{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":70,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":70,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"d932170372cf","filters":{"venue":"Text REtrieval Conference"}},"results":[{"id":"W2294145134","doi":"","title":"Overview of the TREC 2009 Web Track","year":2009,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":94,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Information retrieval; Relevance (law); Task (project management); World Wide Web; Clef; Session (web analytics); Web page; Set (abstract data type); Search engine indexing; Query expansion","authors":[{"name":"Charles L. A. Clarke","is_ca":true},{"name":"Nick Craswell","is_ca":false},{"name":"Ellen M. Voorhees","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07088538982854531,"gpt":0.3088896404462865,"spread":0.2380042506177412,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01084585,0.002111278,0.0017984,0.01376568,0.003691074,0.005978091,0.003287005,0.002587422,0.03598236],"category_scores_gemma":[0.01114587,0.001046202,0.001402201,0.01323636,0.0006920851,0.00550031,0.002230917,0.002426051,0.04407853],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007895012,"about_ca_system_score_gemma":0.01194692,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.104434,"about_ca_topic_score_gemma":0.1276292,"domain_scores_codex":[0.9897261,0.00151706,0.0009223312,0.001416161,0.005186392,0.001231995],"domain_scores_gemma":[0.9841091,0.001146245,0.0007717604,0.001581259,0.01056612,0.001825458],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002312826,0.0002850092,0.001323836,0.0007364437,0.0000347419,0.00004763982,0.00006785495,0.001031731,0.002935444,0.001464586,0.8869195,0.1049219],"study_design_scores_gemma":[0.000106734,0.0003522491,0.01120307,0.0004064845,0.00006408167,0.0001556057,0.000131474,0.002991279,0.003582693,0.001476181,0.979398,0.0001321963],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.02295731,0.03803134,0.03871042,0.01661871,0.009702956,0.01243874,0.4955842,0.03443097,0.3315253],"genre_scores_gemma":[0.03645565,0.01438154,0.05919104,0.004754162,0.00295004,0.006579652,0.7741999,0.003646337,0.09784171],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.104434,"threshold_uncertainty_score":0.2076523,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W305879088","doi":"","title":"Domain-specific synonym expansion and validation for biomedical information retrieval (multitext experiments for trec 2004)","year":2004,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":45,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Synonym (taxonomy); Information retrieval; Domain (mathematical analysis); Natural language processing; Artificial intelligence; Biology; Mathematics","authors":[{"name":"Stefan Büttcher","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":false},{"name":"Gordon V. Cormack","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03568668182173386,"gpt":0.3012554734088946,"spread":0.2655687915871607,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0209553,0.001542624,0.001253281,0.003330541,0.002815798,0.001598169,0.001741625,0.002140542,0.004920068],"category_scores_gemma":[0.0456827,0.0007402484,0.001496863,0.002813688,0.001357674,0.005033787,0.002739514,0.002452417,0.00332447],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00164042,"about_ca_system_score_gemma":0.003035448,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01770778,"about_ca_topic_score_gemma":0.01839791,"domain_scores_codex":[0.9812576,0.01281798,0.001501103,0.0013981,0.002475138,0.0005500527],"domain_scores_gemma":[0.952808,0.03028058,0.001346943,0.006546598,0.007842772,0.001175132],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.009762364,0.007516037,0.02382713,0.002953064,0.001387723,0.001100638,0.003097933,0.02118467,0.1436702,0.004424032,0.2849147,0.4961614],"study_design_scores_gemma":[0.005341374,0.007280997,0.1153985,0.0003521602,0.002494431,0.002911069,0.003691564,0.3371896,0.4283665,0.00904293,0.08718477,0.0007459542],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8750139,0.003602772,0.06394918,0.003172208,0.001272297,0.001874725,0.0329417,0.009107155,0.00906605],"genre_scores_gemma":[0.6908451,0.0009051308,0.1938444,0.0007658031,0.0002264772,0.001525839,0.0998468,0.001376118,0.01066439],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.0209553,"threshold_uncertainty_score":0.1108235,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2604321711","doi":"","title":"TREC 2016 Total Recall Track Overview.","year":2016,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Public Relations and Crisis Communication","field":"Social Sciences","cited_by":29,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Track (disk drive); Recall; Information retrieval; Psychology","authors":[{"name":"Maura R. Grossman","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true},{"name":"Adam Roegiest","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07651398741119439,"gpt":0.3523979570630005,"spread":0.2758839696518061,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0159358,0.007558414,0.004682205,0.02631256,0.004937815,0.008747038,0.008016528,0.004201903,0.04953378],"category_scores_gemma":[0.02351712,0.002060058,0.003360318,0.01651033,0.001282488,0.008481686,0.004772391,0.004178125,0.06721579],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006381209,"about_ca_system_score_gemma":0.01544675,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1286107,"about_ca_topic_score_gemma":0.1692612,"domain_scores_codex":[0.9908072,0.001500592,0.001109514,0.001324949,0.004173547,0.001084161],"domain_scores_gemma":[0.9704561,0.003350082,0.002139282,0.003263534,0.01851459,0.002276417],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002230441,0.00008068767,0.0005971065,0.001308099,0.0001132317,0.00003364811,0.00002964416,0.0005077029,0.001593109,0.0002664422,0.9637483,0.03149892],"study_design_scores_gemma":[0.0006211232,0.0007913409,0.01493429,0.001827919,0.0008913481,0.0005255244,0.0002799914,0.007679473,0.02003321,0.003532581,0.9483554,0.000527849],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.009767732,0.04566389,0.01829249,0.006401176,0.01142607,0.002487053,0.8138567,0.03699686,0.05510803],"genre_scores_gemma":[0.008311318,0.007604382,0.01516404,0.001337101,0.001612427,0.001877229,0.8990116,0.002539326,0.06254254],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1286107,"threshold_uncertainty_score":0.2557243,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3013164143","doi":"","title":"TREC 2015 Total Recall Track Overview.","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Track (disk drive); Computer science; Recall; Information retrieval; Artificial intelligence; Cognitive psychology; Psychology; Operating system","authors":[{"name":"Adam Roegiest","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true},{"name":"Maura