https://metacan.xera.ac/api/v1/cohort?venue=ACM+SIGMOD+Record"},"results":[{"id":"W1974306683","doi":"10.1145/335191.335495","title":"An approximate search engine for structural databases","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","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":"Western University","funders":"","keywords":"Computer science; Parsing; Database; Information retrieval; XML; Set (abstract data type); Subgraph isomorphism problem; Graph database; Graph; Search engine; Heuristic; Data structure; XML database; Data mining; Theoretical computer science; World Wide Web; Programming language; Artificial intelligence","score_opus":0.020294907655088994,"score_gpt":0.28663512953624565,"score_spread":0.26634022188115664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1974306683","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":[0.04198588,0.0025359115,0.9270063,0.0012527156,0.00014283239,0.00030252093,0.0019233814,0.013794337,0.011056085],"genre_scores_gemma":[0.21682738,0.0012108603,0.7697039,0.00032501406,0.00011914665,0.00036348822,0.0047321753,0.0006761663,0.0060419175],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974801,0.00063462334,0.00026887842,0.00042651698,0.0009774657,0.0002125138],"domain_scores_gemma":[0.9942643,0.0025661404,0.00023720406,0.0019160465,0.0008597318,0.00015658626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024226026,0.0007050507,0.0015930526,0.0031100728,0.0011669392,0.003480031,0.0030215238,0.0018402131,0.006959085],"category_scores_gemma":[0.017194767,0.0006955827,0.0008478307,0.0053767944,0.001125534,0.010343605,0.0024902995,0.0011313704,0.0035126284],"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":[0.0015492279,0.0003570989,0.0033401244,0.00083827134,0.00016218041,0.00026767363,0.0005954155,0.0973079,0.010958486,0.27526334,0.04544064,0.56391954],"study_design_scores_gemma":[0.0002317261,0.00018799499,0.0004993098,0.000055568824,0.00009838314,0.00043557634,0.00024738204,0.7359373,0.006211238,0.22570424,0.030349074,0.000042171934],"about_ca_topic_score_codex":0.004085896,"about_ca_topic_score_gemma":0.0061826888,"teacher_disagreement_score":0.006959085,"about_ca_system_score_codex":0.0021247913,"about_ca_system_score_gemma":0.0024711452,"threshold_uncertainty_score":0.023280442},"labels":[],"label_agreement":null},{"id":"W1975270911","doi":"10.1145/601858.601873","title":"Toward autonomic computing with DB2 universal database","year":2002,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","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":"IBM (Canada)","funders":"","keywords":"Computer science; Total cost of ownership; Database; IBM; Unix; Scale (ratio); Software; Operating system","score_opus":0.03708365375936235,"score_gpt":0.23390888169402854,"score_spread":0.1968252279346662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1975270911","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":[0.039123118,0.010466701,0.84916884,0.01760706,0.0019519456,0.0002561576,0.00015704543,0.009308469,0.07196065],"genre_scores_gemma":[0.33965477,0.008156983,0.62361217,0.007203508,0.0016411956,0.0003936905,0.0005742595,0.0007842967,0.017979153],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99750274,0.00056259846,0.00014621801,0.0002932248,0.0011366556,0.00035851562],"domain_scores_gemma":[0.99782723,0.00025049248,0.00009885786,0.00064746285,0.0006324946,0.0005434373],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051236027,0.0005104946,0.0006559832,0.0010167995,0.0012242607,0.0051694815,0.0026147072,0.0014357175,0.0020404826],"category_scores_gemma":[0.00582315,0.00063363614,0.0006900451,0.0015140099,0.0012700411,0.00826508,0.004906474,0.0040711393,0.001164829],"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":[0.00044188273,0.00037097474,0.0060883677,0.00028923078,0.00011534182,0.0005086645,0.0014545096,0.010207325,0.010946211,0.5670201,0.0546818,0.3478756],"study_design_scores_gemma":[0.00012267477,0.00030625684,0.0019034671,0.00025803162,0.00012112479,0.0011056408,0.000737482,0.16552995,0.010842292,0.3697217,0.4492526,0.000098711265],"about_ca_topic_score_codex":0.0021234148,"about_ca_topic_score_gemma":0.0021088135,"teacher_disagreement_score":0.0051694815,"about_ca_system_score_codex":0.00089875935,"about_ca_system_score_gemma":0.0016096529,"threshold_uncertainty_score":0.02709651},"labels":[],"label_agreement":null},{"id":"W1978094234","doi":"10.1145/776985.776996","title":"Peer-to-peer","year":2003,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Peer-to-Peer Network Technologies","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":"University of Alberta","funders":"","keywords":"Computer science; Peer-to-peer; World Wide Web; Programming language; Computer network","score_opus":0.026189099824546435,"score_gpt":0.26945035843242077,"score_spread":0.24326125860787434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1978094234","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":[0.0017476073,0.008452162,0.050773293,0.016358573,0.02989608,0.005020139,0.023820192,0.041455913,0.8224761],"genre_scores_gemma":[0.013604813,0.0035772994,0.00850944,0.0023078122,0.0033074706,0.0014315755,0.01262695,0.0038459736,0.9507887],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9918652,0.0019033132,0.0004430853,0.0019973866,0.0026154716,0.0011755773],"domain_scores_gemma":[0.9797233,0.004701608,0.00076030445,0.005936889,0.005167911,0.0037100127],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005668618,0.0040738527,0.005119011,0.0029670268,0.004850968,0.0083034085,0.006906045,0.004066936,0.8474148],"category_scores_gemma":[0.026823102,0.0024164452,0.0016102568,0.003964817,0.0034040262,0.011084218,0.011617521,0.005678132,0.8847575],"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":[0.00031446444,0.00018275584,0.00011775312,0.001123073,0.00006972371,0.0002278504,0.00010899371,0.0002159148,0.006864589,0.0028257368,0.8838964,0.10405279],"study_design_scores_gemma":[0.00009527686,0.000063739455,0.00024200948,0.00021301405,0.000019672596,0.00015856755,0.000050782157,0.00068566017,0.0008387203,0.0034162204,0.99418503,0.00003134406],"about_ca_topic_score_codex":0.0025990782,"about_ca_topic_score_gemma":0.0028785204,"teacher_disagreement_score":0.8474148,"about_ca_system_score_codex":0.0019476884,"about_ca_system_score_gemma":0.002923073,"threshold_uncertainty_score":0.21764427},"labels":[],"label_agreement":null},{"id":"W1986871321","doi":"10.1145/362084.362142","title":"Knowledge discovery in data warehouses","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Mining Algorithms and Applications","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":"Computer science; Online analytical processing; Data warehouse; Automatic summarization; Terabyte; Knowledge extraction; Data mining; Process (computing); Data science; Information retrieval; Database","score_opus":0.06937896728329003,"score_gpt":0.3180131510689414,"score_spread":0.24863418378565139,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1986871321","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":[0.015636725,0.011949938,0.9507564,0.005821237,0.00027723509,0.00060711976,0.0034066807,0.0022765796,0.009268159],"genre_scores_gemma":[0.10384506,0.009233894,0.87621295,0.0013216918,0.00025523466,0.00045856403,0.0056927963,0.00017628861,0.0028034598],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98622257,0.0046819965,0.0018201215,0.0019953547,0.0047801565,0.00049977476],"domain_scores_gemma":[0.9754484,0.015392447,0.0013439809,0.0047803395,0.0024713958,0.00056336186],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012709628,0.0010738703,0.002919577,0.012086702,0.0023036494,0.014161332,0.004637019,0.0022897068,0.0024289328],"category_scores_gemma":[0.039689656,0.0019351994,0.003017743,0.019263865,0.0028625492,0.017575229,0.0065029846,0.003447293,0.0018548851],"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":[0.0002697584,0.00034005602,0.008212704,0.0036792392,0.00092464976,0.0020725704,0.0034128346,0.05289418,0.00259275,0.45177203,0.02502171,0.44880757],"study_design_scores_gemma":[0.000051805782,0.000053456293,0.0009790881,0.0005878833,0.00017048279,0.0008758942,0.0014026276,0.16042301,0.003199575,0.7658007,0.06637292,0.00008244615],"about_ca_topic_score_codex":0.003744675,"about_ca_topic_score_gemma":0.00376621,"teacher_disagreement_score":0.014161332,"about_ca_system_score_codex":0.0021198038,"about_ca_system_score_gemma":0.0029144345,"threshold_uncertainty_score":0.06721574},"labels":[],"label_agreement":null},{"id":"W1987552238","doi":"10.1145/373626.373697","title":"Constraints for semistructured data and XML","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":107,"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; Data integrity; XML database; Database; XML; Programming language; World Wide Web","score_opus":0.04716285210807767,"score_gpt":0.3007392847096654,"score_spread":0.2535764326015877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1987552238","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":[0.0041223494,0.06307293,0.83961034,0.017699817,0.0040367623,0.00041636062,0.0038067389,0.00083586277,0.066398926],"genre_scores_gemma":[0.107598625,0.09722275,0.73301554,0.0069488715,0.009230489,0.0014757301,0.007854723,0.00067187153,0.035981424],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99187976,0.002309519,0.001538803,0.0010038845,0.0029676766,0.00030043913],"domain_scores_gemma":[0.98884434,0.0067701763,0.0012149531,0.0013411975,0.0014834495,0.00034593305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056467066,0.0011454512,0.0011105235,0.0033817512,0.0019914764,0.0067334925,0.0023073656,0.0027061314,0.011041507],"category_scores_gemma":[0.016371144,0.0008024803,0.0013228782,0.00832441,0.00446942,0.0137761235,0.0030442148,0.0042237886,0.0033559422],"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":[0.000017036102,0.00001303192,0.000112917536,0.0005866311,0.000020133884,0.0003109178,0.00037919392,0.0017800432,0.0005785097,0.93819827,0.015715735,0.042287517],"study_design_scores_gemma":[0.000012270519,0.000018264373,0.00014655088,0.00044723184,0.000018015848,0.0007137829,0.00022577861,0.004314009,0.0008362781,0.7099551,0.28327197,0.000040700364],"about_ca_topic_score_codex":0.0040536025,"about_ca_topic_score_gemma":0.0031748868,"teacher_disagreement_score":0.011041507,"about_ca_system_score_codex":0.002177354,"about_ca_system_score_gemma":0.0021189142,"threshold_uncertainty_score":0.036937475},"labels":[],"label_agreement":null},{"id":"W1998014869","doi":"10.1145/335191.335476","title":"Of XML and databases (panel session)","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"Resverlogix (Canada)","funders":"","keywords":"Computer science; XML database; Database; Session (web analytics); XML; XML Schema Editor; Efficient XML Interchange; World Wide Web; XML Base; Streaming XML; XML validation; Information retrieval; XML Encryption","score_opus":0.04303868654853697,"score_gpt":0.2752964978959816,"score_spread":0.2322578113474446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1998014869","genre_codex":"commentary","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":[0.0013899531,0.10971872,0.0111874,0.36726278,0.23189595,0.0006362856,0.0015295406,0.0006510946,0.2757283],"genre_scores_gemma":[0.0081395125,0.06961299,0.004834391,0.09356745,0.18970487,0.00079339405,0.001831316,0.0005896195,0.63092655],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99822015,0.0004579907,0.00007771046,0.00034993896,0.00061673013,0.0002774325],"domain_scores_gemma":[0.9951515,0.0016509837,0.00020437804,0.00036579225,0.0012563376,0.0013710129],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070897657,0.0013984931,0.0010701795,0.0017934965,0.0034162495,0.008237874,0.0017418799,0.00979725,0.09609054],"category_scores_gemma":[0.0058942,0.0007952493,0.0013471338,0.0025679034,0.0018360881,0.011965331,0.005426606,0.011214568,0.03380161],"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":[0.000020496733,0.00002519401,0.0000909416,0.00006755925,0.000005363894,0.000037508154,0.000063358515,0.000026623453,0.00016209466,0.00543918,0.977371,0.016690645],"study_design_scores_gemma":[0.000013164781,0.000017648777,0.00042383187,0.0002299069,0.0000070860283,0.000055377983,0.000119735596,0.00005825777,0.000079846875,0.004386899,0.99459803,0.000010134318],"about_ca_topic_score_codex":0.004291974,"about_ca_topic_score_gemma":0.013715043,"teacher_disagreement_score":0.09609054,"about_ca_system_score_codex":0.0019885644,"about_ca_system_score_gemma":0.0026874845,"threshold_uncertainty_score":0.32145488},"labels":[],"label_agreement":null},{"id":"W2008481019","doi":"10.1145/1107499.1107509","title":"In memoriam Alberto Oscar Mendelzon","year":2005,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"TD Bank Group","funders":"","keywords":"Tribute; Battle; Friendship; Computer science; Intellect; Art history; Art; History; Ancient history; Theology; Philosophy; Sociology","score_opus":0.16446583863983466,"score_gpt":0.4286982494735129,"score_spread":0.26423241083367827,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2008481019","genre_codex":"editorial","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":[0.00011127402,0.011962084,0.00039366877,0.44122738,0.5401633,0.000023664148,0.00023227312,0.00014335364,0.0057429406],"genre_scores_gemma":[0.0050669652,0.023808626,0.00089467154,0.3562268,0.48585036,0.00012724573,0.000508217,0.0004095677,0.12710747],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9953525,0.00072128425,0.0003825006,0.00086840644,0.0023042068,0.00037103912],"domain_scores_gemma":[0.9713684,0.005537461,0.0012912549,0.00078493985,0.01570215,0.0053157853],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008293007,0.0014328088,0.002027283,0.0015544055,0.0029910097,0.009063325,0.002144347,0.0052607222,0.024049826],"category_scores_gemma":[0.047653064,0.00046634345,0.0010393764,0.00088528806,0.002015631,0.004406773,0.0030508616,0.017632872,0.026801705],"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":[0.000015150468,0.0000035082955,0.000028677423,0.000020747453,0.0000029280664,0.000034773653,0.000026204489,0.000009295274,0.000014580335,0.00039156197,0.99572796,0.0037245757],"study_design_scores_gemma":[0.00000923903,0.000009277906,0.00011495369,0.00013640664,0.000005633988,0.00014085186,0.00007106481,0.000028446493,0.000044414424,0.000722844,0.99870753,0.000009315178],"about_ca_topic_score_codex":0.005001734,"about_ca_topic_score_gemma":0.008374852,"teacher_disagreement_score":0.024049826,"about_ca_system_score_codex":0.003982915,"about_ca_system_score_gemma":0.005416252,"threshold_uncertainty_score":0.08045465},"labels":[],"label_agreement":null},{"id":"W2018758714","doi":"10.1145/945721.945733","title":"The hyperion project","year":2003,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Peer-to-Peer Network Technologies","field":"Computer Science","cited_by":189,"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; University of Toronto","funders":"","keywords":"Computer science; Schema (genetic algorithms); Database; Distributed database; Database schema; Key (lock); Information retrieval; Database design; Computer security","score_opus":0.026853933392331445,"score_gpt":0.2671210670957384,"score_spread":0.24026713370340697,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2018758714","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":[0.01905378,0.022088893,0.14257146,0.029434567,0.010126361,0.0010966925,0.008579178,0.027027408,0.7400216],"genre_scores_gemma":[0.085121274,0.016735863,0.14999476,0.008586554,0.003850811,0.001767662,0.03528403,0.007909112,0.69075],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99589217,0.0013243794,0.00015982325,0.0006665071,0.0013772417,0.00057983905],"domain_scores_gemma":[0.9963883,0.00057686906,0.00019862049,0.00094430236,0.00049590314,0.0013960317],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005429474,0.0010503443,0.0005375269,0.0019067426,0.0011630047,0.007612455,0.0021211223,0.0016451755,0.063551046],"category_scores_gemma":[0.0053494023,0.0003816552,0.0007265592,0.0017804904,0.001077129,0.00961628,0.008716254,0.0019543895,0.025651982],"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":[0.00064296334,0.00018159983,0.0016570551,0.00045867375,0.00005675601,0.00042980557,0.00083380914,0.0013112967,0.003306578,0.18916371,0.4108627,0.39109504],"study_design_scores_gemma":[0.000054172506,0.000048341408,0.00028825193,0.0001099497,0.000010810999,0.0001535462,0.00015548602,0.00092628354,0.001026543,0.013885331,0.9833278,0.000013496181],"about_ca_topic_score_codex":0.0016371625,"about_ca_topic_score_gemma":0.0010694335,"teacher_disagreement_score":0.063551046,"about_ca_system_score_codex":0.0012223509,"about_ca_system_score_gemma":0.0031067953,"threshold_uncertainty_score":0.2125994},"labels":[],"label_agreement":null},{"id":"W2020463925","doi":"10.1145/1462571.1462581","title":"Paper and proposal reviews","year":2008,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","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; Term (time); Data science; Panel discussion; World Wide Web; Engineering