R. Grossman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1064969915600995,"gpt":0.3029353534591744,"spread":0.1964383618990748,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01790139,0.00523095,0.00347806,0.02861147,0.004191015,0.00729555,0.008120014,0.002751923,0.04064232],"category_scores_gemma":[0.02478128,0.001558657,0.002809502,0.01641832,0.001233898,0.007397515,0.004172078,0.00304176,0.04677663],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007479019,"about_ca_system_score_gemma":0.01651502,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1635305,"about_ca_topic_score_gemma":0.202266,"domain_scores_codex":[0.9906389,0.001465594,0.001011395,0.001094683,0.004820417,0.0009690836],"domain_scores_gemma":[0.9587126,0.003149014,0.002600292,0.003435658,0.02958438,0.002518144],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001531379,0.00005926038,0.0008892295,0.000903827,0.0001044729,0.0000197795,0.00002661602,0.000513555,0.0009720424,0.000407827,0.9574878,0.03846243],"study_design_scores_gemma":[0.0002942349,0.0006786833,0.01819615,0.001717911,0.000854241,0.00036837,0.0002188905,0.005425792,0.01562822,0.003365609,0.9528572,0.0003946651],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.0119774,0.07126279,0.03315924,0.009422527,0.0152914,0.002859848,0.7102184,0.04199427,0.1038141],"genre_scores_gemma":[0.02007108,0.01647019,0.02733223,0.002917455,0.00274882,0.002234466,0.8159974,0.003530289,0.1086981],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1635305,"threshold_uncertainty_score":0.3251573,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3013426078","doi":"","title":"Overview of the TREC 2010 Legal Track.","year":2010,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Law, logistics, and international trade","field":"Business, Management and Accounting","cited_by":22,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Track (disk drive); Information retrieval; Data science","authors":[{"name":"Gordon V. Cormack","is_ca":true},{"name":"Maura R. Grossman","is_ca":true},{"name":"Bruce Hedin","is_ca":false},{"name":"Douglas W. Oard","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07474354518727247,"gpt":0.2791428321533284,"spread":0.2043992869660559,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01107644,0.001898623,0.001721076,0.02540879,0.00355488,0.008770743,0.00397326,0.002588564,0.09720081],"category_scores_gemma":[0.01424522,0.0009123529,0.001125856,0.01921236,0.001046457,0.008428385,0.002866226,0.002770861,0.0850139],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007164443,"about_ca_system_score_gemma":0.01813197,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.11759,"about_ca_topic_score_gemma":0.2054895,"domain_scores_codex":[0.9942657,0.0009594034,0.0004367621,0.000480566,0.003225714,0.0006319186],"domain_scores_gemma":[0.9753774,0.00198047,0.001347151,0.001338141,0.01700047,0.002956365],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002642836,0.00004830249,0.0001194598,0.0003732257,0.000008352518,0.00001692496,0.00002047346,0.0001590319,0.0006462284,0.0007913513,0.9579852,0.0398051],"study_design_scores_gemma":[0.00004683541,0.00007045326,0.00290098,0.0005582118,0.00003692454,0.00008240231,0.00009615595,0.0009215798,0.001640822,0.002362899,0.991209,0.00007371049],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.005964506,0.08825539,0.02989885,0.04914515,0.02905602,0.005931933,0.342624,0.02433183,0.4247924],"genre_scores_gemma":[0.01495767,0.04799237,0.05408115,0.01052067,0.008909477,0.004095544,0.4894317,0.005287999,0.3647234],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.11759,"threshold_uncertainty_score":0.3251691,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1878790362","doi":"10.3390/ijms232113127","title":"Overview of the TREC 2012 Contextual Suggestion Track.","year":2012,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Track (disk drive); Computer science; Context (archaeology); Information retrieval; Task (project management); Set (abstract data type); World Wide Web; Data science; Geography; Engineering","authors":[{"name":"Adriel Dean-Hall","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true},{"name":"Jaap Kamps","is_ca":false},{"name":"Paul Thomas","is_ca":false},{"name":"Ellen M Voorhes","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09325439221598353,"gpt":0.3162031300342015,"spread":0.2229487378182179,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00602668,0.0009615357,0.001038782,0.005878786,0.002497256,0.007683615,0.003337694,0.003776979,0.4829487],"category_scores_gemma":[0.01748159,0.0004941833,0.001130247,0.006058857,0.0007175971,0.005218913,0.005706505,0.003392581,0.3447413],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00287945,"about_ca_system_score_gemma":0.01368518,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008686122,"about_ca_topic_score_gemma":0.01840671,"domain_scores_codex":[0.9970899,0.0004217595,0.0002834102,0.0003605199,0.001447226,0.0003971794],"domain_scores_gemma":[0.9850134,0.001158469,0.0007127819,0.0008333559,0.00763557,0.004646414],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000343052,0.00002573007,0.0001700998,0.0003759303,0.000003665739,0.00003431263,0.0000285666,0.00004081648,0.0001297848,0.001788582,0.9035607,0.09380762],"study_design_scores_gemma":[0.000002921957,0.00000907194,0.0001953843,0.0001907089,0.000002208353,0.0000245699,0.00001935067,0.00001396266,0.0000365095,0.0002701392,0.9992304,0.000004870734],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"methods","genre_scores_codex":[0.001226862,0.03003542,0.009792499,0.08924679,0.1128047,0.001690204,0.0449144,0.0102738,0.7000154],"genre_scores_gemma":[0.005274868,0.03588816,0.01375916,0.0241061,0.02471586,0.001488675,0.0541854,0.004052231,0.8365296],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.4829487,"threshold_uncertainty_score":0.7375109,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2887055502","doi":"","title":"Overview of the TREC 2016 Real-Time Summarization Track.","year":2016,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Automatic summarization; Computer science; Track (disk drive); Information retrieval; Multi-document summarization; Natural language processing; Operating system","authors":[{"name":"Jimmy Lin","is_ca":true},{"name":"Salman Mohammed","is_ca":true},{"name":"Royal Sequiera","is_ca":true},{"name":"Luchen Tan","is_ca":true},{"name":"Nimesh Ghelani","is_ca":true},{"name":"Mustafa Abualsaud","is_ca":true},{"name":"Richard McCreadie","is_ca":false},{"name":"Dmitrijs Milajevs","is_ca":false},{"name":"Ellen M. Voorhees","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0542293694693188,"gpt":0.2754098846886667,"spread":0.2211805152193479,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009083415,0.003022748,0.002312666,0.01067446,0.002199662,0.004712859,0.003610437,0.001891015,0.02988203],"category_scores_gemma":[0.009235329,0.001110284,0.001661175,0.009430257,0.0004582205,0.005461639,0.001939816,0.002588303,0.03888039],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002885099,"about_ca_system_score_gemma":0.007480595,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04182118,"about_ca_topic_score_gemma":0.07434688,"domain_scores_codex":[0.9962171,0.0008784421,0.0004455797,0.0006383462,0.001415631,0.0004049813],"domain_scores_gemma":[0.9875771,0.001404049,0.0006893589,0.001222476,0.00797109,0.001136],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002795532,0.0002367126,0.0006854751,0.001803721,0.0001474263,0.00008256924,0.0001202449,0.00174804,0.01345081,0.0008327074,0.7911445,0.1894682],"study_design_scores_gemma":[0.0002384635,0.0008618749,0.008956989,0.0007117036,0.0004110828,0.0003490705,0.0002699152,0.01500544,0.02822268,0.003263327,0.9414247,0.0002847515],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.02211154,0.1127665,0.2366928,0.01815191,0.01756125,0.007845086,0.3472359,0.1305811,0.1070541],"genre_scores_gemma":[0.02605788,0.02215855,0.1819359,0.002472376,0.003704,0.003293958,0.6658248,0.005916283,0.08863618],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.04182118,"threshold_uncertainty_score":0.09996539,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2395887233","doi":"","title":"University of Waterloo at TREC 2015 Microblog Track","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Microblogging; Exploit; Social media; Information retrieval; Relevance (law); Word (group theory); Query expansion; World Wide Web; Mathematics","authors":[{"name":"Luchen Tan","is_ca":true},{"name":"Adam Roegiest","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0386870167005127,"gpt":0.2642413455623248,"spread":0.2255543288618121,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005637045,0.00251254,0.002271261,0.005779509,0.004492575,0.005784139,0.002549804,0.001869338,0.0838113],"category_scores_gemma":[0.01237456,0.001227766,0.00071723,0.005595828,0.001108484,0.006609804,0.002053061,0.002493931,0.05841029],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01115282,"about_ca_system_score_gemma":0.01405893,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.4736371,"about_ca_topic_score_gemma":0.6227235,"domain_scores_codex":[0.9957131,0.0008190097,0.0002593474,0.0009144402,0.001798771,0.0004953261],"domain_scores_gemma":[0.9900215,0.001518137,0.0003365645,0.001449693,0.005340749,0.00133327],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000712699,0.00008953497,0.0004119163,0.0002512846,0.00002077151,0.00002778485,0.00006082977,0.000390792,0.0006510253,0.0007795286,0.976629,0.02061644],"study_design_scores_gemma":[0.0003085034,0.0001435917,0.007259942,0.0002870965,0.00005167482,0.00007500294,0.0003645313,0.01617714,0.00377474,0.003437496,0.9680072,0.000113003],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.01467896,0.009080539,0.01293859,0.0117364,0.002935725,0.002457402,0.7627143,0.02892181,0.1545363],"genre_scores_gemma":[0.02601984,0.002701474,0.02079433,0.001086059,0.0003996948,0.0009350247,0.8446509,0.00162089,0.1017917],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.4736371,"threshold_uncertainty_score":0.9417605,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2402510845","doi":"","title":"Overview of the TREC 2011 Legal Track.","year":2011,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Task (project management); Computer science; Track (disk drive); Rank (graph theory); Information retrieval; Natural language processing; Artificial intelligence","authors":[{"name":"Maura R. Grossman","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true},{"name":"Bruce Hedin","is_ca":false},{"name":"Douglas W. Oard","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1480039556696987,"gpt":0.2856423647114191,"spread":0.1376384090417204,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01058371,0.001729413,0.001084482,0.008979253,0.0020987,0.004006035,0.002149395,0.001647073,0.1078463],"category_scores_gemma":[0.01535853,0.0008581419,0.0008404023,0.009578086,0.000501222,0.004801618,0.001641523,0.001767488,0.09961408],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004322385,"about_ca_system_score_gemma":0.009431419,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.08846956,"about_ca_topic_score_gemma":0.1153503,"domain_scores_codex":[0.9934657,0.00160429,0.0004904693,0.0007366166,0.003065028,0.0006379708],"domain_scores_gemma":[0.9818457,0.002478311,0.0008820911,0.001895813,0.01119321,0.001704841],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007172587,0.00007410086,0.0003250593,0.0003957834,0.000013022,0.00001588921,0.00004072767,0.0004922892,0.001152587,0.0007311499,0.9594452,0.03724253],"study_design_scores_gemma":[0.0001116451,0.0001631582,0.007964171,0.0002886725,0.00005331498,0.0001093185,0.0001634701,0.002459039,0.003138117,0.001927181,0.9835012,0.0001207437],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.007077255,0.01121845,0.024274,0.009921955,0.004335212,0.005591971,0.7396628,0.02393619,0.1739822],"genre_scores_gemma":[0.0188308,0.004652849,0.03859375,0.002108286,0.001246952,0.003746816,0.8229178,0.002837213,0.1050655],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.1078463,"threshold_uncertainty_score":0.3607819,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2121587139","doi":"","title":"York University at TREC 2009: Relevance Feedback Track","year":2009,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Relevance feedback; Relevance (law); Computer science; Weighting; Task (project management); Track (disk drive); Domain (mathematical analysis); Information retrieval; Series (stratigraphy); Artificial intelligence; Mathematics; Engineering; Image retrieval","authors":[{"name":"Zheng Ye","is_ca":true},{"name":"Jimmy Xiangji Huang","is_ca":true},{"name":"Ben He","is_ca":true},{"name":"Hongfei Lin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03342038611951475,"gpt":0.2467342868726649,"spread":0.2133139007531501,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0189547,0.002167927,0.00240281,0.003347807,0.004405282,0.002766868,0.003120235,0.002573419,0.02452498],"category_scores_gemma":[0.02654865,0.001009685,0.00078451,0.002966581,0.0008509378,0.004310098,0.001853188,0.004431332,0.02435691],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00442891,"about_ca_system_score_gemma":0.007254315,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1147386,"about_ca_topic_score_gemma":0.1680787,"domain_scores_codex":[0.9909912,0.003162188,0.0005628322,0.001157213,0.003500853,0.0006257686],"domain_scores_gemma":[0.9733786,0.004607114,0.0006876253,0.003951746,0.01515901,0.002215877],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007287416,0.001454767,0.001447517,0.00045324,0.0001054951,0.00007205534,0.0001647705,0.002374944,0.008225492,0.0008841908,0.9236892,0.0603997],"study_design_scores_gemma":[0.004824075,0.004532778,0.04473497,0.0003018593,0.0003734645,0.0004969133,0.000539904,0.0928252,0.06401194,0.005096456,0.7814369,0.0008255605],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.2025152,0.01350917,0.1023214,0.01792419,0.01224547,0.02988365,0.3752046,0.1114701,0.1349262],"genre_scores_gemma":[0.1365526,0.001943112,0.1421335,0.003308128,0.0009143121,0.00908906,0.5800944,0.003392031,0.1225728],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1147386,"threshold_uncertainty_score":0.2281416,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3013805123","doi":"","title":"Query and Answer Expansion from Conversation History.","year":2019,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Conversation; Computer science; Query expansion; Information retrieval; Linguistics; Philosophy","authors":[{"name":"Jheng-Hong Yang","is_ca":false},{"name":"Sheng-Chieh Lin","is_ca":false},{"name":"Chuan‐Ju Wang","is_ca":false},{"name":"Jimmy Lin","is_ca":true},{"name":"Ming-Feng