ethics; Engineering","score_opus":0.10000165082144481,"score_gpt":0.28754345874430165,"score_spread":0.18754180792285685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2020463925","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009501937,0.020259531,0.006005489,0.5099185,0.30501467,0.0044350037,0.0015850425,0.00064964086,0.15118201],"genre_scores_gemma":[0.01609847,0.035247408,0.014307805,0.3883508,0.13067165,0.009642795,0.0030005176,0.0013165225,0.40136403],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9049464,0.02261877,0.0075567835,0.0049892063,0.054612286,0.005276572],"domain_scores_gemma":[0.7547555,0.0747202,0.009786673,0.009868136,0.13793242,0.012937075],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.056014203,0.0012974661,0.0017002336,0.00765483,0.0052712183,0.017384468,0.0047682202,0.023666764,0.1009122],"category_scores_gemma":[0.2587115,0.0013413733,0.0032674586,0.005036149,0.0030335938,0.010066111,0.0059210327,0.017386943,0.05053826],"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":[0.000043773685,0.00003943949,0.000078411606,0.0008628782,0.000015327327,0.00015229477,0.00029630182,0.000077002645,0.0001791518,0.010336513,0.9629703,0.024948686],"study_design_scores_gemma":[0.000019658404,0.000025525993,0.0000719902,0.00066450465,0.000013534651,0.000045321067,0.00022141074,0.000032341293,0.00009008007,0.001738008,0.99706143,0.000016141934],"about_ca_topic_score_codex":0.0026466476,"about_ca_topic_score_gemma":0.0031932585,"teacher_disagreement_score":0.9439858,"about_ca_system_score_codex":0.009391104,"about_ca_system_score_gemma":0.04116469,"threshold_uncertainty_score":0.3375849},"labels":[],"label_agreement":null},{"id":"W2026730210","doi":"10.1145/362084.362104","title":"Generating spatiotemporal datasets on the WWW","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":54,"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":"Financiadora de Estudos e Projetos; Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Computer science; Search engine indexing; Data mining; Information retrieval; Interface (matter); Database; Operating system","score_opus":0.031612334944660654,"score_gpt":0.2537045766043308,"score_spread":0.22209224165967018,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2026730210","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":[0.04150257,0.00043108655,0.85910887,0.0009229874,0.00026050015,0.00035624334,0.02771948,0.046676893,0.023021324],"genre_scores_gemma":[0.19186608,0.001098403,0.7346586,0.00020002948,0.000099064644,0.0007178136,0.06126637,0.0033643504,0.0067293523],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993032,0.00019863433,0.00008426153,0.00013700625,0.00022922404,0.00004765219],"domain_scores_gemma":[0.99817336,0.000439575,0.00007602882,0.0008195263,0.00041471096,0.00007679887],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009122008,0.0005084283,0.0005684231,0.0018111953,0.0005293263,0.0025775074,0.0011905766,0.0006370652,0.006268343],"category_scores_gemma":[0.0054282253,0.00040619963,0.0008616031,0.0036592064,0.00030331616,0.003000368,0.0023785797,0.00078782806,0.0037873534],"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":[0.0008240543,0.00023917167,0.010262238,0.0009602007,0.0002328676,0.0015307292,0.0012142909,0.10613359,0.027557017,0.18086872,0.21019432,0.45998284],"study_design_scores_gemma":[0.00015732461,0.00012052958,0.0028913203,0.00013761339,0.00007794795,0.0007507475,0.0007431351,0.46694776,0.047075603,0.10954206,0.37147936,0.000076682634],"about_ca_topic_score_codex":0.0015674423,"about_ca_topic_score_gemma":0.002321025,"teacher_disagreement_score":0.006268343,"about_ca_system_score_codex":0.0004055877,"about_ca_system_score_gemma":0.00069451233,"threshold_uncertainty_score":0.020969748},"labels":[],"label_agreement":null},{"id":"W2033301070","doi":"10.1145/1107499.1107519","title":"Christos Faloutsos speaks out","year":2005,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":1,"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":"Very large database; Computer science; Library science; Database; Information retrieval","score_opus":0.02296878735220468,"score_gpt":0.24348310670995268,"score_spread":0.220514319357748,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2033301070","genre_codex":"commentary","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":[0.0029449023,0.032541122,0.0031207467,0.5449676,0.24737293,0.00010004863,0.0011134307,0.0021091131,0.16573012],"genre_scores_gemma":[0.022130687,0.017234566,0.002112233,0.11590288,0.041128237,0.000101677266,0.000623235,0.0011800925,0.7995864],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985083,0.0001905527,0.00005204912,0.00028033834,0.0006848115,0.00028402318],"domain_scores_gemma":[0.99474865,0.00062012405,0.00025761855,0.00014471063,0.0013182205,0.002910797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016954611,0.00090897933,0.0006588525,0.0012561608,0.004575419,0.004163679,0.0008529949,0.0022720085,0.1157567],"category_scores_gemma":[0.007347736,0.0003868321,0.00041008237,0.0008306701,0.0010708087,0.0034795979,0.0026969637,0.00466439,0.072646536],"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":[0.000012777234,0.000008029081,0.00011513435,0.000020969865,9.697651e-7,0.00006887291,0.00011642877,0.000008732866,0.0000918622,0.00092681375,0.98647064,0.0121588325],"study_design_scores_gemma":[0.0000022503486,0.0000072020084,0.000104978804,0.00003181881,9.657755e-7,0.0001496523,0.00044645168,0.00001506862,0.000063633466,0.00043472892,0.9987379,0.000005424305],"about_ca_topic_score_codex":0.0067136534,"about_ca_topic_score_gemma":0.012174453,"teacher_disagreement_score":0.1157567,"about_ca_system_score_codex":0.0013253405,"about_ca_system_score_gemma":0.002959927,"threshold_uncertainty_score":0.3872447},"labels":[],"label_agreement":null},{"id":"W2041835352","doi":"10.1145/945721.945722","title":"Analysis of SIGMOD's co-authorship graph","year":2003,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":138,"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":"Computer science; Graph; Information retrieval; Theoretical computer science","score_opus":0.02584356087544954,"score_gpt":0.30262111552378207,"score_spread":0.27677755464833254,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2041835352","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":[0.5664118,0.00672678,0.3609331,0.006836903,0.00046000333,0.0003268403,0.01585864,0.0019453829,0.04050052],"genre_scores_gemma":[0.92325944,0.0018403267,0.05960416,0.000265343,0.0002130341,0.00020117391,0.007917189,0.00022902175,0.00647014],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9974854,0.00058038963,0.000098019606,0.00035307795,0.0012020848,0.00028113302],"domain_scores_gemma":[0.98769635,0.0069992444,0.0013393211,0.001351241,0.002032941,0.0005808551],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0021134666,0.00057164603,0.00064324745,0.009685114,0.0018104177,0.0019532226,0.00097098714,0.00093357445,0.0043147267],"category_scores_gemma":[0.018679617,0.00022434299,0.00064394117,0.011934563,0.001051358,0.003575298,0.0012361882,0.00087683054,0.00082574243],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039332383,0.00039446427,0.05511972,0.0006199637,0.00041991606,0.00095618854,0.001643345,0.11490048,0.0039810413,0.49330622,0.05902479,0.26924056],"study_design_scores_gemma":[0.000052850683,0.000089663146,0.027627481,0.00011611299,0.00013195563,0.0010629212,0.0009004821,0.48179606,0.0049553025,0.42545336,0.057740055,0.00007377459],"about_ca_topic_score_codex":0.004146787,"about_ca_topic_score_gemma":0.0031927924,"teacher_disagreement_score":0.9903149,"about_ca_system_score_codex":0.0018542169,"about_ca_system_score_gemma":0.0010118141,"threshold_uncertainty_score":0.014434159},"labels":[],"label_agreement":null},{"id":"W2042832124","doi":"10.1145/1361348.1361361","title":"Report on the First International Workshop on Ranking in Databases (DBRank'07)","year":2007,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"University of Waterloo","funders":"","keywords":"Conjunction (astronomy); Ranking (information retrieval); Computer science; Database; World Wide Web; Information retrieval","score_opus":0.047223779236043414,"score_gpt":0.3079342729547288,"score_spread":0.2607104937186854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2042832124","genre_codex":"editorial","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":[0.014520988,0.08783008,0.26955542,0.10492078,0.2729503,0.0040705223,0.040776882,0.013563609,0.19181144],"genre_scores_gemma":[0.02419565,0.053649098,0.14657575,0.020539163,0.055474397,0.002247497,0.1228731,0.009033751,0.56541157],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9862146,0.004322113,0.0006615027,0.0016600188,0.005911164,0.0012305368],"domain_scores_gemma":[0.97160125,0.0051577585,0.0006521276,0.003245155,0.013159515,0.0061841584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019319301,0.0027524356,0.0034426257,0.004156297,0.0019223253,0.009959702,0.0029487312,0.0031269987,0.11849425],"category_scores_gemma":[0.022399683,0.0010010907,0.0021710857,0.0052142777,0.0010107644,0.010023492,0.006838918,0.007033009,0.08831723],"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":[0.00014490714,0.00019105447,0.00026467597,0.00020424117,0.000031274412,0.000052269952,0.00007996849,0.00036353196,0.00070271443,0.002142701,0.9227117,0.07311101],"study_design_scores_gemma":[0.00009291042,0.00021902741,0.0010443738,0.00024816403,0.0000611305,0.00015832079,0.0002128303,0.001632365,0.0016202003,0.0048833424,0.98976433,0.000063130385],"about_ca_topic_score_codex":0.005631926,"about_ca_topic_score_gemma":0.008499938,"teacher_disagreement_score":0.11849425,"about_ca_system_score_codex":0.0021317443,"about_ca_system_score_gemma":0.0048476933,"threshold_uncertainty_score":0.39640272},"labels":[],"label_agreement":null},{"id":"W2044936152","doi":"10.1145/2094114.2094118","title":"Parallel data processing with MapReduce","year":2012,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":555,"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; Abstraction; Data science; Big data; Volume (thermodynamics); Data processing; Simple (philosophy); Node (physics); Parallel processing; Database; Parallel computing; Data mining","score_opus":0.05152052286016848,"score_gpt":0.2733963681575956,"score_spread":0.22187584529742713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2044936152","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":[0.0060021593,0.0047699544,0.9628946,0.0012988654,0.000390638,0.00036850403,0.0007927109,0.007194525,0.016287982],"genre_scores_gemma":[0.17844827,0.010552333,0.7940496,0.0008674129,0.00060545653,0.00062805647,0.0034199473,0.0011143141,0.010314601],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977884,0.0003633494,0.00016771177,0.00041954344,0.0010801991,0.00018079177],"domain_scores_gemma":[0.9986228,0.00033281962,0.000057497593,0.00046798497,0.00044187176,0.00007700716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016113259,0.0009615316,0.0010308159,0.0010828329,0.001047368,0.002933257,0.0022214097,0.00058730756,0.003131885],"category_scores_gemma":[0.003664088,0.0006807301,0.0012674092,0.002734132,0.0006771546,0.0028192983,0.0024292227,0.0017289082,0.002841307],"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":[0.0004410865,0.00033563195,0.0020946271,0.001810659,0.00036828578,0.00039151064,0.0005992492,0.08520278,0.019336047,0.10845895,0.104332395,0.6766288],"study_design_scores_gemma":[0.00012693516,0.00024641101,0.0015355358,0.00020031544,0.00012652087,0.0009527014,0.00037109715,0.33572018,0.033007998,0.23526768,0.39231482,0.00012981183],"about_ca_topic_score_codex":0.002693097,"about_ca_topic_score_gemma":0.0025218332,"teacher_disagreement_score":0.003131885,"about_ca_system_score_codex":0.00078279845,"about_ca_system_score_gemma":0.0022339213,"threshold_uncertainty_score":0.010477245},"labels":[],"label_agreement":null},{"id":"W2046769817","doi":"10.1145/1147376.1147391","title":"Consistent query answering in databases","year":2006,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Logic, Reasoning, and Knowledge","field":"Computer Science","cited_by":201,"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; Database; Set (abstract data type); Query optimization; Query language; View; Sargable; Data integrity; Information retrieval; Distributed database; Web search query; Database design; Programming language; Search engine","score_opus":0.02901131692479943,"score_gpt":0.2536003828049169,"score_spread":0.22458906588011748,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2046769817","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":[0.012682657,0.0055321045,0.962248,0.010721473,0.00023965782,0.00028660972,0.0008170172,0.0012973374,0.006174994],"genre_scores_gemma":[0.25715604,0.0038828673,0.72466123,0.004543471,0.0011564209,0.0007447725,0.0034275893,0.00041797318,0.0040097674],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.94670576,0.026773699,0.0058387476,0.0066428995,0.01233647,0.0017025196],"domain_scores_gemma":[0.93134224,0.04979386,0.002462218,0.009696869,0.005890435,0.0008144685],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.034602795,0.00093269243,0.0030790353,0.0041263546,0.0029852933,0.011917599,0.0057066865,0.005320413,0.0038819364],"category_scores_gemma":[0.078813404,0.0019872796,0.0022904389,0.008777403,0.006738278,0.017948518,0.008526813,0.0051659322,0.0012137867],"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":[0.00040094947,0.00018530693,0.002248037,0.0011121226,0.00033121565,0.0008303344,0.0021237938,0.024487205,0.0022961013,0.81646216,0.01825614,0.13126665],"study_design_scores_gemma":[0.00009506322,0.000044362612,0.00018978884,0.000112356225,0.000089636735,0.0003660146,0.0003826328,0.063520245,0.0012568723,0.9201823,0.0137263145,0.00003439394],"about_ca_topic_score_codex":0.003173642,"about_ca_topic_score_gemma":0.001962613,"teacher_disagreement_score":0.034602795,"about_ca_system_score_codex":0.003002761,"about_ca_system_score_gemma":0.003352671,"threshold_uncertainty_score":0.18299925},"labels":[],"label_agreement":null},{"id":"W2064853889","doi":"10.1145/335191.335372","title":"Mining frequent patterns without candidate generation","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":6360,"is_retracted":false,"has_abstract":true,"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; Data mining; Scalability; Set (abstract data type); Tree (set theory); GSP Algorithm; Trie; Apriori algorithm; Association rule learning; Database transaction; Data structure; Database; Mathematics","score_opus":0.02961243143666664,"score_gpt":0.2688342363591712,"score_spread":0.23922180492250456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064853889","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":[0.092139326,0.0011316038,0.8995709,0.00043904272,0.00012886345,0.000554885,0.0017245291,0.0018245466,0.0024863668],"genre_scores_gemma":[0.31372544,0.00072446873,0.6758033,0.0002281819,0.00015492039,0.0006648764,0.0061734836,0.00013046637,0.0023949298],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961926,0.00071675074,0.00043328703,0.0008190486,0.0015685359,0.00026971198],"domain_scores_gemma":[0.9865072,0.007654811,0.0010216712,0.0023200556,0.0022599164,0.00023628607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002773499,0.0010560468,0.0015197528,0.0033173126,0.0010026896,0.0020651596,0.0023299793,0.0013650684,0.0018598416],"category_scores_gemma":[0.020350711,0.0007856847,0.001484656,0.0049026296,0.0007125326,0.0044252025,0.0016331488,0.0010364935,0.0016855483],"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":[0.0010564004,0.00046187502,0.018379858,0.00074728165,0.00032900326,0.0018769072,0.00041830604,0.060169913,0.019364689,0.02295568,0.009558297,0.86468184],"study_design_scores_gemma":[0.0001956015,0.00081906974,0.003836875,0.0001240209,0.00024286074,0.002747715,0.00032901953,0.9142375,0.020698236,0.044000197,0.012703483,0.00006536376],"about_ca_topic_score_codex":0.0011205769,"about_ca_topic_score_gemma":0.0013202778,"teacher_disagreement_score":0.0033173126,"about_ca_system_score_codex":0.00040084432,"about_ca_system_score_gemma":0.0017851382,"threshold_uncertainty_score":0.014667809},"labels":[],"label_agreement":null},{"id":"W2066407644","doi":"10.1145/1041410.1041416","title":"Kanata","year":2004,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Semantic Web and Ontologies","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":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Metadata; Focus (optics); World Wide Web; Database; Software engineering","score_opus":0.02648632253714379,"score_gpt":0.25709268722599943,"score_spread":0.23060636468885565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2066407644","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":[0.010112878,0.0105389515,0.03771711,0.020389546,0.0062189065,0.00050251,0.013682659,0.017416028,0.8834215],"genre_scores_gemma":[0.031197729,0.0036236215,0.017969895,0.0020305265,0.00029680145,0.00018234854,0.0081058135,0.002522501,0.9340708],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976,0.00025363674,0.000074182564,0.00053578865,0.0011340476,0.0004022826],"domain_scores_gemma":[0.99551153,0.00028657427,0.00010782188,0.0006978338,0.0018042111,0.0015920459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002257729,0.00088274846,0.00088074896,0.0020544874,0.0055448036,0.00907957,0.0015177765,0.0019012585,0.1817472],"category_scores_gemma":[0.004455176,0.0008199981,0.00070989324,0.0031996379,0.0018802582,0.0035964001,0.005626833,0.002059986,0.113291286],"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":[0.0003583202,0.00005854254,0.0025659606,0.0004562815,0.000021840313,0.00039645447,0.002621094,0.00039001595,0.0031258578,0.07154197,0.72418755,0.19427603],"study_design_scores_gemma":[0.00001369405,0.000010764022,0.0005036806,0.000037958253,0.00000867451,0.00006356296,0.00032326794,0.00015802341,0.0005434742,0.0020070535,0.99631447,0.000015358184],"about_ca_topic_score_codex":0.20290878,"about_ca_topic_score_gemma":0.3954291,"teacher_disagreement_score":0.20290878,"about_ca_system_score_codex":0.011326221,"about_ca_system_score_gemma":0.016166715,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2070556972","doi":"10.1145/2380776.2380784","title":"dbTrento","year":2012,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":3,"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":"Universidad