Tsai","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02976830130571452,"gpt":0.2219263047048665,"spread":0.192158003399152,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002592092,0.001085039,0.001057259,0.004337746,0.0008422394,0.001551812,0.001257562,0.001146098,0.01088092],"category_scores_gemma":[0.01292215,0.0004848553,0.0008092202,0.002302055,0.0003488438,0.003601085,0.001774415,0.001371448,0.006040028],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008105735,"about_ca_system_score_gemma":0.001494318,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008273984,"about_ca_topic_score_gemma":0.01196104,"domain_scores_codex":[0.9974945,0.001148832,0.000143172,0.0005213657,0.0004703082,0.0002218908],"domain_scores_gemma":[0.9942871,0.003980014,0.0001722479,0.0004300662,0.0009028879,0.0002278134],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002424582,0.0005346725,0.006180443,0.0009994353,0.0002746838,0.0003576495,0.001273957,0.01693345,0.03078776,0.008126564,0.09425326,0.8378536],"study_design_scores_gemma":[0.0001816322,0.0003564571,0.008406175,0.0001903106,0.0003635522,0.0003591951,0.0008223002,0.8933335,0.02064158,0.02211376,0.05314454,0.0000869801],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.146845,0.0142875,0.7514319,0.004780795,0.001424513,0.00197897,0.02730376,0.01975204,0.03219563],"genre_scores_gemma":[0.6907883,0.002275292,0.2397438,0.0004901854,0.001383875,0.001001325,0.04329241,0.000900784,0.02012403],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01088092,"threshold_uncertainty_score":0.03640032,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W94318738","doi":"","title":"Synonym-Based Expansion and Boosting-Based Re-Ranking: A Two-phase Approach for Genomic Information Retrieval.","year":2005,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Boosting (machine learning); WordNet; Information retrieval; Artificial intelligence; Ranking (information retrieval); Machine learning","authors":[{"name":"Zhongmin Shi","is_ca":true},{"name":"Baohua Gu","is_ca":true},{"name":"Fred Popowich","is_ca":true},{"name":"Anoop Sarkar","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02894459750801155,"gpt":0.2989211651388505,"spread":0.269976567630839,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005392751,0.00117083,0.001808679,0.007026738,0.001040141,0.001456901,0.001894101,0.001159405,0.002986779],"category_scores_gemma":[0.009490428,0.0005584182,0.001177884,0.00516084,0.0005774494,0.002687557,0.001772463,0.001182711,0.004304155],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006187105,"about_ca_system_score_gemma":0.001305351,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001937428,"about_ca_topic_score_gemma":0.004664458,"domain_scores_codex":[0.9958961,0.001704541,0.0002527652,0.0006063238,0.001316864,0.0002234168],"domain_scores_gemma":[0.995863,0.00128404,0.0003141618,0.000865693,0.001512692,0.0001605117],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003744726,0.0004446272,0.00270353,0.000447616,0.0002130195,0.0002111561,0.0004088,0.009178377,0.05300895,0.005177511,0.02060935,0.9072226],"study_design_scores_gemma":[0.0002983659,0.001238483,0.0167299,0.0001580266,0.0005721902,0.002293222,0.0006190575,0.6996439,0.117108,0.04860067,0.1123248,0.0004135777],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02056872,0.001985152,0.9645828,0.0003393569,0.0002523824,0.0007570464,0.0006477695,0.006997914,0.003868911],"genre_scores_gemma":[0.1104763,0.0006553431,0.8806005,0.0003231873,0.0002267184,0.0004263466,0.002515841,0.0003501831,0.004425517],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007026738,"threshold_uncertainty_score":0.02851993,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3173838993","doi":"","title":"Overview of the TREC 2020 Health Misinformation Track.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Misinformation; Track (disk drive); Computer science; Information retrieval; Data science; Computer security","authors":[{"name":"Charles L. A. Clarke","is_ca":true},{"name":"S. Nasir Aziz Rizvi","is_ca":false},{"name":"Mark D. Smucker","is_ca":true},{"name":"Maria Maistro","is_ca":false},{"name":"Guido Zuccon","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1556287675719629,"gpt":0.3753961097522171,"spread":0.2197673421802542,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01593824,0.001940425,0.00140445,0.02173156,0.002696482,0.008695533,0.002880062,0.002478431,0.05403712],"category_scores_gemma":[0.01326886,0.0006940336,0.001190364,0.01497256,0.0008488221,0.007179159,0.002661017,0.002513537,0.03912666],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007022777,"about_ca_system_score_gemma":0.01817399,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1078189,"about_ca_topic_score_gemma":0.2165781,"domain_scores_codex":[0.9938671,0.001269178,0.0004136892,0.0003670145,0.003446485,0.0006364513],"domain_scores_gemma":[0.9700852,0.003965015,0.001725141,0.001185007,0.01873118,0.004308484],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005902556,0.00006111777,0.0003431033,0.000977296,0.0000265558,0.00002257106,0.00004698717,0.0002962097,0.0009723405,0.0007745069,0.925262,0.07115819],"study_design_scores_gemma":[0.00004943596,0.0001540869,0.004134604,0.0009695716,0.00009366894,0.00009494867,0.0001751856,0.001314565,0.00227573,0.001776225,0.988865,0.00009704526],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"review","genre_scores_codex":[0.008467169,0.1925066,0.02523356,0.1002153,0.03140583,0.005513674,0.3015836,0.01791935,0.3171549],"genre_scores_gemma":[0.02991194,0.1051057,0.06758911,0.02325463,0.01503808,0.003471215,0.4563071,0.003302035,0.2960202],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.1078189,"threshold_uncertainty_score":0.2143826,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2132630274","doi":"","title":"York University at TREC 2012: Medical Records Track","year":2012,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Track (disk drive); Computer science; Information retrieval; Medical record; Work (physics); Artificial intelligence; Medicine; Engineering","authors":[{"name":"Jun Miao","is_ca":true},{"name":"Zheng Ye","is_ca":false},{"name":"Jimmy Xiangji Huang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03818721144490055,"gpt":0.2545070467609054,"spread":0.2163198353160048,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02224003,0.004561493,0.004415937,0.009036218,0.004073566,0.00575712,0.004554925,0.004161833,0.05565176],"category_scores_gemma":[0.0326659,0.001126923,0.001867051,0.008430782,0.001197868,0.006674117,0.003105791,0.004708813,0.04642735],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006942063,"about_ca_system_score_gemma":0.01171594,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1027113,"about_ca_topic_score_gemma":0.1540458,"domain_scores_codex":[0.9862541,0.004077873,0.001257223,0.001399558,0.005976236,0.001035001],"domain_scores_gemma":[0.9615148,0.006752898,0.002158585,0.005678092,0.01964117,0.004254486],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002064366,0.0001683102,0.000442349,0.0005055803,0.00005970492,0.00003418656,0.00002884901,0.0003923673,0.001060528,0.0002931753,0.980867,0.0159415],"study_design_scores_gemma":[0.001630395,0.00104166,0.02624767,0.0006141818,0.0003355977,0.0004396406,0.0003129735,0.01850154,0.009926459,0.003230595,0.937322,0.0003972358],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"methods","genre_scores_codex":[0.01646362,0.01545749,0.01115247,0.0227003,0.008755158,0.003727584,0.846627,0.02893306,0.0461833],"genre_scores_gemma":[0.01209203,0.002769531,0.01424167,0.002364589,0.001188213,0.001627346,0.9285212,0.001002668,0.03619266],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.1027113,"threshold_uncertainty_score":0.2042269,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2885586573","doi":"","title":"MRG_UWaterloo and WaterlooCormack Participation in the TREC 2017 Common Core Track.","year":2017,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Advanced Bandit Algorithms Research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Track (disk drive); Core (optical fiber); Telecommunications; Operating system","authors":[{"name":"Maura