de Zaragoza; University of Alberta","keywords":"Citation; Computer science; Library science; Database; World Wide Web","score_opus":0.03247796109390363,"score_gpt":0.278870893944344,"score_spread":0.24639293285044037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2070556972","genre_codex":"other","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":[0.00052484235,0.0020318246,0.0058511994,0.0032113905,0.0033210516,0.00045835142,0.20071699,0.10022576,0.68365854],"genre_scores_gemma":[0.0038316841,0.0022600738,0.0047379863,0.0035764915,0.0012742758,0.00038106932,0.2548102,0.05486334,0.6742649],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976561,0.00022082518,0.00017404988,0.00040309635,0.0010806946,0.000465279],"domain_scores_gemma":[0.99196106,0.00071701745,0.0003232276,0.00258943,0.0025456694,0.0018635709],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019095804,0.0025373972,0.0025980636,0.00648763,0.0024408102,0.016589466,0.004804303,0.0028369494,0.85040903],"category_scores_gemma":[0.009651038,0.0017261608,0.0011980514,0.009366105,0.0007282617,0.008430945,0.007961803,0.0040570092,0.8815767],"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":[0.000032825035,0.000017824515,0.0000661893,0.00015545137,0.000004893867,0.000012085114,0.000025156429,0.000016190714,0.0001398032,0.00083977974,0.98037016,0.018319644],"study_design_scores_gemma":[0.000009796448,0.0000036397194,0.0001394407,0.000025164241,0.0000018786573,0.00002562107,0.000015590109,0.000026288657,0.00013146635,0.0002991831,0.9993131,0.000008644205],"about_ca_topic_score_codex":0.0080660535,"about_ca_topic_score_gemma":0.009739088,"teacher_disagreement_score":0.85040903,"about_ca_system_score_codex":0.002642148,"about_ca_system_score_gemma":0.004401224,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2078181649","doi":"10.1145/507338.507366","title":"Report on the 18 <sup>th</sup> British National Conference on Databases (BNCOD)","year":2002,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":0,"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":"Terabyte; Library science; Session (web analytics); Presentation (obstetrics); Database; Service (business); Computer science; World Wide Web; Medicine","score_opus":0.14909883946530322,"score_gpt":0.32737886912988584,"score_spread":0.17828002966458262,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2078181649","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":[0.0038918876,0.05075631,0.024479823,0.09539157,0.34237084,0.0046524433,0.12231336,0.0068434523,0.34930038],"genre_scores_gemma":[0.006438048,0.022402754,0.0113988565,0.017157432,0.03003247,0.0032790045,0.13268723,0.0039255624,0.7726786],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9902969,0.00123358,0.00067248073,0.0010294815,0.00587661,0.00089099805],"domain_scores_gemma":[0.96756,0.0039855796,0.0012414518,0.0023450002,0.017881133,0.0069867736],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.01166213,0.0018068736,0.0027139205,0.0050614127,0.0024780356,0.013719771,0.0034305037,0.0035964362,0.3755006],"category_scores_gemma":[0.018959217,0.0009186246,0.0014312433,0.007232178,0.0010179674,0.008162072,0.006254745,0.0053976867,0.2803801],"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":[0.000042486776,0.000018363446,0.00012437404,0.000110483175,0.0000042194256,0.000013831611,0.000010591016,0.000028374454,0.0001034136,0.00036636108,0.98521686,0.013960671],"study_design_scores_gemma":[0.000020076777,0.000024362502,0.00072797504,0.000109563174,0.000006457526,0.000029197083,0.000049644175,0.00011938891,0.0001641593,0.0002673383,0.9984713,0.000010577478],"about_ca_topic_score_codex":0.0116819,"about_ca_topic_score_gemma":0.012360141,"teacher_disagreement_score":0.3755006,"about_ca_system_score_codex":0.0038208046,"about_ca_system_score_gemma":0.0065795346,"threshold_uncertainty_score":0.8907726},"labels":[],"label_agreement":null},{"id":"W2079337221","doi":"10.1145/603867.603886","title":"Career-enhancing services at SIGMOD online","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"University of Toronto","funders":"","keywords":"Computer science; Database; Architecture","score_opus":0.15690084962122752,"score_gpt":0.37431087168082067,"score_spread":0.21741002205959314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2079337221","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":[0.0030678108,0.00838882,0.022221027,0.03531689,0.019048497,0.0011114022,0.0754104,0.08696255,0.7484726],"genre_scores_gemma":[0.007512614,0.006973403,0.022106368,0.005929462,0.010755735,0.0005795056,0.0451973,0.008402266,0.8925434],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967534,0.000696944,0.0002829071,0.000316631,0.0016252382,0.00032485207],"domain_scores_gemma":[0.9859979,0.001678534,0.0008839105,0.0028760252,0.0027629568,0.005800606],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.007996119,0.002017459,0.0016627238,0.010802147,0.0023929554,0.008016187,0.0018969426,0.002035746,0.6590756],"category_scores_gemma":[0.018733457,0.0010674018,0.0007631988,0.015521737,0.0005223231,0.0061752894,0.006568452,0.0028721003,0.65537953],"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":[0.00002206012,0.000042113857,0.00019392635,0.000063119965,0.000002908637,0.00001621126,0.000036133923,0.000019267749,0.00014545204,0.0011328644,0.95176893,0.04655702],"study_design_scores_gemma":[0.000019000336,0.000014039581,0.00032320555,0.000036655936,0.0000020892865,0.000045606568,0.00003661461,0.000116624164,0.00009343468,0.00084023556,0.99846375,0.000008744969],"about_ca_topic_score_codex":0.0012477344,"about_ca_topic_score_gemma":0.002227352,"teacher_disagreement_score":0.6590756,"about_ca_system_score_codex":0.0011177836,"about_ca_system_score_gemma":0.003257061,"threshold_uncertainty_score":0.48628724},"labels":[],"label_agreement":null},{"id":"W2083283415","doi":"10.1145/2694413.2694425","title":"Report on the First International Workshop on Exploratory Search in Databases and the Web (ExploreDB 2014)","year":2014,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management 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 British Columbia","funders":"","keywords":"Computer science; Database; World Wide Web; Information retrieval","score_opus":0.0560095100511173,"score_gpt":0.2822348339987544,"score_spread":0.22622532394763706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2083283415","genre_codex":"review","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017980199,0.7049705,0.011725719,0.076226495,0.066078015,0.0020947857,0.010539444,0.001112753,0.12545428],"genre_scores_gemma":[0.009824626,0.58508366,0.015273345,0.035210792,0.020363266,0.0019971954,0.024315396,0.0015188227,0.30641288],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9944311,0.0013343828,0.00049240695,0.0004898118,0.0027217893,0.00053050957],"domain_scores_gemma":[0.9750748,0.005003347,0.0010656047,0.0014913165,0.014398841,0.0029660068],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010622045,0.001090392,0.0017410758,0.008596987,0.0008926755,0.0057589603,0.0018413048,0.002550642,0.13861053],"category_scores_gemma":[0.02031806,0.0006455397,0.0021253566,0.008365041,0.0006472298,0.0073048137,0.004978236,0.002651387,0.07325722],"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":[0.000074877804,0.00004795553,0.00020793857,0.0030618848,0.000055246717,0.00004532335,0.000093126786,0.00007052927,0.0006141204,0.0014630146,0.83258975,0.16167624],"study_design_scores_gemma":[0.000014830706,0.000031836575,0.00044085656,0.0018238794,0.000048908554,0.00005082771,0.0000819902,0.000026669139,0.0002189796,0.0006914757,0.9965593,0.000010413591],"about_ca_topic_score_codex":0.0060100984,"about_ca_topic_score_gemma":0.011382837,"teacher_disagreement_score":0.13861053,"about_ca_system_score_codex":0.00308098,"about_ca_system_score_gemma":0.009591543,"threshold_uncertainty_score":0.4636984},"labels":[],"label_agreement":null},{"id":"W2094817912","doi":"10.1145/1084805.1084825","title":"Exchange, integration, and consistency of data","year":2005,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":true,"ca_institutions":"IBM (Canada); University of Victoria; York University; Carleton University","funders":"","keywords":"Computer science; IBM; Data exchange; Consistency (knowledge bases); Data consistency; Data integration; Database; Operating system; Programming language","score_opus":0.057434164822293146,"score_gpt":0.30152373536708393,"score_spread":0.2440895705447908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2094817912","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":[0.022291671,0.003637989,0.95303833,0.0064801956,0.00054905465,0.0004606436,0.00069936836,0.002432297,0.010410481],"genre_scores_gemma":[0.29433817,0.003682106,0.68176824,0.0021059362,0.00090121146,0.00093247415,0.0047614407,0.0019388427,0.009571611],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9025375,0.03998942,0.012523609,0.011771278,0.028909812,0.00426837],"domain_scores_gemma":[0.80970275,0.043451875,0.010209693,0.11412131,0.020086598,0.0024277947],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.07596805,0.0010237952,0.0025939397,0.0056424704,0.00474472,0.025118962,0.0076039303,0.0045179715,0.0029262714],"category_scores_gemma":[0.20889042,0.0031258424,0.0021165975,0.009057824,0.009996237,0.035976518,0.019499104,0.006687119,0.0028412626],"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":[0.0005237047,0.00027824295,0.013101656,0.00057243224,0.0003010116,0.0005579981,0.0054144617,0.012786256,0.0041040266,0.7287231,0.018054422,0.2155827],"study_design_scores_gemma":[0.00017890455,0.0001910155,0.002651162,0.0005086768,0.00033320874,0.0010372677,0.0013897496,0.042735014,0.012291938,0.8173557,0.12114097,0.00018643712],"about_ca_topic_score_codex":0.005371878,"about_ca_topic_score_gemma":0.0023625395,"teacher_disagreement_score":0.07596805,"about_ca_system_score_codex":0.0027142975,"about_ca_system_score_gemma":0.00981249,"threshold_uncertainty_score":0.40176225},"labels":[],"label_agreement":null},{"id":"W2098416578","doi":"10.1145/335191.335419","title":"Efficient and extensible algorithms for multi query optimization","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":409,"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; Query optimization; Heuristics; Benchmark (surveying); Overhead (engineering); Greedy algorithm; Heuristic; Algorithm; Data mining; Artificial intelligence","score_opus":0.04290867931345283,"score_gpt":0.2807725920124819,"score_spread":0.23786391269902907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098416578","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":[0.004667192,0.00056318997,0.9861907,0.00017247215,0.000051220395,0.00020325306,0.00017112051,0.0055048056,0.0024759239],"genre_scores_gemma":[0.07820176,0.00054705853,0.9176114,0.00017988861,0.00009716884,0.00048460503,0.0006516078,0.00065620756,0.0015702947],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99578613,0.0009963744,0.0005028958,0.0007386413,0.0015119823,0.00046399987],"domain_scores_gemma":[0.9935731,0.0035087087,0.00044126844,0.0016773692,0.0006742534,0.00012519065],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003566887,0.0029344202,0.0017107454,0.0025227533,0.0010040456,0.0026525422,0.0045367344,0.001996858,0.0048064147],"category_scores_gemma":[0.011892577,0.0011619609,0.0017486123,0.0042608213,0.0012537069,0.0043402384,0.0033920587,0.0031181625,0.0023914773],"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":[0.000404842,0.00047695445,0.0016505545,0.00036258128,0.00013017678,0.00014419055,0.00023688046,0.3925352,0.006174304,0.038432084,0.014707178,0.54474515],"study_design_scores_gemma":[0.00013008706,0.00007580968,0.00021414887,0.000024916668,0.000037327627,0.00010265397,0.00006868371,0.9586068,0.0028698472,0.032729834,0.0051128343,0.00002704596],"about_ca_topic_score_codex":0.0053370357,"about_ca_topic_score_gemma":0.0062771523,"teacher_disagreement_score":0.0053370357,"about_ca_system_score_codex":0.0017277885,"about_ca_system_score_gemma":0.0023004024,"threshold_uncertainty_score":0.018863738},"labels":[],"label_agreement":null},{"id":"W2099464320","doi":"10.1145/1107499.1107507","title":"An apples-to-apples comparison of two database journals","year":2005,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Mining Algorithms and Applications","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; Database; Information retrieval; World Wide Web","score_opus":0.06868933506628278,"score_gpt":0.39365361095319384,"score_spread":0.32496427588691107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2099464320","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":[0.67279464,0.027124783,0.058298137,0.009973531,0.013653671,0.0013081328,0.020843316,0.004853982,0.19114977],"genre_scores_gemma":[0.91211396,0.0036782941,0.055460297,0.0013255904,0.002686883,0.0005346321,0.010270983,0.0017379815,0.0121913785],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9529158,0.012804044,0.0061562154,0.001996337,0.024829546,0.0012980036],"domain_scores_gemma":[0.67890537,0.1307683,0.043596826,0.01604848,0.11973731,0.010943646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.022460813,0.0006125753,0.0013539993,0.047731884,0.00185725,0.00978656,0.0008375396,0.0009793045,0.009220402],"category_scores_gemma":[0.220461,0.00027878265,0.0010703865,0.030052518,0.001700874,0.009164975,0.0043409416,0.0012074054,0.003016908],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003764472,0.00051130744,0.19375613,0.0026380967,0.0014792972,0.00046384474,0.005504593,0.0020598196,0.015010394,0.041554447,0.08337661,0.64988106],"study_design_scores_gemma":[0.0003371421,0.0043943673,0.53362155,0.0010764579,0.0012387297,0.0021668642,0.01564019,0.010493718,0.015741026,0.035976898,0.37875256,0.0005604422],"about_ca_topic_score_codex":0.0014830522,"about_ca_topic_score_gemma":0.0034039288,"teacher_disagreement_score":0.047731884,"about_ca_system_score_codex":0.0027615854,"about_ca_system_score_gemma":0.0016659446,"threshold_uncertainty_score":0.11878556},"labels":[],"label_agreement":null},{"id":"W2116493296","doi":"10.1145/335191.335409","title":"XTRACT","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":202,"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; XML validation; Document Structure Description; Document type definition; RELAX NG; XML Schema Editor; Well-formed document; XML Schema (W3C); Streaming XML; XML; Information retrieval; Programming language; Simple API for XML; Efficient XML Interchange; XML database; XML Signature; World Wide Web","score_opus":0.010340759165822527,"score_gpt":0.24748946711200087,"score_spread":0.23714870794617834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2116493296","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":[0.0038420162,0.0016874342,0.46492332,0.0021098177,0.0015109935,0.00076432247,0.047426786,0.39055887,0.0871764],"genre_scores_gemma":[0.06455163,0.0032887494,0.45505536,0.002851344,0.00090719457,0.0014040242,0.29233906,0.06987658,0.10972606],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99645114,0.0005564816,0.00033741887,0.00078689546,0.0015994415,0.0002685711],"domain_scores_gemma":[0.9952721,0.0013324555,0.000336269,0.0016800149,0.0011418713,0.00023739513],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030113338,0.0016049226,0.0011287858,0.0026478022,0.0013350561,0.0059946673,0.0044262614,0.002401274,0.09221158],"category_scores_gemma":[0.015199302,0.0011082339,0.0021465293,0.0031899498,0.0009739241,0.006280416,0.0035056586,0.0028912798,0.08255686],"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":[0.0006169067,0.000097695236,0.0015964333,0.00078722124,0.00011401977,0.0005178604,0.0002718069,0.0067927954,0.0045273965,0.05063822,0.6613179,0.27272174],"study_design_scores_gemma":[0.00010315516,0.00009399517,0.0006987663,0.00014519184,0.00004608314,0.000500553,0.000059825365,0.04813132,0.0076695993,0.040860407,0.90160626,0.00008482553],"about_ca_topic_score_codex":0.005217343,"about_ca_topic_score_gemma":0.0040334105,"teacher_disagreement_score":0.09221158,"about_ca_system_score_codex":0.0015295597,"about_ca_system_score_gemma":0.0025373467,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2123331342","doi":"10.1145/335191.336589","title":"DISIMA","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"","field":"","cited_by":9,"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":"Computer science; Programming