R. Grossman","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.354013297376704,"gpt":0.5097599940313238,"spread":0.1557466966546198,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01041401,0.002055707,0.003371284,0.004953814,0.006420569,0.005583501,0.003114104,0.003738261,0.2490491],"category_scores_gemma":[0.01455788,0.0007335806,0.00100699,0.00492352,0.001444096,0.003908163,0.005228227,0.002663233,0.1071184],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01320011,"about_ca_system_score_gemma":0.02151597,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.2834327,"about_ca_topic_score_gemma":0.5706774,"domain_scores_codex":[0.9900663,0.001319249,0.0002196266,0.001231911,0.004837827,0.002325034],"domain_scores_gemma":[0.9700944,0.001132228,0.0003882394,0.001599642,0.01658386,0.01020162],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001429266,0.0000628948,0.0002180447,0.00003975508,0.000007518736,0.00005030178,0.00006582303,0.00007563177,0.001009528,0.001018364,0.9819577,0.01535156],"study_design_scores_gemma":[0.00009848219,0.00005607997,0.001325004,0.00003136766,0.00001350139,0.00003487731,0.0002906611,0.001011627,0.001488631,0.001085191,0.9945303,0.00003424244],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.01678748,0.008997363,0.01932538,0.1729889,0.07243968,0.004417844,0.143698,0.01088666,0.5504586],"genre_scores_gemma":[0.02111732,0.0009416685,0.0093009,0.007186915,0.003343778,0.0004692684,0.04858671,0.001585611,0.9074678],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.2834327,"threshold_uncertainty_score":0.8331523,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W161736330","doi":"","title":"Approaches to High Accuracy Retrieval: Phrase-Based Search Experiments in the HARD Track.","year":2004,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Focus (optics); Phrase; Selection (genetic algorithm); Natural language processing; Word (group theory); Artificial intelligence; Noun phrase; Noun; Information retrieval; Mathematics","authors":[{"name":"Olga Vechtomova","is_ca":true},{"name":"Murat Karamuftuoglu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1651441880432227,"gpt":0.3311107469396994,"spread":0.1659665588964767,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02607267,0.002312389,0.00297154,0.002615442,0.002558791,0.00318797,0.00382225,0.00575083,0.009843691],"category_scores_gemma":[0.1100871,0.001082522,0.001629103,0.003434694,0.001994166,0.008370861,0.003697697,0.005019682,0.003722688],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001774889,"about_ca_system_score_gemma":0.002005849,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0106017,"about_ca_topic_score_gemma":0.009047964,"domain_scores_codex":[0.9774818,0.0141716,0.002023522,0.001671179,0.003814867,0.0008368884],"domain_scores_gemma":[0.8522637,0.1297124,0.002298871,0.008496322,0.004538774,0.002689964],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0617849,0.04685712,0.02785149,0.01622354,0.00447653,0.002369485,0.008600711,0.09867968,0.04193493,0.0100818,0.1054858,0.575654],"study_design_scores_gemma":[0.01808015,0.04322507,0.03478437,0.000553309,0.00196447,0.002539945,0.004663461,0.7953876,0.04086543,0.02873338,0.02843002,0.0007728626],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.897703,0.01035955,0.05544893,0.002715199,0.001152555,0.004998323,0.007413398,0.005371933,0.01483718],"genre_scores_gemma":[0.8256688,0.001573073,0.1422076,0.001437152,0.000602637,0.003095445,0.01543994,0.0009842828,0.008991082],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02607267,"threshold_uncertainty_score":0.1378871,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2885556514","doi":"","title":"UWaterlooMDS at the TREC 2017 Common Core Track.","year":2017,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Track (disk drive); Computer science; Core (optical fiber); Telecommunications; Operating system","authors":[{"name":"Haotian Zhang","is_ca":true},{"name":"Mustafa Abualsaud","is_ca":true},{"name":"Nimesh Ghelani","is_ca":true},{"name":"Angshuman Ghosh","is_ca":false},{"name":"Mark D. Smucker","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true},{"name":"Maura R. Grossman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06543898827919384,"gpt":0.3335348181136187,"spread":0.2680958298344248,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002813818,0.001666888,0.002707295,0.005047725,0.002608035,0.006534286,0.002418604,0.001705443,0.636452],"category_scores_gemma":[0.005871141,0.0008303203,0.0007212191,0.007028039,0.0006837763,0.00636939,0.003411903,0.002013799,0.4508407],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00351377,"about_ca_system_score_gemma":0.005592791,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.07814548,"about_ca_topic_score_gemma":0.2151276,"domain_scores_codex":[0.998261,0.000247145,0.00008696298,0.0003259145,0.0008501312,0.0002287765],"domain_scores_gemma":[0.9948584,0.0004360389,0.0001254693,0.0005615867,0.002478758,0.001539794],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00002566561,0.00002829287,0.00004332743,0.00004876966,0.000003015682,0.000009653846,0.00001019777,0.00003185324,0.0002666159,0.0004243945,0.9867306,0.01237769],"study_design_scores_gemma":[0.00004370348,0.00002608922,0.0008075878,0.00007192578,0.000007761277,0.00002463294,0.00008208418,0.0005169187,0.0006108429,0.002451681,0.9953354,0.00002139208],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.003226726,0.006974822,0.01267128,0.0169588,0.01479909,0.001011148,0.3481355,0.02509473,0.5711279],"genre_scores_gemma":[0.005177018,0.002569291,0.008296989,0.001892977,0.001568531,0.0002717434,0.177854,0.004037347,0.798332],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.636452,"threshold_uncertainty_score":0.5185571,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2145841197","doi":"","title":"Selecting versus Describing: A Preliminary Analysis of the Efficacy of Categories in Exploring the Web","year":2001,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Task (project management); Session (web analytics); Directory; Computer science; Information retrieval; Point (geometry); Rating scale; Certainty; Scale (ratio); Process (computing); Psychology; World Wide Web; Mathematics","authors":[{"name":"E. G. Torns","is_ca":true},{"name":"Rick Kopak","is_ca":false},{"name":"Joan C. Bartlett","is_ca":false},{"name":"Luanne Freund","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1460100374333013,"gpt":0.3010006766775535,"spread":0.1549906392442522,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02029542,0.0007963474,0.0009951589,0.002086711,0.00098515,0.002720416,0.001072729,0.001319645,0.0077946],"category_scores_gemma":[0.1350298,0.0006615304,0.001383892,0.0009927006,0.001571931,0.002776963,0.001911685,0.001648753,0.001460176],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007884593,"about_ca_system_score_gemma":0.0008801713,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001914867,"about_ca_topic_score_gemma":0.002160346,"domain_scores_codex":[0.9873676,0.006303018,0.002022405,0.0009969643,0.002203785,0.00110631],"domain_scores_gemma":[0.6617123,0.2984511,0.01346678,0.007973975,0.01479145,0.003604348],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.02346682,0.005915126,0.6864419,0.003343509,0.001035884,0.0006009404,0.1064742,0.001759408,0.02143824,0.001403093,0.005490336,0.1426305],"study_design_scores_gemma":[0.0003616194,0.00687166,0.9470105,0.0002611509,0.0006299578,0.0002730175,0.0304024,0.002908299,0.006418208,0.0008036398,0.003891455,0.0001682561],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9913037,0.0001588207,0.0008631539,0.0001109835,0.00002087005,0.0004468313,0.000474523,0.0000810964,0.00653993],"genre_scores_gemma":[0.9938188,0.0001225275,0.001708228,0.0000733509,0.00003111535,0.00110816,0.0006381111,0.0001051768,0.002394403],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02029542,"threshold_uncertainty_score":0.1073337,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W413120383","doi":"","title":"Frequent Itemset Mining for Query Expansion in Microblog Ad-hoc Search","year":2012,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Microblogging; Computer science; Social media; Data mining; Information retrieval; Process (computing); Volume (thermodynamics); World Wide Web","authors":[{"name":"Younos Aboulnaga","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07056403817857194,"gpt":0.3173098186451487,"spread":0.2467457804665768,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002871201,0.0006548507,0.001509589,0.004269013,0.0007866964,0.001071705,0.001389954,0.000741164,0.001156915],"category_scores_gemma":[0.01008817,0.0003727048,0.000727421,0.004361131,0.0003056076,0.001505078,0.0006418093,0.0008002723,0.0006831712],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007711084,"about_ca_system_score_gemma":0.001055804,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005292965,"about_ca_topic_score_gemma":0.004566489,"domain_scores_codex":[0.9982621,0.0005893293,0.0002468042,0.000260102,0.0005069486,0.0001347468],"domain_scores_gemma":[0.9916207,0.005738162,0.0006237926,0.0005853488,0.001160107,0.000271906],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002434334,0.001611368,0.05936617,0.0007651269,0.0005253642,0.001032846,0.0005840472,0.1036664,0.02806484,0.003543699,0.01014927,0.7882566],"study_design_scores_gemma":[0.00005842866,0.0002375908,0.00579174,0.00001843411,0.00006174045,0.0003729329,0.000203165,0.9815336,0.008045097,0.002651989,0.001005661,0.00001955507],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.6925686,0.002149186,0.2960187,0.0008883944,0.0001546899,0.0006453604,0.001985685,0.003574227,0.002015158],"genre_scores_gemma":[0.7944177,0.0002510266,0.2029929,0.0000857807,0.00006824036,0.0001473401,0.001455557,0.00004068896,0.0005406972],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.005292965,"threshold_uncertainty_score":0.01518452,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3013646476","doi":"","title":"WaterlooClarke at the TREC 2019 Conversational Assistant Track.","year":2019,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Track (disk drive); Artificial intelligence; Natural language processing; World Wide Web; Multimedia; Operating system","authors":[{"name":"Charles L. A. Clarke","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01528003651922833,"gpt":0.2526781807060182,"spread":0.2373981441867898,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004510692,0.001532551,0.001938905,0.001970043,0.003663611,0.005401331,0.001709337,0.001901205,0.3146465],"category_scores_gemma":[0.005313545,0.0006013637,0.0005520565,0.002070989,0.0006956467,0.004983251,0.002422709,0.002481227,0.1532771],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004086189,"about_ca_system_score_gemma":0.00661627,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1240986,"about_ca_topic_score_gemma":0.3946297,"domain_scores_codex":[0.998183,0.000432087,0.00006721081,0.0003603443,0.0006767044,0.0002806717],"domain_scores_gemma":[0.9960487,0.0005342073,0.00007979164,0.0002603163,0.0018218,0.001255195],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00007305862,0.00005271658,0.00006307087,0.00006269767,0.000007070028,0.00002024613,0.00003272738,0.00004011843,0.0008711079,0.0005153955,0.9858078,0.01245415],"study_design_scores_gemma":[0.00009765957,0.00007961549,0.001243385,0.00007472182,0.0000213628,0.00004272155,0.0002485358,0.001382455,0.001967641,0.002002412,0.9927984,0.00004103746],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.01177957,0.01490483,0.05663486,0.0600019,0.04553545,0.003193659,0.2102665,0.02868075,0.5690025],"genre_scores_gemma":[0.01170356,0.002491622,0.01587678,0.003845547,0.002607864,0.0005372119,0.08272737,0.001675164,0.8785349],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.3146465,"threshold_uncertainty_score":0.9775735,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W165780999","doi":"","title":"DalTREC 2005 Spam Track: Spam Filtering using N-gram-based Techniques","year":2005,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Spam and Phishing Detection","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; n-gram; Track (disk drive); Spambot; Computer network; Spamming; World Wide Web; Artificial intelligence; Operating system; The Internet","authors":[{"name":"Vlado Kešelj","is_ca":true},{"name":"Evangelos Milios","is_ca":true},{"name":"Andrew Tuttle","is_ca":true},{"name":"Singer Wang","is_ca":true},{"name":"Roger Zhang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04681525418723725,"gpt":0.2834078574464462,"spread":0.2365926032592089,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004587616,0.003198449,0.003651095,0.008038163,0.002838713,0.00428377,0.003746504,0.004015462,0.01677662],"category_scores_gemma":[0.01272241,0.001149255,0.0009294952,0.00452114,0.0008253717,0.004198157,0.002365871,0.002222673,0.02768152],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001428054,"about_ca_system_score_gemma":0.002533415,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01957658,"about_ca_topic_score_gemma":0.02205114,"domain_scores_codex":[0.9957045,0.00078234,0.0003145399,0.0005543639,0.002289576,0.0003546964],"domain_scores_gemma":[0.9926092,0.001303963,0.0003931403,0.001843098,0.003376203,0.0004743921],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001319403,0.0006466247,0.00205862,0.0007262903,0.0002867913,0.0002292492,0.0001546966,0.003652613,0.01291998,0.001365296,0.8126759,0.1639644],"study_design_scores_gemma":[0.0011969,0.001663249,0.007841817,0.0001682527,0.0004575088,0.0007092048,0.0002085757,0.3826893,0.1170591,0.007337323,0.4802099,0.0004586857],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"software","genre_gemma":"empirical","genre_scores_codex":[0.04498069,0.004158555,0.2066639,0.003058714,0.003243051,0.002557008,0.1359865,0.570745,0.0286066],"genre_scores_gemma":[0.1035892,0.001202684,0.3185437,0.001658392,0.0009354349,0.001484764,0.4991119,0.01296816,0.06050579],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01957658,"threshold_uncertainty_score":0.05612338,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2573384236","doi":"","title":"Waterloo (Cormack) Participation in the TREC 2015 Total Recall Track.","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Recall; Track (disk drive); Computer science; Information retrieval; Artificial intelligence; Cognitive psychology; Psychology","authors":[{"name":"Gordon V. Cormack","is_ca":true},{"name":"Maura