language; Database","score_opus":0.01982543029966458,"score_gpt":0.2650928007484224,"score_spread":0.24526737044875782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123331342","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":[0.0011913407,0.0036211575,0.0015651801,0.0054312674,0.00641447,0.00029406088,0.027816178,0.005872194,0.9477943],"genre_scores_gemma":[0.005453289,0.0019657891,0.0005267356,0.001079447,0.0006070344,0.000109130546,0.01633025,0.0008115189,0.9731169],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994716,0.000063071464,0.000026190346,0.00010923181,0.00021736935,0.00011257567],"domain_scores_gemma":[0.99859303,0.00015434268,0.00004386735,0.00025526775,0.00039443848,0.0005590639],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00068191095,0.0011298998,0.0010390754,0.0017339655,0.00096419017,0.0037265578,0.0014831404,0.0014431064,0.8948823],"category_scores_gemma":[0.002974283,0.000551713,0.0005129945,0.0014006243,0.00054322754,0.0018321694,0.0021650738,0.0018199128,0.8786451],"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":[0.00032203647,0.000080474834,0.00014059778,0.00031343388,0.000009457548,0.00007429005,0.00002506908,0.00003430863,0.0014589847,0.0013340976,0.9074214,0.08878585],"study_design_scores_gemma":[0.00004110071,0.00004368752,0.00030534674,0.000061627085,0.0000065208133,0.00007208338,0.000023792736,0.000041624116,0.00049391104,0.00037797313,0.998528,0.0000043778045],"about_ca_topic_score_codex":0.0023242503,"about_ca_topic_score_gemma":0.003513209,"teacher_disagreement_score":0.10511768,"about_ca_system_score_codex":0.00068584975,"about_ca_system_score_gemma":0.0011175723,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2126321570","doi":"10.1145/335191.335393","title":"How to roll a join","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Distributed systems and fault tolerance","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; Materialized view; Asynchronous communication; Process (computing); Join (topology); Point (geometry); Refresh rate; Distributed computing; Database; Operating system; Computer hardware; Computer network; View","score_opus":0.016190014463895588,"score_gpt":0.23275630997701277,"score_spread":0.21656629551311718,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126321570","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":[0.034586802,0.0013619134,0.83250564,0.004628137,0.0022782194,0.00094101205,0.0016696589,0.034631457,0.0873972],"genre_scores_gemma":[0.3640887,0.0011815768,0.5752281,0.0012275785,0.0006115616,0.0003539118,0.0027717808,0.0035442014,0.050992496],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99860483,0.00018804867,0.00008811134,0.0002767307,0.0006453299,0.00019695793],"domain_scores_gemma":[0.9985393,0.00033161358,0.000071684866,0.00059807196,0.00027657876,0.00018264237],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009820093,0.00068324804,0.0005416619,0.00063633005,0.0012999412,0.003170004,0.0010056937,0.0011382982,0.025873285],"category_scores_gemma":[0.0044496004,0.0005543829,0.00082807365,0.00044503532,0.00079073093,0.0033457454,0.001491314,0.001591917,0.00986842],"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":[0.0010022062,0.00033353092,0.0029285757,0.0003664983,0.00009890034,0.000743416,0.0007332409,0.019931836,0.031271823,0.0749215,0.12163683,0.74603164],"study_design_scores_gemma":[0.00062583195,0.00060668506,0.0041811853,0.00028666845,0.00022532367,0.0013957996,0.0017264613,0.28346097,0.057276,0.18650545,0.46338373,0.00032581773],"about_ca_topic_score_codex":0.0040008556,"about_ca_topic_score_gemma":0.0038816913,"teacher_disagreement_score":0.025873285,"about_ca_system_score_codex":0.00036482426,"about_ca_system_score_gemma":0.00087202183,"threshold_uncertainty_score":0.086554706},"labels":[],"label_agreement":null},{"id":"W2128114397","doi":"10.1145/1228268.1228284","title":"Normalization theory for XML","year":2006,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":50,"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":"University of Toronto","keywords":"Computer science; Relational database; Normalization (sociology); XML; XML database; Database; Relational model; Information retrieval; Simple (philosophy); World Wide Web","score_opus":0.012011270581554807,"score_gpt":0.24309843180001883,"score_spread":0.23108716121846404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128114397","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":[0.004196031,0.0068479124,0.8306817,0.010184603,0.0014848216,0.00014470676,0.0013463973,0.0015374551,0.14357647],"genre_scores_gemma":[0.29805434,0.017861504,0.5152355,0.012074397,0.007858055,0.0018739449,0.0069281673,0.0023703303,0.13774389],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9950105,0.0015706728,0.00043893288,0.0010328336,0.0016258053,0.00032126985],"domain_scores_gemma":[0.99513704,0.002109293,0.0003023488,0.001259502,0.001034198,0.00015761705],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056144963,0.00093564036,0.0012092239,0.0044302344,0.0037166306,0.006433757,0.0025961476,0.002470553,0.01733558],"category_scores_gemma":[0.010211608,0.0009094419,0.002106896,0.005285158,0.0059106904,0.01862656,0.004168019,0.006128062,0.0069993134],"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":[0.0000033707793,0.0000046018204,0.0000413515,0.000022540322,0.000004103442,0.000028115359,0.00008019608,0.0002545024,0.00007751093,0.9917413,0.0029787724,0.004763614],"study_design_scores_gemma":[0.000003456095,0.000003403727,0.000051711148,0.000026484819,0.0000053600616,0.00006782719,0.000035818375,0.0018089339,0.00017360035,0.9624922,0.035322968,0.000008149869],"about_ca_topic_score_codex":0.0046069105,"about_ca_topic_score_gemma":0.0027302643,"teacher_disagreement_score":0.01733558,"about_ca_system_score_codex":0.0043192585,"about_ca_system_score_gemma":0.0022654997,"threshold_uncertainty_score":0.057993352},"labels":[],"label_agreement":null},{"id":"W2132851185","doi":"10.1145/776985.776986","title":"Issues in data stream management","year":2003,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":916,"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; Timestamp; Data stream mining; Database; Query language; Streaming data; Data stream; Data management; Data science; Data mining; Computer security","score_opus":0.04335612978330646,"score_gpt":0.30557491160712486,"score_spread":0.2622187818238184,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132851185","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":[0.00808037,0.061158802,0.6739842,0.21418989,0.008426955,0.00035920372,0.00052719977,0.0015942327,0.03167909],"genre_scores_gemma":[0.31275308,0.11741233,0.45961818,0.035003345,0.045564663,0.0012661286,0.0012540765,0.001033601,0.026094599],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9835491,0.005442186,0.0017780894,0.0016845855,0.00700507,0.00054094393],"domain_scores_gemma":[0.9556658,0.028759127,0.0017510623,0.0044588367,0.008038112,0.0013270503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.030363197,0.0007013565,0.00130674,0.0018749357,0.002013788,0.014773264,0.0049120625,0.0051175444,0.0038195888],"category_scores_gemma":[0.074914835,0.0008204041,0.0012784171,0.0054957964,0.0064228084,0.029337067,0.0044361716,0.007979317,0.0020194289],"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":[0.00009991324,0.00007210702,0.0012182039,0.0005896482,0.000048880444,0.0001879617,0.000920778,0.0060246177,0.0005887382,0.78515166,0.042378217,0.16271941],"study_design_scores_gemma":[0.00005056511,0.0001106434,0.00040772994,0.00048958865,0.000038000024,0.000786826,0.0010877388,0.040999442,0.0010118416,0.736633,0.2183205,0.00006414104],"about_ca_topic_score_codex":0.0022975334,"about_ca_topic_score_gemma":0.0012642677,"teacher_disagreement_score":0.030363197,"about_ca_system_score_codex":0.002601151,"about_ca_system_score_gemma":0.0033452064,"threshold_uncertainty_score":0.1605779},"labels":[],"label_agreement":null},{"id":"W2139422684","doi":"10.1145/335191.336572","title":"Towards data mining benchmarking","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Mining Algorithms and Applications","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; Data mining; Benchmarking; Data stream mining; Database transaction; Association rule learning; Apriori algorithm; Set (abstract data type); GSP Algorithm; Database; Relational database","score_opus":0.05942132737752803,"score_gpt":0.2999806763587709,"score_spread":0.24055934898124287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139422684","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0127102975,0.008093272,0.904648,0.02819348,0.0026669954,0.0014120243,0.0023988665,0.0091613205,0.030715713],"genre_scores_gemma":[0.1305668,0.0060761445,0.8335759,0.0072931326,0.0013268857,0.002414981,0.011137007,0.0022373828,0.0053717145],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9033338,0.048775017,0.008951464,0.009044292,0.027283413,0.0026119724],"domain_scores_gemma":[0.8826397,0.038210843,0.0038993736,0.04006594,0.031718552,0.003465533],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.08084901,0.002452552,0.0036637764,0.006299605,0.0021011953,0.017717518,0.008426043,0.004187661,0.0061568916],"category_scores_gemma":[0.19353804,0.001373507,0.0022828782,0.011315529,0.0036101767,0.023646332,0.012806977,0.010353582,0.006223383],"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":[0.000494394,0.0009710113,0.009077013,0.001994987,0.00033869475,0.00023880045,0.0010002431,0.045109514,0.0041297763,0.3878365,0.08371024,0.4650989],"study_design_scores_gemma":[0.00016906341,0.0007037732,0.0026139244,0.0017192086,0.000113422706,0.0004603865,0.0014112898,0.22657391,0.0117822215,0.4658794,0.28840753,0.00016582545],"about_ca_topic_score_codex":0.0015298664,"about_ca_topic_score_gemma":0.00087495486,"teacher_disagreement_score":0.919151,"about_ca_system_score_codex":0.0047153383,"about_ca_system_score_gemma":0.0074580163,"threshold_uncertainty_score":0.42757553},"labels":[],"label_agreement":null},{"id":"W2144182447","doi":"10.1145/335191.335388","title":"LOF","year":2000,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":5181,"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":"Outlier; Local outlier factor; Computer science; Object (grammar); Anomaly detection; Degree (music); Data mining; Property (philosophy); Binary number; Theoretical computer science; Artificial intelligence; Mathematics","score_opus":0.012959734804963171,"score_gpt":0.24427321013530048,"score_spread":0.2313134753303373,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144182447","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":[0.0047607697,0.0015721257,0.9304909,0.0011951259,0.0007168557,0.00038163888,0.0047446582,0.019237107,0.03690088],"genre_scores_gemma":[0.14478856,0.0021222695,0.735836,0.0024399953,0.001199064,0.0011460681,0.02564131,0.004455762,0.0823709],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971035,0.00033823666,0.00020856546,0.0006498533,0.001344405,0.0003553829],"domain_scores_gemma":[0.9959002,0.00110259,0.00029576087,0.0013331191,0.0012240745,0.00014421274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026942564,0.0014255403,0.0011877273,0.004829217,0.0018137947,0.0036786469,0.0026734625,0.0024944262,0.07034544],"category_scores_gemma":[0.013886158,0.00047847716,0.0016199896,0.0041016955,0.0011141766,0.005126272,0.0027148137,0.0021741353,0.050086793],"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":[0.00035662903,0.00020253408,0.003929155,0.00041798662,0.00008228906,0.00031970075,0.00021487389,0.011751341,0.0048197773,0.06885535,0.13557279,0.77347755],"study_design_scores_gemma":[0.00012508048,0.00029257004,0.0031781548,0.00023467088,0.00008453245,0.0020246753,0.00037710564,0.20038846,0.013505581,0.15270224,0.62691957,0.00016723879],"about_ca_topic_score_codex":0.0052188826,"about_ca_topic_score_gemma":0.0065624053,"teacher_disagreement_score":0.07034544,"about_ca_system_score_codex":0.0012002616,"about_ca_system_score_gemma":0.0017717363,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W2161694107","doi":"10.1145/2034863.2034873","title":"Repeatability and workability evaluation of SIGMOD 2011","year":2011,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Scientific Computing and Data Management","field":"Decision Sciences","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":"Repeatability; Executable; Computer science; Process (computing); Statistics; Operating system; Mathematics","score_opus":0.5113276891992731,"score_gpt":0.42903251205096915,"score_spread":0.08229517714830392,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2161694107","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8342363,0.006395853,0.067668565,0.006827115,0.0038862466,0.004973188,0.01730976,0.018419012,0.040284004],"genre_scores_gemma":[0.9151257,0.0006348221,0.048169516,0.00052144844,0.0009424173,0.0032997755,0.022781527,0.002143001,0.0063817096],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.7594117,0.10751649,0.025623249,0.01595291,0.086711735,0.004784016],"domain_scores_gemma":[0.43158448,0.23245682,0.041688155,0.124103546,0.15959546,0.010571545],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.1583614,0.0011548867,0.0015055714,0.012224635,0.0032150794,0.0073471833,0.0027218186,0.0018634099,0.0029385653],"category_scores_gemma":[0.44759086,0.00066537596,0.0016529299,0.008223243,0.0020483804,0.004864079,0.006249575,0.001836283,0.003149112],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006542405,0.003093462,0.3048047,0.003570271,0.00217541,0.00043965367,0.010785635,0.010049593,0.01430078,0.009156674,0.1304938,0.5045876],"study_design_scores_gemma":[0.0010732103,0.0072075673,0.668603,0.00081102847,0.00094088394,0.0014129432,0.0058715977,0.052933257,0.043465257,0.017777795,0.1989438,0.0009597395],"about_ca_topic_score_codex":0.002596654,"about_ca_topic_score_gemma":0.0025492616,"teacher_disagreement_score":0.84163857,"about_ca_system_score_codex":0.0026070236,"about_ca_system_score_gemma":0.0036168792,"threshold_uncertainty_score":0.83750516},"labels":[],"label_agreement":null},{"id":"W2162294668","doi":"10.1145/373626.373713","title":"The Clio project","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":343,"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; Set (abstract data type); Data integration; Transformation (genetics); Data science; Software engineering; Programming language; Database","score_opus":0.031767974344315664,"score_gpt":0.29049034626259007,"score_spread":0.2587223719182744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2162294668","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":[0.025234452,0.017048687,0.33040324,0.014100209,0.0029549962,0.002244452,0.035516497,0.13701755,0.43547997],"genre_scores_gemma":[0.101530135,0.01094154,0.36165553,0.006476815,0.0011302202,0.00350873,0.1967767,0.04293669,0.27504367],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9921158,0.0022171233,0.00037406076,0.0011825673,0.0030408183,0.0010695338],"domain_scores_gemma":[0.9905483,0.0019113199,0.0005395493,0.002748453,0.0022532272,0.0019990997],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010678826,0.0023819946,0.0010695506,0.00590226,0.0024827253,0.008610901,0.0044870893,0.003244307,0.06462744],"category_scores_gemma":[0.017023856,0.0008432479,0.0014452093,0.004840857,0.0019867085,0.009700947,0.012852394,0.0034774179,0.039841276],"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":[0.00094317005,0.0003191301,0.0024004881,0.0010526369,0.00010099943,0.00045070102,0.0015234882,0.0020249446,0.0052158725,0.11205075,0.5481361,0.32578176],"study_design_scores_gemma":[0.00015807367,0.0000936694,0.00057687465,0.0002416912,0.000031261763,0.00027208688,0.00043800718,0.0030637844,0.0019234482,0.01632028,0.9768395,0.000041346284],"about_ca_topic_score_codex":0.008760191,"about_ca_topic_score_gemma":0.005299137,"teacher_disagreement_score":0.06462744,"about_ca_system_score_codex":0.0024226175,"about_ca_system_score_gemma":0.0060404465,"threshold_uncertainty_score":0.21620029},"labels":[],"label_agreement":null},{"id":"W2792572948","doi":"10.1145/3186549.3186559","title":"Data Quality","year":2018,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Quality and Management","field":"Decision Sciences","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; University of Waterloo","funders":"","keywords":"Computer science; Scope (computer science); Quality (philosophy); Data quality; Empiricism; Data science; Action (physics); Data mining; Epistemology; Programming language; Engineering","score_opus":0.6848192121373361,"score_gpt":0.55904053398058,"score_spread":0.12577867815675614,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792572948","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":[0.006542586,0.01501703,0.46633467,0.31668064,0.008321189,0.0024976784,0.007229406,0.0020884564,0.17528841],"genre_scores_gemma":[0.2743502,0.017660363,0.5273928,0.09971249,0.009577886,0.0051124454,0.015738104,0.0026552517,0.047800545],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.7142184,0.13483924,0.03182683,0.018473815,0.09605062,0.004591084],"domain_scores_gemma":[0.36518064,0.23131172,0.0320962,0.18640155,0.17247032,0.012539674],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.23770383,0.00086830824,0.0020890315,0.008266796,0.005483202,0.02800883,0.0064671193,0.004075619,0.027704155],"category_scores_gemma":[0.50652224,0.0011696044,0.0027114619,0.012324504,0.011077821,0.025510427,0.014574925,0.008119707,0.009645497],"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":[0.00014278141,0.000095048,0.01157377,0.0017202668,0.00019285818,0.00019059239,0.002376759,0.001489982,0.0004210506,0.62495565,0.11219919,0.24464203],"study_design_scores_gemma":[0.000053706175,0.00011642106,0.0033375511,0.002637476,0.00006697126,0.00034630817,0.0014342512,0.0016993791,0.0006453721,0.30305153,0.68653667,0.0000742903],"about_ca_topic_score_codex":0.010360619,"about_ca_topic_score_gemma":0.006116215,"teacher_disagreement_score":0.23770383,"about_ca_system_score_codex":0.013526931,"about_ca_system_score_gemma":0.037954606,"threshold_uncertainty_score":0.9400469},"labels":[],"label_agreement":null},{"id":"W2924482967","doi":"","title":"SIGMOD officers, committees, and awardees","year":2018,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Operations research; Engineering","score_opus":0.1562836903849191,"score_gpt":0.39958671011275426,"score_spread":0.24330301972783516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2924482967","genre_codex":"editorial","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":[0.008409055,0.020405123,0.01167927,0.34162176,0.3693281,0.0039665652,0.007491607,0.002507641,0.23459077],"genre_scores_gemma":[0.021450678,0.012711585,0.008195883,0.019626608,0.055908415,0.002091556,0.0063677477,0.0008973735,0.8727502],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9753772,0.0037180143,0.0016253865,0.0015534778,0.013520168,0.0042058122],"domain_scores_gemma":[0.8498933,0.0038760058,0.0039078733,0.0056662187,0.07162243,0.06503422],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.031384017,0.0018923961,0.0028704447,0.0062618232,0.0058074193,0.013927511,0.0021030067,0.0052241087,0.11911113],"category_scores_gemma":[0.05174795,0.0012676965,0.0012741388,0.0061551146,0.0013311881,0.0063640983,0.008761388,0.0071367314,0.11176997],"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":[0.00010011587,0.000091910384,0.0012497953,0.00007306883,0.000010137992,0.00003280937,0.000046909045,0.000069237205,0.00021231589,0.001442365,0.93997264,0.056698613],"study_design_scores_gemma":[0.000069757174,0.00007173434,0.0026942552,0.00010867611,0.000015331087,0.000039136463,0.00038436565,0.00029653206,0.00018961968,0.0017115449,0.9943955,0.000023723116],"about_ca_topic_score_codex":0.005472243,"about_ca_topic_score_gemma":0.01114165,"teacher_disagreement_score":0.11911113,"about_ca_system_score_codex":0.0032489283,"about_ca_system_score_gemma":0.020646283,"threshold_uncertainty_score":0.3984664},"labels":[],"label_agreement":null},{"id":"W2999873292","doi":"10.1145/376284.375664","title":"Efficient computation of Iceberg cubes with complex measures","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management and Algorithms","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":"Simon Fraser University","funders":"","keywords":"Pruning; Computer science; Online analytical processing; Data cube; Computation; Scalability; Cube (algebra); Data mining; Measure (data warehouse); Tree (set theory); Theoretical computer science; Algorithm; Data warehouse; Database; Mathematics; Combinatorics","score_opus":0.042388281774761956,"score_gpt":0.2641887323800171,"score_spread":0.22180045060525516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2999873292","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":[0.032441907,0.0001995195,0.9645909,0.00015084211,0.000021937845,0.000068453366,0.0002186759,0.00087729545,0.0014304298],"genre_scores_gemma":[0.20555419,0.00023087565,0.79214716,0.00007059908,0.000030492374,0.00016534593,0.0008316444,0.00016410615,0.0008055709],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99830735,0.00032810675,0.00017797563,0.00021924409,0.0008098092,0.00015749341],"domain_scores_gemma":[0.9943363,0.0027579335,0.0005731922,0.00094285974,0.0012015008,0.000188189],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015574953,0.00079946936,0.0013231976,0.0017752809,0.00087126356,0.0024585684,0.0017908665,0.00066212803,0.001469706],"category_scores_gemma":[0.010469437,0.0006402285,0.001005185,0.0028727655,0.0010443392,0.0047466233,0.0024789541,0.0012854047,0.00046108582],"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":[0.00030719495,0.00009294182,0.0053642727,0.00035351884,0.00016014706,0.00030978033,0.0005536324,0.51284295,0.009890629,0.13347912,0.007221444,0.32942435],"study_design_scores_gemma":[0.000012321984,0.00002829124,0.00030991694,0.000012209676,0.000011104576,0.000064310385,0.00008765049,0.93634814,0.0045898557,0.0564762,0.0020475413,0.000012488533],"about_ca_topic_score_codex":0.004800087,"about_ca_topic_score_gemma":0.0064613954,"teacher_disagreement_score":0.004800087,"about_ca_system_score_codex":0.0011290105,"about_ca_system_score_gemma":0.0019358221,"threshold_uncertainty_score":0.009544313},"labels":[],"label_agreement":null},{"id":"W3041888157","doi":"10.1145/376284.375685","title":"Independence is good","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":54,"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; Curse of dimensionality; Data mining; Histogram; Independence (probability theory); Exploit; Set (abstract data type); Algorithm; Theoretical computer science; Artificial intelligence; Mathematics; Image (mathematics)","score_opus":0.025710358788209507,"score_gpt":0.2582570222251933,"score_spread":0.2325466634369838,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3041888157","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032417987,0.0032310933,0.8589856,0.009185302,0.0007949766,0.00033392545,0.0048040967,0.003306734,0.08694024],"genre_scores_gemma":[0.7009514,0.006058804,0.21639119,0.00663498,0.0021116873,0.0009393914,0.013439586,0.0020486596,0.05142418],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9886979,0.0022245646,0.0008166891,0.003751446,0.0036419085,0.000867553],"domain_scores_gemma":[0.97638,0.008175113,0.0012838108,0.010256324,0.0030233741,0.000881431],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006546968,0.0013303618,0.0022482579,0.0027397885,0.002525625,0.0059406552,0.0021121425,0.0019672092,0.031556774],"category_scores_gemma":[0.049819168,0.0012149063,0.0016528536,0.0050619864,0.0034067316,0.017440377,0.006522203,0.004075538,0.014375335],"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":[0.000878419,0.00023059499,0.015591348,0.00082245487,0.00047503988,0.0007181868,0.0017374853,0.03195253,0.0047108536,0.44283807,0.11889157,0.3811535],"study_design_scores_gemma":[0.000096996424,0.00018144421,0.006437684,0.0002255331,0.00016665841,0.0024717946,0.000992563,0.08963518,0.0036626637,0.74958473,0.14639276,0.00015188854],"about_ca_topic_score_codex":0.0042968797,"about_ca_topic_score_gemma":0.0038271276,"teacher_disagreement_score":0.031556774,"about_ca_system_score_codex":0.0017451787,"about_ca_system_score_gemma":0.0028238855,"threshold_uncertainty_score":0.10556799},"labels":[],"label_agreement":null},{"id":"W3111145856","doi":"10.1145/3444831.3444841","title":"Advice from SIGMOD/PODS 2020","year":2020,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Conferences and Exhibitions Management","field":"Social 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":"Western University","funders":"","keywords":"Notice; Computer science; Advice (programming); Component (thermodynamics); World Wide Web; Political science; Programming language","score_opus":0.046734223800662306,"score_gpt":0.3025489882476701,"score_spread":0.2558147644470078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111145856","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009835328,0.018942341,0.008575372,0.35754278,0.0783818,0.0006129358,0.033370595,0.023652148,0.47793847],"genre_scores_gemma":[0.0071160756,0.028318837,0.030188186,0.10745922,0.041450523,0.0005781429,0.06670578,0.014410366,0.7037728],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.979652,0.00430342,0.0017985437,0.0014060051,0.01161164,0.0012283832],"domain_scores_gemma":[0.953247,0.0038169092,0.0016536504,0.003386259,0.028537259,0.009358951],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020569876,0.0019191187,0.0014303412,0.008262606,0.0028324758,0.011597603,0.0027578743,0.00501031,0.21011117],"category_scores_gemma":[0.045313682,0.00075350865,0.0009926684,0.008683868,0.0006680877,0.009914114,0.004512795,0.0058164345,0.28716046],"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":[0.000011387309,0.000008642163,0.000044182834,0.000028080272,9.133617e-7,0.000006580182,0.000011299662,0.000008326542,0.000031287476,0.00017561289,0.9877389,0.011934771],"study_design_scores_gemma":[0.000006279287,0.0000041606386,0.0000558678,0.000039344024,0.0000012921648,0.0000125707775,0.000030022036,0.00002122455,0.000040196384,0.00013537602,0.9996493,0.0000043379064],"about_ca_topic_score_codex":0.012742224,"about_ca_topic_score_gemma":0.016624339,"teacher_disagreement_score":0.21011117,"about_ca_system_score_codex":0.0026714327,"about_ca_system_score_gemma":0.008503399,"threshold_uncertainty_score":0.7028919},"labels":[],"label_agreement":null},{"id":"W3111316923","doi":"10.1145/376284.375749","title":"Exploiting constraint-like data characterizations in query optimization","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","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":"IBM (Canada); York University","funders":"","keywords":"Computer science; Query optimization; Constraint (computer-aided design); Data integrity; Binary constraint; Exploit; Constraint programming; USable; Data mining; Database; Constraint satisfaction; Constraint logic programming; Mathematical optimization; Probabilistic logic; Artificial intelligence","score_opus":0.0655448349157439,"score_gpt":0.28878375790549454,"score_spread":0.22323892298975065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111316923","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":[0.013516058,0.00053514296,0.97772753,0.0012988885,0.00007417897,0.00022288981,0.0016169639,0.0016996318,0.0033087023],"genre_scores_gemma":[0.25153238,0.00085434155,0.7365792,0.0010510901,0.00026094675,0.00076257816,0.00386407,0.0019154547,0.0031799872],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97685224,0.0062505715,0.0032154787,0.002623151,0.009821761,0.001236855],"domain_scores_gemma":[0.94747037,0.032541674,0.0049756127,0.009590062,0.0048651593,0.0005569927],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012322414,0.0018573639,0.0020265568,0.0038236992,0.0014397112,0.008370128,0.003457537,0.0017059271,0.0039816382],"category_scores_gemma":[0.052416146,0.0016825604,0.0028108,0.008371071,0.0035572944,0.016683936,0.0044934647,0.005209405,0.0010982346],"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":[0.0006956754,0.00028262933,0.0076555107,0.0010914839,0.00032882366,0.0005723924,0.0016873954,0.30866703,0.012508991,0.47679797,0.012657132,0.17705506],"study_design_scores_gemma":[0.000054396474,0.00009577501,0.0007337798,0.000120030636,0.00007058344,0.0002334465,0.0003703543,0.6821553,0.009037998,0.28916076,0.017853035,0.000114547904],"about_ca_topic_score_codex":0.008247627,"about_ca_topic_score_gemma":0.010639227,"teacher_disagreement_score":0.012322414,"about_ca_system_score_codex":0.003131245,"about_ca_system_score_gemma":0.003382669,"threshold_uncertainty_score":0.065167904},"labels":[],"label_agreement":null},{"id":"W3136198913","doi":"10.1145/3261861","title":"Session details: Database principle","year":2006,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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 Toronto","funders":"","keywords":"Computer science; Session (web analytics); Database; Information retrieval; Programming language; World Wide Web","score_opus":0.020976961408618672,"score_gpt":0.2764887543090294,"score_spread":0.25551179290041076,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3136198913","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":[0.0010556744,0.009137281,0.017437464,0.028591411,0.015303446,0.00047521113,0.014675155,0.0033484807,0.9099759],"genre_scores_gemma":[0.02340777,0.0066188565,0.0054102065,0.004402653,0.009811939,0.0002849167,0.008401434,0.0010711061,0.9405912],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981817,0.00041448427,0.00010118701,0.00049650914,0.0006401483,0.00016595008],"domain_scores_gemma":[0.9963882,0.0011610233,0.000121307996,0.001080489,0.0006243886,0.00062465714],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0025432429,0.0010684233,0.002055082,0.0013379406,0.003033555,0.008892196,0.0022716983,0.0039307293,0.7449129],"category_scores_gemma":[0.007563874,0.0008691293,0.0012786089,0.0030365817,0.0013556508,0.005613128,0.00378286,0.0046311095,0.63862675],"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":[0.00015374333,0.000058575402,0.00012637179,0.00023346764,0.000015806214,0.000042888045,0.00004426866,0.00007450191,0.0010381818,0.038155094,0.9222961,0.037760913],"study_design_scores_gemma":[0.00006828202,0.00002727544,0.00026549867,0.00006151946,0.000012773444,0.0001241126,0.000032963744,0.00024950146,0.00056436367,0.012964362,0.98561674,0.0000124860935],"about_ca_topic_score_codex":0.00095629215,"about_ca_topic_score_gemma":0.0011247194,"teacher_disagreement_score":0.7449129,"about_ca_system_score_codex":0.0020604725,"about_ca_system_score_gemma":0.001719207,"threshold_uncertainty_score":0.3638507},"labels":[],"label_agreement":null},{"id":"W3175620523","doi":"10.1145/3471485.3471506","title":"Scaling Dynamic Hash Tables on Real Persistent Memory","year":2021,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Parallel Computing and Optimization Techniques","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":"Simon Fraser University","funders":"","keywords":"Computer science; Hash function; Scalability; Emulation; Parallel computing; Hash table; Latency (audio); Embedded system; Operating system; Programming language","score_opus":0.021875223175109013,"score_gpt":0.2662600561580582,"score_spread":0.2443848329829492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3175620523","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":[0.44947046,0.0044547454,0.50748324,0.0014240679,0.0010176913,0.00021812397,0.0009795787,0.013176721,0.021775445],"genre_scores_gemma":[0.88069165,0.0007888601,0.111065835,0.00021043401,0.00019474108,0.00014936927,0.0005639264,0.0003463718,0.005988839],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988794,0.00018684063,0.00010008167,0.000204469,0.0004346222,0.00019466113],"domain_scores_gemma":[0.9945148,0.001784693,0.00023613156,0.0021579778,0.0010809287,0.00022546233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013588216,0.00066887157,0.0005850234,0.0010426303,0.00056963065,0.0019029566,0.0021495197,0.00057035865,0.0070410687],"category_scores_gemma":[0.009530989,0.00048416466,0.00027083527,0.0014395666,0.00086027867,0.0055168006,0.0021871035,0.00074735604,0.0017509176],"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":[0.0027266198,0.00041517726,0.0055603636,0.00070796255,0.00012977692,0.0004241693,0.0006663574,0.30512252,0.07804462,0.098739356,0.031526837,0.4759362],"study_design_scores_gemma":[0.0002538356,0.00062405184,0.00092661317,0.00006844081,0.00006693028,0.0002878549,0.00032005296,0.87583077,0.04069316,0.05797785,0.022881249,0.00006911812],"about_ca_topic_score_codex":0.0013428547,"about_ca_topic_score_gemma":0.0013067115,"teacher_disagreement_score":0.0070410687,"about_ca_system_score_codex":0.0010535929,"about_ca_system_score_gemma":0.0009850481,"threshold_uncertainty_score":0.023554742},"labels":[],"label_agreement":null},{"id":"W3175911760","doi":"10.1145/3471485.3471494","title":"Efficient Directed Densest Subgraph Discovery","year":2021,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Web Data Mining and Analysis","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 British Columbia","funders":"","keywords":"Scalability; Computer science; Induced subgraph isomorphism problem; Graph; Enhanced Data Rates for GSM Evolution; Subgraph isomorphism problem; Core (optical fiber); Theoretical computer science; Graph factorization; Efficient algorithm; Algorithm; Database; Line graph; Artificial intelligence; Voltage graph","score_opus":0.01560761602411701,"score_gpt":0.23701589540386872,"score_spread":0.22140827937975172,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3175911760","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":[0.05953621,0.001437167,0.91809434,0.0015176787,0.00015416765,0.00050292996,0.0046239356,0.00794359,0.006189918],"genre_scores_gemma":[0.22555059,0.00054795924,0.75403965,0.0004047903,0.00009434208,0.0003013645,0.013296161,0.0005298563,0.0052353996],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99785465,0.00042533423,0.00011526602,0.0006622122,0.00068283937,0.0002597783],"domain_scores_gemma":[0.99507153,0.0020252871,0.0004126014,0.0015211515,0.00071022875,0.00025928483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013756008,0.0014002289,0.0019965654,0.0034538913,0.0014033555,0.00216501,0.0031527968,0.0016420207,0.0041606375],"category_scores_gemma":[0.009554587,0.00093083165,0.0018441174,0.0054556434,0.0008949503,0.00408294,0.0029121686,0.0014927705,0.0017282045],"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":[0.00048032985,0.00049685634,0.009645627,0.000995839,0.00028105275,0.0005265068,0.00073066796,0.23649108,0.010193095,0.048327584,0.07449201,0.6173393],"study_design_scores_gemma":[0.00008589556,0.00005413729,0.0010392214,0.000034712655,0.000055965473,0.00034231256,0.00028185107,0.92299825,0.0037124283,0.06293352,0.008440258,0.00002142733],"about_ca_topic_score_codex":0.012571457,"about_ca_topic_score_gemma":0.028330615,"teacher_disagreement_score":0.012571457,"about_ca_system_score_codex":0.0017899616,"about_ca_system_score_gemma":0.004011743,"threshold_uncertainty_score":0.024996579},"labels":[],"label_agreement":null},{"id":"W3215306199","doi":"10.1145/3516431.3516435","title":"Federated