R. Grossman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08988319202190065,"gpt":0.3078099765277643,"spread":0.2179267845058637,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01527652,0.001574214,0.001985346,0.005138173,0.004013502,0.005405969,0.003005023,0.001558145,0.1158521],"category_scores_gemma":[0.01451483,0.000574408,0.0008288709,0.004713631,0.00137445,0.002719883,0.003308538,0.001550992,0.04376391],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01314385,"about_ca_system_score_gemma":0.02112383,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.4872061,"about_ca_topic_score_gemma":0.7099874,"domain_scores_codex":[0.9927526,0.001569024,0.0003142912,0.001170028,0.003238274,0.000955828],"domain_scores_gemma":[0.9765185,0.00139305,0.0003783196,0.00172975,0.01701834,0.002962084],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001730713,0.00007674229,0.0004122843,0.0002136523,0.00001982291,0.00002517597,0.0001089574,0.0001780908,0.001751921,0.0008511915,0.9588298,0.03735934],"study_design_scores_gemma":[0.0001324623,0.0001188416,0.005951969,0.0001352615,0.00004841395,0.00005135765,0.0004013225,0.001443875,0.004504443,0.001361607,0.9857976,0.00005282706],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.02922198,0.01632137,0.03222805,0.03930332,0.0153124,0.004886643,0.3560568,0.01400587,0.4926636],"genre_scores_gemma":[0.03626962,0.002294009,0.02023806,0.004572867,0.001112907,0.001067812,0.1857923,0.001956603,0.7466957],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.4872061,"threshold_uncertainty_score":0.9687405,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3175902990","doi":"","title":"H2oloo at TREC 2020: When all you got is a hammer... Deep Learning, Health Misinformation, and Precision Medicine.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Misinformation; Hammer; Computer science; Artificial intelligence; Deep learning; Data science; Engineering; Computer security; Mechanical engineering","authors":[{"name":"Ronak Pradeep","is_ca":true},{"name":"Xueguang Ma","is_ca":true},{"name":"Xinyu Zhang","is_ca":true},{"name":"Hang Cui","is_ca":false},{"name":"Ruizhou Xu","is_ca":false},{"name":"Rodrigo Nogueira","is_ca":true},{"name":"Jimmy Lin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05694637064717668,"gpt":0.2868362349313853,"spread":0.2298898642842086,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01149506,0.002348424,0.001345279,0.002667091,0.002032303,0.005002543,0.002520171,0.004764943,0.04682727],"category_scores_gemma":[0.01750206,0.0004853339,0.001012189,0.00176016,0.001793906,0.004501351,0.003082178,0.004276824,0.01863619],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004015107,"about_ca_system_score_gemma":0.007653739,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0801805,"about_ca_topic_score_gemma":0.2027159,"domain_scores_codex":[0.9969049,0.001304041,0.000121544,0.0002847994,0.001006792,0.0003778348],"domain_scores_gemma":[0.9836233,0.006292259,0.0006661137,0.00100912,0.004869177,0.003539946],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001202866,0.00002732696,0.0002638735,0.0002214678,0.0000293159,0.00001209293,0.0000161301,0.0004036861,0.0002245507,0.0002845316,0.9913121,0.007084661],"study_design_scores_gemma":[0.001433197,0.0005151397,0.0118705,0.001200703,0.000290029,0.0001574872,0.0006167041,0.01612787,0.005188035,0.02174767,0.9405427,0.0003099946],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.01727866,0.03688133,0.01924152,0.171645,0.06058912,0.001493132,0.5981041,0.01110668,0.08366036],"genre_scores_gemma":[0.08486571,0.007941765,0.02941872,0.04291128,0.01551072,0.001127663,0.6851285,0.002914323,0.1301813],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0801805,"threshold_uncertainty_score":0.1594276,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2115057477","doi":"","title":"York University at TREC 2006: Enterprise Email Discussion Search","year":2006,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Computer science; Information retrieval; Thread (computing); Ranking (information retrieval); Search engine; Word (group theory); Document retrieval; Rank (graph theory); Query expansion; World Wide Web","authors":[{"name":"Yu Fan","is_ca":false},{"name":"Jimmy Xiangji Huang","is_ca":true},{"name":"Aijun An","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02514964510748901,"gpt":0.2407946580660498,"spread":0.2156450129585608,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007950339,0.001603596,0.001868663,0.00561936,0.003480827,0.002942711,0.001906436,0.002129144,0.04494273],"category_scores_gemma":[0.01570276,0.0006361064,0.0006544197,0.004326717,0.0005421361,0.004201123,0.001778739,0.002149923,0.02559722],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003605829,"about_ca_system_score_gemma":0.003579342,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05365471,"about_ca_topic_score_gemma":0.08222938,"domain_scores_codex":[0.9956963,0.001629493,0.0004465065,0.000531535,0.001234352,0.0004618245],"domain_scores_gemma":[0.9891411,0.003109242,0.0004747967,0.001440861,0.004506763,0.001327205],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004872643,0.0006787258,0.001600734,0.0007829539,0.00006155766,0.0001073791,0.0003409448,0.00195333,0.005079172,0.001824119,0.8913915,0.09569229],"study_design_scores_gemma":[0.001570473,0.001243763,0.0301419,0.0003644905,0.0002012783,0.0003538704,0.00119057,0.06220024,0.03513097,0.007048899,0.8600481,0.0005055028],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1995817,0.0146013,0.07025097,0.01538992,0.005443461,0.0104899,0.2711028,0.1040812,0.3090588],"genre_scores_gemma":[0.2144273,0.002169886,0.1344256,0.002136337,0.001065423,0.005800012,0.4089664,0.003307949,0.227701],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.05365471,"threshold_uncertainty_score":0.1503484,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2166281503","doi":"","title":"DalTREC 2004: Question Answering using Regular Expression Rewriting","year":2004,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Dalhousie University","funders":"","keywords":"Rewriting; Regular expression; Computer science; Expression (computer science); Question answering; Search engine; Information retrieval; Track (disk drive); Programming language; Natural language processing; World Wide Web; Artificial intelligence; Operating system","authors":[{"name":"Vlado Kešelj","is_ca":true},{"name":"Anthony Cox","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02825001249352575,"gpt":0.2986974757111274,"spread":0.2704474632176017,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005057315,0.001410804,0.001673488,0.001758084,0.0009932478,0.003024858,0.003565608,0.002287433,0.01858698],"category_scores_gemma":[0.01051046,0.000895262,0.001245938,0.001162159,0.001083848,0.004356581,0.002328982,0.003061239,0.01112674],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00208018,"about_ca_system_score_gemma":0.00190752,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0125215,"about_ca_topic_score_gemma":0.01173183,"domain_scores_codex":[0.9944887,0.002135918,0.0003399394,0.001352857,0.001348785,0.0003337977],"domain_scores_gemma":[0.9940136,0.002186389,0.0002003637,0.001783615,0.001572591,0.0002434963],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001484157,0.0009335515,0.001584075,0.001932573,0.0003814906,0.0007738547,0.001271727,0.0179405,0.08060946,0.02193745,0.3731783,0.4979728],"study_design_scores_gemma":[0.001070493,0.0007966018,0.003957591,0.0001442709,0.0002500442,0.001388685,0.0005377437,0.2815157,0.1685955,0.02655242,0.5148561,0.0003349071],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05558458,0.003463844,0.5704606,0.003089212,0.0009544568,0.002345422,0.01587123,0.3080067,0.04022398],"genre_scores_gemma":[0.2739885,0.001241686,0.5707157,0.002241914,0.0004075281,0.001047162,0.08058278,0.01279177,0.05698301],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01858698,"threshold_uncertainty_score":0.06217957,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2947689917","doi":"","title":"KlickLabs at TREC 2018 Precision Medicine