Data Science to Break Down Silos [Vision]","year":2022,"lang":"en","type":"preprint","venue":"ACM SIGMOD Record","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":"Concordia University","funders":"","keywords":"Computer science; Data sharing; Focus (optics); Data science; Open science; Pipeline transport; Open data; Open research; World Wide Web; Engineering","score_opus":0.28799195367350366,"score_gpt":0.46530460369348814,"score_spread":0.17731265001998447,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3215306199","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":[0.0077208714,0.0009812609,0.9650272,0.0048899315,0.00045041263,0.0003312076,0.00075933785,0.00867615,0.0111636035],"genre_scores_gemma":[0.08205165,0.00076972,0.9035294,0.0017585141,0.00018914751,0.00029225808,0.0034446637,0.0015528337,0.006411823],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98609585,0.004471441,0.001212217,0.0029924791,0.004279564,0.0009483972],"domain_scores_gemma":[0.95863867,0.0058684032,0.002447697,0.026866836,0.0042120903,0.0019662154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02379974,0.001247634,0.001536975,0.0060937805,0.004882351,0.011194012,0.0046411715,0.003242303,0.006358399],"category_scores_gemma":[0.032043997,0.0013191503,0.0030113026,0.0054638125,0.0072487844,0.030564299,0.021370517,0.0058691157,0.0033038745],"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":[0.00024366456,0.00025646674,0.0027934837,0.00062052207,0.00022321586,0.0002411443,0.0022419575,0.011848011,0.0042849905,0.79831743,0.02883026,0.15009885],"study_design_scores_gemma":[0.000048986356,0.00008223755,0.00065409107,0.0003283531,0.00007727146,0.00029679286,0.0006461132,0.0356362,0.005784871,0.77174336,0.1846176,0.000084121064],"about_ca_topic_score_codex":0.00400276,"about_ca_topic_score_gemma":0.0033524,"teacher_disagreement_score":0.02379974,"about_ca_system_score_codex":0.0032517202,"about_ca_system_score_gemma":0.0123438295,"threshold_uncertainty_score":0.12586653},"labels":[],"label_agreement":null},{"id":"W3217479643","doi":"10.1145/3503780.3503786","title":"VLDB Panel Summary","year":2021,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","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":"Computer science; Very large database; SQL; Database; Point (geometry); World Wide Web; Mathematics education; Information retrieval","score_opus":0.12357829945750257,"score_gpt":0.28860616529085836,"score_spread":0.1650278658333558,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3217479643","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":[0.00067480374,0.005397174,0.0018824348,0.088919505,0.038818642,0.00069740525,0.05947886,0.0016425067,0.80248857],"genre_scores_gemma":[0.0045147804,0.004659554,0.0010466145,0.033006404,0.01167058,0.00049677264,0.038526215,0.000691768,0.9053874],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981153,0.000194279,0.0001242448,0.0003892859,0.00094291643,0.00023392809],"domain_scores_gemma":[0.99265516,0.00070824573,0.00023577725,0.0006186873,0.004885912,0.0008963457],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0033775868,0.0006946123,0.0008173629,0.0018182088,0.0017550805,0.0079326,0.0021010023,0.0043375934,0.43359518],"category_scores_gemma":[0.00618391,0.00041212555,0.0006295074,0.0031867083,0.00047514314,0.003952252,0.0017424321,0.0035106177,0.3815129],"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":[0.0000090836675,0.0000055859305,0.000045654233,0.0000332128,7.631382e-7,0.00000643975,0.0000025747563,0.000011088374,0.000042246666,0.0005721973,0.99261284,0.006658267],"study_design_scores_gemma":[0.0000060603215,0.000004927956,0.00034307287,0.00007316591,0.0000012994999,0.0000072346024,0.000011812648,0.0000133467975,0.000041253123,0.00030256866,0.9991922,0.0000031408829],"about_ca_topic_score_codex":0.0072864834,"about_ca_topic_score_gemma":0.0091265645,"teacher_disagreement_score":0.43359518,"about_ca_system_score_codex":0.0020533341,"about_ca_system_score_gemma":0.0035056134,"threshold_uncertainty_score":0.8079077},"labels":[],"label_agreement":null},{"id":"W4232924440","doi":"10.1145/3262559","title":"Session details: Database principles","year":2004,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Mathematics, Computing, and Information Processing","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; Session (web analytics); Database; Information retrieval; World Wide Web","score_opus":0.04760966187545459,"score_gpt":0.2783804266532686,"score_spread":0.23077076477781402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232924440","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":[0.0010962941,0.01502549,0.02452165,0.024976477,0.017286014,0.001180166,0.028955592,0.010531543,0.87642676],"genre_scores_gemma":[0.007632138,0.0076329173,0.0048894994,0.003021419,0.004544468,0.0003758716,0.010680815,0.00169691,0.9595259],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99834335,0.00033234846,0.00011088762,0.00044902935,0.0006178106,0.00014650935],"domain_scores_gemma":[0.995488,0.0013386883,0.00014274371,0.0012074899,0.0008308839,0.0009921469],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0031181194,0.0016303121,0.0029013641,0.0018621731,0.0031150987,0.010574529,0.0024457905,0.0043966756,0.82640326],"category_scores_gemma":[0.0075316383,0.001234428,0.001192831,0.0052719964,0.001086837,0.0059558637,0.003918667,0.0042874753,0.8058476],"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":[0.00007911057,0.000067925284,0.000109401466,0.00020110249,0.000012125569,0.000027605443,0.00003353102,0.00007002989,0.0008693375,0.006623418,0.954427,0.037479233],"study_design_scores_gemma":[0.000047281163,0.000028224822,0.0002270522,0.00007119654,0.00000876784,0.000074313866,0.000028175311,0.00022321487,0.00038759442,0.0035355417,0.9953566,0.000011958245],"about_ca_topic_score_codex":0.0012095195,"about_ca_topic_score_gemma":0.0017300879,"teacher_disagreement_score":0.82640326,"about_ca_system_score_codex":0.001758741,"about_ca_system_score_gemma":0.0017134537,"threshold_uncertainty_score":0.24761462},"labels":[],"label_agreement":null},{"id":"W4238142139","doi":"10.1145/3262194","title":"Session details: Database principles","year":2006,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Mathematics, Computing, and Information Processing","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; Session (web analytics); Database; World Wide Web; Information retrieval","score_opus":0.03559642318195582,"score_gpt":0.2633261416628337,"score_spread":0.22772971848087786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4238142139","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":[0.0010830645,0.014415919,0.024756957,0.024121765,0.017703582,0.0012264139,0.029987024,0.0104651395,0.8762401],"genre_scores_gemma":[0.007715449,0.007402672,0.0049133482,0.0029477952,0.004563891,0.00038761823,0.010612463,0.0017019733,0.95975477],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983222,0.0003389968,0.000113144124,0.0004633322,0.000613474,0.00014883894],"domain_scores_gemma":[0.99540377,0.0013609033,0.00014335601,0.0012470616,0.0008432836,0.0010016247],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0031452735,0.001647362,0.0029419803,0.0018128243,0.0030800719,0.010622596,0.0024835747,0.0043195477,0.83018225],"category_scores_gemma":[0.0076550143,0.0012108836,0.0011994627,0.005159655,0.0010883097,0.005798278,0.0039467057,0.004279446,0.80997986],"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":[0.000080465165,0.0000694258,0.000110316156,0.00020371078,0.000012369341,0.000027827648,0.00003310327,0.00007021804,0.00088352076,0.0065555396,0.9546696,0.037283927],"study_design_scores_gemma":[0.000048170557,0.000029298224,0.00023126365,0.000073057105,0.000008724645,0.00007489595,0.000027922484,0.00023274362,0.0003945492,0.0034288003,0.9954385,0.000012118423],"about_ca_topic_score_codex":0.0011925101,"about_ca_topic_score_gemma":0.00169544,"teacher_disagreement_score":0.83018225,"about_ca_system_score_codex":0.0017585971,"about_ca_system_score_gemma":0.0017375876,"threshold_uncertainty_score":0.2422244},"labels":[],"label_agreement":null},{"id":"W4238783465","doi":"10.1145/376284.375729","title":"Data-driven understanding and refinement of schema mappings","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","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; Schema (genetic algorithms); Schema evolution; Database schema; Theoretical computer science; Programming language; Information retrieval; Data mining; Database design","score_opus":0.13553835996809832,"score_gpt":0.3042947998079359,"score_spread":0.16875643983983757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4238783465","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":[0.012513582,0.00021952407,0.9807535,0.0011355805,0.00004818093,0.00015676397,0.00028035042,0.0011198766,0.0037726746],"genre_scores_gemma":[0.10220309,0.00046154237,0.89327186,0.0003206947,0.000031809574,0.00021010042,0.0010644493,0.0005285489,0.0019079648],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.98668694,0.0058123986,0.00142797,0.0018264668,0.0038078094,0.0004385038],"domain_scores_gemma":[0.9715062,0.013892281,0.0013318717,0.007936252,0.00496078,0.00037266462],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015410757,0.001073986,0.00084076915,0.0027778153,0.0015185052,0.0071611647,0.0032632449,0.0021471188,0.003020016],"category_scores_gemma":[0.055120893,0.0014320455,0.0018751998,0.0022797447,0.0041021374,0.01617909,0.0062756357,0.004562275,0.0007389252],"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":[0.00013355253,0.00010270762,0.0030220908,0.00048381748,0.00006745932,0.00087664597,0.017892819,0.019719634,0.009585577,0.8238911,0.006660381,0.1175642],"study_design_scores_gemma":[0.00004326381,0.000065912966,0.0006466902,0.00040527986,0.00007805056,0.0008721027,0.003285484,0.14082493,0.022499492,0.6979947,0.1331951,0.00008895947],"about_ca_topic_score_codex":0.004231218,"about_ca_topic_score_gemma":0.004064025,"teacher_disagreement_score":0.015410757,"about_ca_system_score_codex":0.0016280542,"about_ca_system_score_gemma":0.0027204342,"threshold_uncertainty_score":0.08150083},"labels":[],"label_agreement":null},{"id":"W4241938281","doi":"10.1145/376284.375672","title":"Data bubbles","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Clustering Algorithms Research","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 British Columbia","funders":"","keywords":"Cluster analysis; Computer science; Data mining; CURE data clustering algorithm; Data stream clustering; Canopy clustering algorithm; Hierarchical clustering; Correlation clustering; Set (abstract data type); Data set; Data compression; Key (lock); Constrained clustering; Determining the number of clusters in a data set; Fuzzy clustering; Algorithm; Artificial intelligence","score_opus":0.123803096386827,"score_gpt":0.366511782884858,"score_spread":0.24270868649803096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241938281","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":[0.016694076,0.00478333,0.9249112,0.008200641,0.002105885,0.0007143421,0.003767193,0.0040272474,0.034796122],"genre_scores_gemma":[0.18941064,0.0041444814,0.75151587,0.00576521,0.001150831,0.0013564934,0.009064232,0.0019103991,0.035681877],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9950486,0.0011151726,0.0003746202,0.0010955689,0.0021119881,0.0002540601],"domain_scores_gemma":[0.9850007,0.0068379696,0.0007109692,0.004603016,0.002271746,0.0005755757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004386181,0.0007071546,0.0009842973,0.0022737775,0.0017261559,0.0038416295,0.0030857143,0.001957915,0.024653232],"category_scores_gemma":[0.02955338,0.00058529276,0.0010680081,0.0029977176,0.002020341,0.011695137,0.006867439,0.0025594598,0.007902516],"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":[0.00052630843,0.00012455783,0.0030848607,0.00084606075,0.000083555926,0.0006828432,0.0013627026,0.008804914,0.0053592813,0.5522953,0.07293121,0.35389844],"study_design_scores_gemma":[0.00007453088,0.00017459349,0.0008361073,0.00022629075,0.0000469375,0.001445506,0.00079367036,0.03794724,0.007958464,0.3173856,0.6330255,0.000085509586],"about_ca_topic_score_codex":0.001573575,"about_ca_topic_score_gemma":0.0011476026,"teacher_disagreement_score":0.024653232,"about_ca_system_score_codex":0.0013069317,"about_ca_system_score_gemma":0.0012555909,"threshold_uncertainty_score":0.08247328},"labels":[],"label_agreement":null},{"id":"W4245210711","doi":"10.1145/376284.375666","title":"Iceberg-cube computation with PC clusters","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Management and Algorithms","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 Hospital","funders":"","keywords":"Computer science; Computation; Cube (algebra); Cuboid; Parallel computing; Algorithm; Data cube; Data mining; Mathematics","score_opus":0.020947862199152976,"score_gpt":0.246467051847856,"score_spread":0.22551918964870302,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4245210711","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":[0.12862606,0.0007934883,0.83054835,0.0011080959,0.00026132702,0.00026585805,0.0008908684,0.0071731615,0.030332778],"genre_scores_gemma":[0.4207923,0.0005673945,0.5675953,0.0002394524,0.000075744734,0.00029346746,0.0015545599,0.00040050625,0.008481237],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99874985,0.00024914474,0.00007345533,0.00027401192,0.00044116855,0.00021242505],"domain_scores_gemma":[0.9975884,0.00082607626,0.00009971041,0.000873143,0.00047612723,0.0001365337],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009944371,0.00080232363,0.0012103089,0.0007164747,0.0012758401,0.0024842657,0.0032085427,0.00077384675,0.006522969],"category_scores_gemma":[0.0047108554,0.0005301471,0.000704361,0.0031945824,0.0008903866,0.0044657756,0.002447761,0.0011067105,0.0019961302],"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":[0.0015996727,0.00034672668,0.0058821877,0.00041362824,0.00020101057,0.00040660318,0.0006149264,0.5958847,0.00792232,0.09634133,0.038787514,0.2515994],"study_design_scores_gemma":[0.00006893655,0.00007254632,0.00031999263,0.000010560748,0.000017422677,0.00007250053,0.0001398966,0.95666873,0.0047247508,0.030622495,0.007266431,0.000015790927],"about_ca_topic_score_codex":0.016354175,"about_ca_topic_score_gemma":0.017040685,"teacher_disagreement_score":0.016354175,"about_ca_system_score_codex":0.0014783272,"about_ca_system_score_gemma":0.0021207056,"threshold_uncertainty_score":0.03251797},"labels":[],"label_agreement":null},{"id":"W4246773344","doi":"10.1145/3262443","title":"Session details: Database principles","year":2002,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Mathematics, Computing, and Information Processing","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; Session (web analytics); Database; Information retrieval; World Wide Web","score_opus":0.07280586807241568,"score_gpt":0.2691107056749638,"score_spread":0.19630483760254813,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4246773344","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":[0.001038337,0.014823786,0.023466563,0.02385145,0.017388616,0.0011486395,0.027604254,0.010061724,0.88061666],"genre_scores_gemma":[0.007221246,0.0076998165,0.0046839593,0.0030200814,0.004509101,0.00037069304,0.010094285,0.001660816,0.96073985],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984043,0.00031992956,0.00010697736,0.0004338448,0.0005937807,0.00014113808],"domain_scores_gemma":[0.99563307,0.0012892687,0.0001397656,0.0011617974,0.0008068754,0.00096923986],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0030687733,0.0015986072,0.002836782,0.0018180781,0.0030105324,0.010244796,0.0023851683,0.0043226457,0.8268066],"category_scores_gemma":[0.0074224826,0.0012193193,0.0011600886,0.005104647,0.0010869177,0.005793709,0.0038995002,0.0042834054,0.80930394],"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":[0.00007408854,0.00006570056,0.00010528508,0.00019842196,0.000011546624,0.00002660724,0.000032616572,0.00006720743,0.0008637563,0.0065846085,0.9549736,0.03699653],"study_design_scores_gemma":[0.000045181518,0.000026716652,0.00022341478,0.000071397095,0.000008398289,0.00007246882,0.000026372702,0.00021137623,0.0003665031,0.003366705,0.9955699,0.000011632725],"about_ca_topic_score_codex":0.0011526544,"about_ca_topic_score_gemma":0.0016333596,"teacher_disagreement_score":0.8268066,"about_ca_system_score_codex":0.0017099458,"about_ca_system_score_gemma":0.0016553033,"threshold_uncertainty_score":0.24703938},"labels":[],"label_agreement":null},{"id":"W4247707248","doi":"10.1145/3262132","title":"Session details: Database principles","year":2003,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Mathematics, Computing, and Information Processing","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; Session (web analytics); Database; Database design; Information retrieval; World Wide Web","score_opus":0.051649521499406346,"score_gpt":0.2777147422639968,"score_spread":0.22606522076459043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4247707248","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":[0.0010385785,0.01520652,0.024230763,0.024702564,0.017255064,0.001143365,0.027879393,0.010056918,0.8784869],"genre_scores_gemma":[0.0072752023,0.007821352,0.004806279,0.0029953767,0.004531268,0.00036859067,0.0103483405,0.0016758349,0.9601776],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99836725,0.000326258,0.000110414556,0.00043804405,0.00061612553,0.00014191742],"domain_scores_gemma":[0.9955237,0.0013192651,0.0001444012,0.0011967034,0.00083953695,0.00097640883],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0031062106,0.0015878448,0.0028140054,0.0018478339,0.0030330154,0.010397274,0.0024098596,0.004300499,0.82330376],"category_scores_gemma":[0.0075815558,0.001231088,0.0011487659,0.0052186875,0.0010779281,0.0059278146,0.00390945,0.0042806226,0.8044345],"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":[0.00007228613,0.000062847255,0.000104847415,0.00019802064,0.00001136176,0.000026181699,0.00003278582,0.000067209556,0.00084313535,0.006639969,0.95532256,0.03661878],"study_design_scores_gemma":[0.000043558415,0.000025861957,0.00021745067,0.00007057873,0.00000826966,0.000071402654,0.000026432139,0.00020955488,0.0003671115,0.0034311,0.9955173,0.000011442099],"about_ca_topic_score_codex":0.001172115,"about_ca_topic_score_gemma":0.0016815517,"teacher_disagreement_score":0.82330376,"about_ca_system_score_codex":0.0017281059,"about_ca_system_score_gemma":0.0017032234,"threshold_uncertainty_score":0.25203574},"labels":[],"label_agreement":null},{"id":"W4248260714","doi":"10.1145/376284.375693","title":"SPARTAN","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Bayesian Modeling and Causal Inference","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":"Bell (Canada)","funders":"","keywords":"Computer science; Spartan; Lossy compression; Data mining; Table (database); Data compression; Data structure; Theoretical computer science; Algorithm; Artificial intelligence; Programming language","score_opus":0.04318286130473845,"score_gpt":0.272622653250019,"score_spread":0.22943979194528055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4248260714","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":[0.014511814,0.0029990154,0.32283086,0.0008038355,0.0011158236,0.00076955813,0.031204194,0.4928765,0.13288835],"genre_scores_gemma":[0.1654998,0.004323076,0.3529821,0.0026615898,0.0004108267,0.0013713076,0.19595745,0.04060916,0.23618463],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99884367,0.00009517878,0.000099499484,0.00030883454,0.0005329144,0.0001198115],"domain_scores_gemma":[0.998002,0.0003194879,0.000103099854,0.00095478893,0.00053386565,0.0000867452],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010222339,0.0012510563,0.0010206437,0.0017094902,0.0007667274,0.0025086685,0.0037104578,0.0010152062,0.08956731],"category_scores_gemma":[0.0042513334,0.0007226873,0.00082075765,0.002191444,0.00050292024,0.0039260015,0.0023861772,0.001496404,0.06338695],"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":[0.0014057119,0.0002189918,0.0021255484,0.00069023104,0.00016655172,0.00032581142,0.00027128434,0.005714366,0.010004045,0.025959874,0.44073218,0.5123855],"study_design_scores_gemma":[0.00025399745,0.0002583699,0.0015722158,0.00012564173,0.00010129573,0.0009653523,0.00014069595,0.06504061,0.040080458,0.03232243,0.85900646,0.00013255636],"about_ca_topic_score_codex":0.002547307,"about_ca_topic_score_gemma":0.0031946525,"teacher_disagreement_score":0.08956731,"about_ca_system_score_codex":0.0008352586,"about_ca_system_score_gemma":0.0017615176,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4250963083","doi":"10.1145/376284.375677","title":"Filtering algorithms and implementation for very fast publish/subscribe systems","year":2001,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Peer-to-Peer Network Technologies","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 Toronto","funders":"","keywords":"Computer science; Publication; Scalability; Implementation; Event (particle physics); Workload; Polling; Database; Distributed computing; Computer network; Programming language; Operating system","score_opus":0.03398909794582906,"score_gpt":0.2917580653035587,"score_spread":0.25776896735772964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4250963083","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":[0.0027633754,0.00023004871,0.9878792,0.00020290114,0.000102081845,0.00018491315,0.00013318262,0.0066356463,0.0018687024],"genre_scores_gemma":[0.07208786,0.00050816964,0.9198536,0.00018598605,0.00014995108,0.0005707068,0.00088877487,0.0005108662,0.005244039],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949144,0.0008483362,0.0006658893,0.0006961261,0.0023052276,0.00057009864],"domain_scores_gemma":[0.9907782,0.0032110075,0.0004333434,0.0028846734,0.0024025172,0.00029025707],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0059346803,0.0011211977,0.0015683358,0.0029116382,0.0027924187,0.0069341613,0.005942703,0.0031318031,0.010097185],"category_scores_gemma":[0.017275922,0.0014252451,0.0015421315,0.004657581,0.0013867458,0.0077312896,0.002335577,0.0024946586,0.006779351],"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":[0.0008830143,0.00052462757,0.0027451843,0.0006135188,0.00025266357,0.00027882057,0.0006243823,0.11282192,0.01688608,0.18353517,0.031412147,0.64942247],"study_design_scores_gemma":[0.00028860327,0.00017725037,0.00063609483,0.000094835035,0.000113261,0.00029852227,0.00019723592,0.7942119,0.019216858,0.13419175,0.050443925,0.00012974559],"about_ca_topic_score_codex":0.008979181,"about_ca_topic_score_gemma":0.006601335,"teacher_disagreement_score":0.010097185,"about_ca_system_score_codex":0.003037913,"about_ca_system_score_gemma":0.0032699301,"threshold_uncertainty_score":0.03377843},"labels":[],"label_agreement":null},{"id":"W4288696362","doi":"10.1145/3552490.3552499","title":"Reminiscences on Influential Papers","year":2022,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Big Data Technologies and Applications","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":"University of Waterloo","funders":"","keywords":"Column (typography); Computer science; Value (mathematics); Citation; Key (lock); Reading (process); Library science; World Wide Web; Operations research; History; Telecommunications; Law; Computer security; Political science; Mathematics","score_opus":0.22026839757357825,"score_gpt":0.3827781944790289,"score_spread":0.16250979690545064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288696362","genre_codex":"editorial","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016805159,0.038655005,0.00094804395,0.28787377,0.64309734,0.000042158696,0.00033059856,0.00023828277,0.027134283],"genre_scores_gemma":[0.032704327,0.040513817,0.0011923189,0.22208175,0.60377854,0.0001647816,0.0005550676,0.0011592003,0.09785026],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9847339,0.003974069,0.00091485557,0.0018823085,0.006872904,0.0016220526],"domain_scores_gemma":[0.93229485,0.01999638,0.004321875,0.003423074,0.025538165,0.014425734],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014233461,0.0022165289,0.0013565808,0.007445313,0.010938416,0.021296136,0.0032121378,0.007203442,0.018969595],"category_scores_gemma":[0.09396681,0.00073070114,0.0022873234,0.007955117,0.0037451305,0.010962461,0.007622335,0.019315636,0.013745218],"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":[0.00002638135,0.000017538541,0.00021380505,0.00018511739,0.00002700071,0.00033842487,0.0015809508,0.000027347895,0.00010838389,0.004167404,0.98129934,0.012008267],"study_design_scores_gemma":[0.0000063356392,0.000011290589,0.00015701338,0.00025820875,0.000018283508,0.000274687,0.0013149527,0.000021955655,0.00009774662,0.0014704289,0.99634796,0.000021234113],"about_ca_topic_score_codex":0.0018345811,"about_ca_topic_score_gemma":0.0038590885,"teacher_disagreement_score":0.021296136,"about_ca_system_score_codex":0.0059338557,"about_ca_system_score_gemma":0.0041517876,"threshold_uncertainty_score":0.07527459},"labels":[],"label_agreement":null},{"id":"W4309505072","doi":"10.1145/3572751.3572765","title":"Characterizing I/O in Machine Learning with MLPerf Storage","year":2022,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Data Storage Technologies","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":"McGill University","funders":"","keywords":"Computer science; Focus (optics); Inference; Software; Machine learning; Computer data storage; Training set; Training (meteorology); Data access; Database; Artificial intelligence; Computer engineering; Operating system","score_opus":0.017438022209979483,"score_gpt":0.23393867511933225,"score_spread":0.21650065290935278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309505072","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":[0.66368085,0.010513806,0.27622098,0.0036429723,0.0007888303,0.000406122,0.0049188714,0.012904651,0.026922967],"genre_scores_gemma":[0.9519822,0.0014047562,0.037587494,0.00033296557,0.00024559654,0.0003001923,0.0021332996,0.0008043721,0.0052092355],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99699533,0.00027965606,0.00023574523,0.000617286,0.0010939072,0.00077806425],"domain_scores_gemma":[0.9870204,0.006247272,0.000933396,0.0032915252,0.0021252513,0.00038221103],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020214105,0.0012883309,0.0010253756,0.0025546446,0.0013328043,0.0027627873,0.0031053291,0.001020149,0.0073777027],"category_scores_gemma":[0.021449689,0.00060637784,0.0005541351,0.0033386345,0.0015613373,0.009864399,0.001944495,0.0013411044,0.002185463],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036265776,0.0009542507,0.037673533,0.0011081309,0.00016439527,0.0011341223,0.00095485605,0.27574992,0.05095065,0.08170464,0.04517791,0.500801],"study_design_scores_gemma":[0.0001034206,0.00052143564,0.008130414,0.00015538305,0.00008726779,0.0006186168,0.00039753618,0.8534338,0.04457492,0.07404394,0.01785126,0.00008193642],"about_ca_topic_score_codex":0.0046003135,"about_ca_topic_score_gemma":0.003972217,"teacher_disagreement_score":0.0073777027,"about_ca_system_score_codex":0.002506387,"about_ca_system_score_gemma":0.0028115097,"threshold_uncertainty_score":0.024680912},"labels":[],"label_agreement":null},{"id":"W4317952578","doi":"10.1145/3582302.3582312","title":"Mid-Career Researcher, huh?","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Health and Medical Research Impacts","field":"Medicine","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":"Wonder; Mistake; Negotiation; Promotion (chess); Ask price; Computer science; Work (physics); Law; Psychology; Social psychology; Political science; Business; Engineering","score_opus":0.4168772941303401,"score_gpt":0.4880706105098954,"score_spread":0.07119331637955534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317952578","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008476906,0.012136159,0.0021186233,0.3451948,0.1166999,0.00025364305,0.0043363255,0.0029321162,0.50785154],"genre_scores_gemma":[0.013611582,0.0036948966,0.00083581835,0.05351416,0.0045282654,0.000083866755,0.00080693926,0.00029410297,0.9226305],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99906737,0.0001865674,0.00003829354,0.00016170995,0.00027451233,0.00027145317],"domain_scores_gemma":[0.99201196,0.0002355135,0.00020561746,0.00024839104,0.0010132192,0.006285306],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0014060101,0.00035885943,0.00051849615,0.00089137256,0.0035884976,0.004681198,0.00088839565,0.0018331235,0.46813932],"category_scores_gemma":[0.008357118,0.00034237892,0.00028679147,0.0011929112,0.0008010532,0.003692048,0.0034392232,0.0019687572,0.25459728],"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":[0.000021288082,0.00002066177,0.00084820605,0.0000424212,0.0000013104012,0.00012737671,0.0001915321,0.0000035194698,0.0001151687,0.0009172077,0.9672769,0.030434398],"study_design_scores_gemma":[0.000010875083,0.0000409453,0.0015005576,0.000090925794,0.0000022414283,0.00029537716,0.0019131424,0.000016500584,0.00008115779,0.00053095125,0.9955077,0.000009685857],"about_ca_topic_score_codex":0.003151754,"about_ca_topic_score_gemma":0.016097799,"teacher_disagreement_score":0.998594,"about_ca_system_score_codex":0.0015101712,"about_ca_system_score_gemma":0.0027480496,"threshold_uncertainty_score":0.7586347},"labels":[],"label_agreement":null},{"id":"W4317952630","doi":"10.1145/3582302.3582314","title":"Report on the First International Workshop on Data Systems Education (DataEd '22)","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Teaching and Learning Programming","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 Toronto","funders":"","keywords":"Session (web analytics); Computer science; Bridging (networking); Panel discussion; Data management; Conjunction (astronomy); Data science; Engineering management; Library science; World Wide Web; Database; Engineering","score_opus":0.08682560969649208,"score_gpt":0.3362625693619047,"score_spread":0.2494369596654126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317952630","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":[0.021270534,0.042515397,0.056983735,0.16848391,0.27215645,0.002954713,0.016806027,0.0019182747,0.41691092],"genre_scores_gemma":[0.04588489,0.01587187,0.023898518,0.020676464,0.019297868,0.001716126,0.016219229,0.0021811936,0.85425377],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98869425,0.0034078732,0.000451551,0.0012578145,0.0041547306,0.0020337657],"domain_scores_gemma":[0.98258346,0.0025281147,0.000331016,0.0009798363,0.006078839,0.0074986797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017408447,0.001630288,0.001104907,0.0027906997,0.0026761247,0.0118524395,0.0022740914,0.0046741515,0.1138034],"category_scores_gemma":[0.013252183,0.0006345196,0.0016493127,0.0025029727,0.0013153245,0.0063697184,0.011439102,0.007755717,0.04202347],"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":[0.00026340358,0.0004763931,0.0006630269,0.00037996913,0.000023744467,0.00022106346,0.00096986693,0.00041832752,0.0012269759,0.006100728,0.8971931,0.09206345],"study_design_scores_gemma":[0.000034011093,0.000087514505,0.0006370473,0.0001971809,0.000010430162,0.0000491693,0.0008130715,0.00017032944,0.0006718954,0.0012264971,0.9960846,0.000018257542],"about_ca_topic_score_codex":0.0054864744,"about_ca_topic_score_gemma":0.00970119,"teacher_disagreement_score":0.1138034,"about_ca_system_score_codex":0.0029091171,"about_ca_system_score_gemma":0.009094537,"threshold_uncertainty_score":0.3807103},"labels":[],"label_agreement":null},{"id":"W4379797318","doi":"10.1145/3604437.3604439","title":"TECHNICAL PERSPECTIVE: Ad Hoc Transactions: What They Are and Why We Should Care","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"Computer science; Perspective (graphical); Set (abstract data type); Data science; Programming language; Artificial intelligence","score_opus":0.04067911802497789,"score_gpt":0.3036587340324514,"score_spread":0.26297961600747355,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379797318","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006473368,0.047884073,0.11488879,0.75153434,0.006639385,0.000094207426,0.00033047723,0.00033264366,0.071822725],"genre_scores_gemma":[0.46752176,0.08010297,0.12092388,0.2569579,0.02662958,0.0004093403,0.000596398,0.0008807864,0.045977432],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98589104,0.0069203973,0.00090165855,0.001966021,0.0033138597,0.0010070164],"domain_scores_gemma":[0.9379139,0.03679234,0.0031332038,0.0054027922,0.013220209,0.0035375452],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015778873,0.0009781562,0.0009173563,0.0020925435,0.0032078964,0.015887482,0.0028814867,0.00981984,0.010096075],"category_scores_gemma":[0.042235024,0.0009843502,0.0006755573,0.002557318,0.026475355,0.036370583,0.003151754,0.013747375,0.0043179076],"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":[0.000064187436,0.00008189101,0.0019262319,0.00079593627,0.000030848616,0.0002529192,0.003994606,0.00051027193,0.0005492696,0.8554536,0.08733588,0.049004365],"study_design_scores_gemma":[0.000043458054,0.00006107516,0.00095202925,0.00088744593,0.00003010877,0.00096698187,0.006598729,0.0013607497,0.00057409756,0.65606385,0.33240768,0.000053861644],"about_ca_topic_score_codex":0.004608076,"about_ca_topic_score_gemma":0.003056904,"teacher_disagreement_score":0.015887482,"about_ca_system_score_codex":0.0054168687,"about_ca_system_score_gemma":0.006526047,"threshold_uncertainty_score":0.083447635},"labels":[],"label_agreement":null},{"id":"W4379800102","doi":"10.1145/3604437.3604444","title":"Efficiently Making Cross-Engine Transactions Consistent","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"Simon Fraser University","funders":"","keywords":"Computer science; Limiting; Isolation (microbiology); Correctness; Database; Operating system; Programming language","score_opus":0.032594072378710856,"score_gpt":0.3036519960292631,"score_spread":0.27105792365055226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379800102","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":[0.131368,0.00070984644,0.850181,0.0010490136,0.00030458544,0.0007091996,0.00045809022,0.0094503695,0.0057699215],"genre_scores_gemma":[0.5190859,0.0005305575,0.46761855,0.0006136614,0.00021280405,0.000499474,0.0021022945,0.0026391323,0.0066976044],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9679672,0.009050073,0.0043172254,0.004023287,0.011628584,0.0030136271],"domain_scores_gemma":[0.8914718,0.032990605,0.005672946,0.04858431,0.019705359,0.0015749469],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.024833618,0.001588095,0.0022636177,0.0021353401,0.0014845739,0.0063611497,0.006797702,0.0020633051,0.0044004535],"category_scores_gemma":[0.08720197,0.002644193,0.001299412,0.004507081,0.0025184888,0.011829067,0.01082798,0.003796693,0.0028480415],"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":[0.0026445955,0.0011495615,0.019459285,0.00077342737,0.0005615664,0.0011464872,0.0031879263,0.13033369,0.057238586,0.0802378,0.022610925,0.68065614],"study_design_scores_gemma":[0.0009879722,0.0010949888,0.0031058085,0.00023999084,0.00061080174,0.00078011584,0.0023161045,0.6859716,0.09037988,0.16989775,0.044439565,0.00017548697],"about_ca_topic_score_codex":0.0028943,"about_ca_topic_score_gemma":0.0035732652,"teacher_disagreement_score":0.024833618,"about_ca_system_score_codex":0.0010108497,"about_ca_system_score_gemma":0.0065078815,"threshold_uncertainty_score":0.1313343},"labels":[],"label_agreement":null},{"id":"W4379932427","doi":"10.1145/3604437.3604458","title":"Accurate Summary-based Cardinality Estimation Through the Lens of Cardinality Estimation Graphs","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Advanced Database Systems and Queries","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":"Estimator; Cardinality (data modeling); Computer science; Graph; Joins; Proxy (statistics); Context (archaeology); Theoretical computer science; Mathematical optimization; Mathematics; Statistics; Data mining; Machine learning","score_opus":0.0549768818470338,"score_gpt":0.31388724108751587,"score_spread":0.25891035924048206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379932427","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":[0.018265795,0.0005975461,0.9788296,0.00054024864,0.000021513106,0.000025900805,0.00022142188,0.0003176435,0.0011802682],"genre_scores_gemma":[0.64190465,0.0012383953,0.3535456,0.00033230678,0.0002181594,0.00013601709,0.00079058344,0.00029321748,0.0015410616],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99371403,0.002701094,0.00028756532,0.001121107,0.0018196418,0.00035649133],"domain_scores_gemma":[0.9453303,0.0398453,0.005074413,0.006354347,0.0028462757,0.0005493555],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009126582,0.00088211783,0.001346527,0.0028645764,0.0007589264,0.0040199743,0.0024969783,0.001455837,0.0014926142],"category_scores_gemma":[0.0838493,0.00096194627,0.00056610035,0.0032978598,0.0019574985,0.011179333,0.0028350372,0.0027690586,0.0003047623],"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":[0.00023705246,0.000051534764,0.0077653527,0.00024285422,0.00010724819,0.00018250234,0.0006181816,0.34764233,0.0027109173,0.5525037,0.0032472871,0.084690966],"study_design_scores_gemma":[0.000010567747,0.000026294732,0.000741105,0.00003657262,0.000022777687,0.00007431861,0.00007831089,0.7698411,0.0017587516,0.22455211,0.0028349336,0.000023272303],"about_ca_topic_score_codex":0.0031498398,"about_ca_topic_score_gemma":0.0021175474,"teacher_disagreement_score":0.009126582,"about_ca_system_score_codex":0.0024143688,"about_ca_system_score_gemma":0.0013428708,"threshold_uncertainty_score":0.04826659},"labels":[],"label_agreement":null},{"id":"W4388233567","doi":"10.1145/3631504.3631514","title":"Kùzu: A Database Management System For \"Beyond Relational\" Workloads","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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; Relational database; Relational database management system; Set (abstract data type); Database; Relational model; World Wide Web; Programming language","score_opus":0.02996732447430424,"score_gpt":0.25584146233573296,"score_spread":0.22587413786142874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388233567","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":[0.090080395,0.0056681084,0.4617988,0.0066555957,0.0014873623,0.0013335415,0.00920177,0.3705149,0.05325957],"genre_scores_gemma":[0.4250575,0.0052095884,0.47496137,0.0036058212,0.00077068136,0.0011462013,0.02435787,0.012362369,0.052528646],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99902797,0.00015786223,0.00013949693,0.00019030728,0.00031811107,0.0001662635],"domain_scores_gemma":[0.99765587,0.00030691532,0.00015883791,0.0009994588,0.0006059163,0.00027284573],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017600391,0.0007086145,0.0008048905,0.0013742939,0.0011821908,0.002929693,0.0027506629,0.0008713491,0.015912335],"category_scores_gemma":[0.0056450074,0.0009097132,0.00053859333,0.00216756,0.00058117445,0.010479484,0.0032738382,0.0018943686,0.007322221],"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":[0.0018501998,0.0005859873,0.008052633,0.0014547439,0.00024389224,0.0005235547,0.0015147579,0.006197792,0.048028063,0.0558291,0.42660028,0.44911885],"study_design_scores_gemma":[0.000967418,0.0010674449,0.015498092,0.0004583744,0.00040695549,0.0029269282,0.0014088756,0.13664019,0.03814737,0.04990919,0.75200003,0.0005690629],"about_ca_topic_score_codex":0.0029801328,"about_ca_topic_score_gemma":0.0038544624,"teacher_disagreement_score":0.015912335,"about_ca_system_score_codex":0.0008118528,"about_ca_system_score_gemma":0.0020474235,"threshold_uncertainty_score":0.053232014},"labels":[],"label_agreement":null},{"id":"W4391055211","doi":"10.1145/3641832.3641842","title":"Future Database Engine Development: You Will Only Need One Programming Language","year":2023,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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; SPARK (programming language); Database; Java; Programming language; Productivity; Operating system","score_opus":0.021180013553948315,"score_gpt":0.2615194378698532,"score_spread":0.24033942431590488,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391055211","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0085743535,0.0054992656,0.7981849,0.05877003,0.0024775919,0.00033022053,0.0018316751,0.06992202,0.05441001],"genre_scores_gemma":[0.03535736,0.005181353,0.840425,0.018631592,0.0010374383,0.0003165238,0.0040983222,0.012067486,0.082884945],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976376,0.00028619982,0.00016050879,0.0004069068,0.0011933412,0.00031541212],"domain_scores_gemma":[0.9920942,0.0009394995,0.0002985315,0.002088395,0.0031845088,0.0013948835],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007967077,0.0008559296,0.00055204413,0.0005674771,0.0010484613,0.005702648,0.0033822365,0.0017172603,0.0315596],"category_scores_gemma":[0.008460742,0.0011097604,0.0010952485,0.0005820817,0.00082335947,0.015701078,0.0033870572,0.007260653,0.046966616],"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":[0.00029523094,0.00037328093,0.0034359272,0.00065993686,0.000059694827,0.00021860389,0.0008277502,0.00080309366,0.027608367,0.06903489,0.45965368,0.43702954],"study_design_scores_gemma":[0.00007314654,0.00010090537,0.001053973,0.00024974247,0.0000522122,0.00073810003,0.00019857501,0.0039640535,0.010923037,0.03130871,0.95125276,0.000084833184],"about_ca_topic_score_codex":0.0016292382,"about_ca_topic_score_gemma":0.0020605302,"teacher_disagreement_score":0.0315596,"about_ca_system_score_codex":0.0009443751,"about_ca_system_score_gemma":0.003028231,"threshold_uncertainty_score":0.10557741},"labels":[],"label_agreement":null},{"id":"W4396883491","doi":"10.1145/3665252.3665266","title":"Technical Perspective: Synthetic Data Needs a Reproducibility Benchmark","year":2024,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Scientific Computing and Data 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 Waterloo","funders":"","keywords":"Computer science; Benchmark (surveying); Perspective (graphical); Reproducibility; Data mining; Data science; Database; Artificial intelligence; Statistics; Mathematics","score_opus":0.22098713037544487,"score_gpt":0.44072847411390287,"score_spread":0.219741343738458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396883491","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027869834,0.0029228916,0.87527066,0.039002556,0.004691218,0.0011437141,0.009299514,0.004425143,0.03537443],"genre_scores_gemma":[0.42734295,0.0021591885,0.5144117,0.015736395,0.0047881124,0.0029647674,0.021054866,0.00411565,0.0074263774],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8744061,0.07847718,0.0070812358,0.011495867,0.026680768,0.0018588277],"domain_scores_gemma":[0.4617153,0.291689,0.014152699,0.16034189,0.06737428,0.0047269203],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.15268917,0.0017914249,0.0023078893,0.0041190404,0.0026215394,0.011407062,0.0065284315,0.005589562,0.011363756],"category_scores_gemma":[0.49222928,0.0010460742,0.0019248055,0.0073239533,0.005634068,0.015636316,0.00788222,0.0066472813,0.0056608072],"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":[0.0021337008,0.0012180519,0.037501086,0.0030371689,0.0010131139,0.0011469147,0.002084033,0.13002078,0.014470075,0.40004453,0.14361712,0.26371345],"study_design_scores_gemma":[0.0002865542,0.0010236565,0.007141934,0.0015764284,0.00022400086,0.0019421363,0.0017429448,0.20229933,0.014428466,0.6251168,0.14397244,0.00024524285],"about_ca_topic_score_codex":0.004236171,"about_ca_topic_score_gemma":0.0023716807,"teacher_disagreement_score":0.84731084,"about_ca_system_score_codex":0.0032580567,"about_ca_system_score_gemma":0.0070818285,"threshold_uncertainty_score":0.80750716},"labels":[],"label_agreement":null},{"id":"W4406379941","doi":"10.1145/3712311.3712323","title":"A Roadmap to Graph Analytics","year":2025,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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; Analytics; Data science; Graph; Data analysis; Product (mathematics); Task (project management); Data mining; Theoretical computer science; Engineering; Systems engineering","score_opus":0.013973507860048258,"score_gpt":0.2632285770793657,"score_spread":0.24925506921931745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406379941","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":[0.0048736385,0.1525407,0.66577977,0.11891862,0.0069260015,0.00046377405,0.0051656556,0.0074134246,0.037918407],"genre_scores_gemma":[0.08245387,0.17539732,0.6817009,0.023486003,0.012570184,0.00070892187,0.010366917,0.0023905633,0.010925347],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9933901,0.0026119228,0.0002968734,0.0012151678,0.002126935,0.00035899354],"domain_scores_gemma":[0.9470007,0.037106264,0.0007934028,0.0067096506,0.0065409252,0.0018490549],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008150072,0.002454279,0.0031388095,0.0071502863,0.0013891722,0.00849444,0.0061319745,0.0049211676,0.020377524],"category_scores_gemma":[0.03356067,0.0012971739,0.003171891,0.008868039,0.0045065163,0.025280079,0.006420058,0.010349738,0.009524294],"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":[0.0001727639,0.00028350102,0.0017086641,0.002404097,0.00024567795,0.00015308194,0.00038090342,0.013555537,0.0007621523,0.52482605,0.17305407,0.28245345],"study_design_scores_gemma":[0.000023983719,0.000054405904,0.00047977595,0.00044614155,0.000029562238,0.000101585465,0.00023563937,0.025758311,0.0002258258,0.80530286,0.16729821,0.000043664622],"about_ca_topic_score_codex":0.0055818255,"about_ca_topic_score_gemma":0.0039293296,"teacher_disagreement_score":0.020377524,"about_ca_system_score_codex":0.003148708,"about_ca_system_score_gemma":0.0035848524,"threshold_uncertainty_score":0.068169594},"labels":[],"label_agreement":null},{"id":"W4406379983","doi":"10.1145/3712311.3712321","title":"Multi-Analyst Differential Privacy with Fine-Grained Provenance for Databases","year":2025,"lang":"en","type":"article","venue":"ACM SIGMOD Record","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":"Differential privacy; Privilege (computing); Computer science; Internet privacy; Safeguarding; Information privacy; Personally identifiable information; Computer security; Data mining","score_opus":0.0500052720731431,"score_gpt":0.3125504361115713,"score_spread":0.2625451640384282,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406379983","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":[0.009308852,0.0010920101,0.98308855,0.0029824912,0.00012243429,0.00016652903,0.00047456537,0.00042768294,0.0023369177],"genre_scores_gemma":[0.65993685,0.0014240264,0.33197823,0.00096776226,0.00067863235,0.00027982853,0.0009791597,0.00017802768,0.0035775308],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97385424,0.009674724,0.0019416307,0.0049278475,0.008436843,0.0011647645],"domain_scores_gemma":[0.9165019,0.03970913,0.0060643367,0.029486058,0.00646981,0.0017687974],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.021367053,0.00080885284,0.0015736124,0.001672185,0.0021530145,0.006329953,0.003251729,0.0026272498,0.0024341086],"category_scores_gemma":[0.07520528,0.0010660442,0.0017042697,0.0033642217,0.004567609,0.016941544,0.009481632,0.005027968,0.0005558083],"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":[0.00062075385,0.00014949171,0.0052053435,0.00059102627,0.0001774498,0.0004418619,0.0010631104,0.07006955,0.0050404957,0.807133,0.005441833,0.10406607],"study_design_scores_gemma":[0.000055710203,0.00011717827,0.0005335805,0.00009304965,0.000072454975,0.00053516135,0.00012747681,0.18061934,0.0040537766,0.80268013,0.011068969,0.000043098145],"about_ca_topic_score_codex":0.0017905666,"about_ca_topic_score_gemma":0.0015222816,"teacher_disagreement_score":0.021367053,"about_ca_system_score_codex":0.0031548773,"about_ca_system_score_gemma":0.0039610774,"threshold_uncertainty_score":0.11300117},"labels":[],"label_agreement":null},{"id":"W4409918354","doi":"10.1145/3733620.3733635","title":"Reservoir Sampling over Joins","year":2025,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Reservoir Engineering and Simulation Methods","field":"Engineering","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":"Joins; Sampling (signal processing); Computer science; Programming language; Computer vision","score_opus":0.03661353942592897,"score_gpt":0.3200537669503436,"score_spread":0.2834402275244146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409918354","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":[0.01697906,0.00028601836,0.9799485,0.00019528226,0.000049857015,0.00007180731,0.00016201082,0.00092535315,0.0013820785],"genre_scores_gemma":[0.48539197,0.00056655554,0.5078584,0.00030891373,0.00017877114,0.00028444277,0.001135211,0.00036910374,0.0039065685],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998052,0.00047204804,0.00012813014,0.0004916806,0.0006492654,0.00020687273],"domain_scores_gemma":[0.99442935,0.0030355128,0.0003819928,0.0012341867,0.00066308776,0.00025585163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025683271,0.00069650274,0.0015602836,0.00086020044,0.0011501231,0.002114881,0.002427608,0.001020494,0.002418376],"category_scores_gemma":[0.011832937,0.0005465624,0.0010032232,0.0017880107,0.0013748104,0.005193471,0.0029242896,0.001560326,0.0005108699],"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":[0.00054887845,0.00016194549,0.0054669846,0.00026690747,0.00013437112,0.00029913094,0.00032743352,0.66906196,0.013916345,0.13632993,0.007274947,0.1662111],"study_design_scores_gemma":[0.000014637631,0.000040481256,0.0001629353,0.00000699477,0.000011513437,0.000059755574,0.000030859723,0.9653066,0.003156757,0.029259648,0.0019397436,0.000010099508],"about_ca_topic_score_codex":0.004282934,"about_ca_topic_score_gemma":0.004612875,"teacher_disagreement_score":0.004282934,"about_ca_system_score_codex":0.0010896854,"about_ca_system_score_gemma":0.0018120835,"threshold_uncertainty_score":0.013582826},"labels":[],"label_agreement":null},{"id":"W4412402624","doi":"10.1145/3749116.3749124","title":"Reminiscences on Influential Papers","year":2025,"lang":"en","type":"article","venue":"ACM SIGMOD Record","topic":"Data Quality and 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":"University of Toronto","funders":"","keywords":"Library science; Computer science","score_opus":0.12901287901589728,"score_gpt":0.439895915516474,"score_spread":0.3108830365005767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412402624","genre_codex":"editorial","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005165441,0.027086008,0.0005240776,0.26941237,0.68992203,0.000024578112,0.00034772418,0.00015776634,0.012008882],"genre_scores_gemma":[0.010675368,0.028647874,0.000665988,0.20398211,0.6858434,0.00010590806,0.0005440365,0.00074754684,0.06878771],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9866274,0.0028050735,0.00089343486,0.001874378,0.006215362,0.0015843437],"domain_scores_gemma":[0.9427771,0.014986353,0.003229316,0.0035339943,0.02289605,0.0125771165],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014682682,0.002185228,0.0015420105,0.0074329693,0.0071141804,0.020329736,0.0036204983,0.009311438,0.028519727],"category_scores_gemma":[0.07773223,0.0007191594,0.0024038178,0.008096801,0.0030637705,0.011638043,0.0070827105,0.020955391,0.023077236],"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":[0.00001775556,0.000012773846,0.000089454814,0.00011065427,0.000016844617,0.0001281919,0.00027884988,0.000019810459,0.00005320147,0.0024559407,0.98920166,0.0076147295],"study_design_scores_gemma":[0.0000076681345,0.000007810088,0.00012856002,0.00021145913,0.00001362817,0.000116782605,0.00032607635,0.000017444981,0.000070746435,0.0015700468,0.99751544,0.0000143580755],"about_ca_topic_score_codex":0.0018161469,"about_ca_topic_score_gemma":0.003478072,"teacher_disagreement_score":0.028519727,"about_ca_system_score_codex":0.0056923,"about_ca_system_score_gemma":0.004240258,"threshold_uncertainty_score":0.09540796},"labels":[],"label_agreement":null}]}