track.","year":2018,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Track (disk drive); Precision medicine; Information retrieval; Artificial intelligence; Medicine","authors":[{"name":"Lediona Nishani","is_ca":false},{"name":"Maheedhar Kolla","is_ca":true},{"name":"Gaurav Baruah","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04302524158150629,"gpt":0.3146855815370476,"spread":0.2716603399555413,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009426671,0.004717301,0.003718633,0.01109041,0.003455893,0.008887012,0.00532603,0.003476197,0.1835662],"category_scores_gemma":[0.02271781,0.001075642,0.001986421,0.008555487,0.001077322,0.01119517,0.004880049,0.003768427,0.1829371],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006463628,"about_ca_system_score_gemma":0.008451271,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.07185958,"about_ca_topic_score_gemma":0.1127988,"domain_scores_codex":[0.9941239,0.001200474,0.0004547167,0.001026046,0.002481566,0.0007132689],"domain_scores_gemma":[0.9804698,0.003541829,0.0007079813,0.002460223,0.01031441,0.00250581],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005638383,0.00004581788,0.00008601863,0.0001743187,0.00001608106,0.00001845155,0.00001116886,0.0001175551,0.0002978091,0.0002977946,0.9896768,0.009201858],"study_design_scores_gemma":[0.0004139721,0.0001540069,0.003030906,0.0003752885,0.0001276161,0.0001702516,0.0002420317,0.005810654,0.00442948,0.006621255,0.9785097,0.0001149404],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.004207891,0.01071939,0.0137089,0.02440011,0.01175343,0.001047832,0.8263688,0.04214679,0.06564679],"genre_scores_gemma":[0.005906092,0.0021198,0.01493904,0.002180103,0.001210663,0.0005395524,0.9035081,0.002217618,0.06737916],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1835662,"threshold_uncertainty_score":0.6140901,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2578945560","doi":"","title":"Real Time Filtering of Tweets Using Wikipedia Concepts and Google Tri-gram Semantic Relatedness","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; Information retrieval; n-gram; Semantic similarity; Word (group theory); Social media; World Wide Web; Natural language processing; Language model","authors":[{"name":"Anh Duc Dang","is_ca":true},{"name":"Raheleh Makki","is_ca":true},{"name":"Abidalrahman Moh’d","is_ca":true},{"name":"Aminul Islam","is_ca":true},{"name":"Vlado Kešelj","is_ca":true},{"name":"Evangelos Milios","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07837344284233229,"gpt":0.3094833050650553,"spread":0.231109862222723,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001019977,0.001464406,0.001045533,0.01125887,0.001144069,0.001429148,0.0006428831,0.0008962401,0.001694818],"category_scores_gemma":[0.005173184,0.0002693048,0.001027007,0.005165648,0.0003010747,0.002671326,0.0009942116,0.0007721198,0.002620385],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007016975,"about_ca_system_score_gemma":0.0009506127,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01083807,"about_ca_topic_score_gemma":0.01624328,"domain_scores_codex":[0.9986695,0.0002659899,0.0001376662,0.0003681298,0.0004247991,0.0001339713],"domain_scores_gemma":[0.9982017,0.0005624624,0.0002036956,0.0001599977,0.0007584417,0.0001135743],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002362785,0.00142356,0.02747598,0.00170989,0.0005965793,0.001430453,0.002094018,0.01695572,0.10253,0.005449547,0.0458358,0.7921356],"study_design_scores_gemma":[0.0001348757,0.0009880048,0.04634593,0.0001829496,0.0003468475,0.001177295,0.00245571,0.8449103,0.05352491,0.01233835,0.03733975,0.0002550446],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5646859,0.003639693,0.3740354,0.001137115,0.0009863785,0.001555434,0.01720816,0.02322814,0.01352379],"genre_scores_gemma":[0.6769998,0.0007001322,0.2939443,0.000160422,0.000344566,0.0005274453,0.0233479,0.0004239961,0.003551478],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01125887,"threshold_uncertainty_score":0.02154994,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2604217576","doi":"","title":"Word embeddings and Global Preference for Contextual Suggestion.","year":2016,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":4,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Lakehead University; Université Laval","funders":"","keywords":"Preference; Computer science; Word (group theory); Natural language processing; Artificial intelligence; Speech recognition; Linguistics; Mathematics; Statistics; Philosophy","authors":[{"name":"Jian Mo","is_ca":false},{"name":"Luc Lamontagne","is_ca":true},{"name":"Richard Khoury","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05459598172282607,"gpt":0.3217472922575974,"spread":0.2671513105347713,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001724369,0.0003452922,0.0002971024,0.0009123207,0.0003888021,0.002382191,0.0004426503,0.0008560479,0.008598678],"category_scores_gemma":[0.03225145,0.0002436102,0.0002656697,0.000903325,0.0006479841,0.004293395,0.001065464,0.001203605,0.0009996092],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004091698,"about_ca_system_score_gemma":0.0003564442,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007498011,"about_ca_topic_score_gemma":0.001011368,"domain_scores_codex":[0.9987527,0.0005923565,0.00009228885,0.0002463471,0.0002091145,0.0001072775],"domain_scores_gemma":[0.9878649,0.007458156,0.001528482,0.001092431,0.001385537,0.0006705104],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.007291966,0.001182257,0.136025,0.002196233,0.0007117398,0.0008367235,0.00732712,0.008990762,0.1642216,0.08844464,0.01804306,0.564729],"study_design_scores_gemma":[0.0009900099,0.003413072,0.4308605,0.0006498081,0.001496157,0.003548996,0.01016797,0.1776686,0.04474949,0.2978743,0.02795039,0.0006307028],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9280536,0.002311602,0.03885195,0.001068119,0.0002483979,0.0001136044,0.0005296944,0.0004464721,0.02837654],"genre_scores_gemma":[0.9925956,0.0002314254,0.005284213,0.00008453761,0.00004792555,0.00002400418,0.0003615155,0.00008828781,0.00128253],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.008598678,"threshold_uncertainty_score":0.02876544,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3176967269","doi":"","title":"TREC 2020 Podcasts Track Overview.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Track (disk drive); Computer science; Information retrieval; Operating system","authors":[{"name":"Rosie Jones","is_ca":false},{"name":"Ben Carterette","is_ca":false},{"name":"Ann Clifton","is_ca":true},{"name":"Jussi Karlgren","is_ca":false},{"name":"Aasish Pappu","is_ca":false},{"name":"Sravana Reddy","is_ca":false},{"name":"Yongze Yu","is_ca":false},{"name":"Maria Eskevich","is_ca":false},{"name":"Gareth J. F. Jones","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06050055887350622,"gpt":0.310150801677166,"spread":0.2496502428036598,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0118054,0.004233426,0.002574337,0.01184147,0.003228608,0.009065449,0.005937845,0.002947149,0.1602651],"category_scores_gemma":[0.01458224,0.001438144,0.00179706,0.009026304,0.0009472689,0.006936122,0.004330505,0.003504569,0.207521],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004519215,"about_ca_system_score_gemma":0.01701537,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1191248,"about_ca_topic_score_gemma":0.1834547,"domain_scores_codex":[0.9951609,0.0007914446,0.0002406284,0.0003777597,0.00270037,0.0007290249],"domain_scores_gemma":[0.9749223,0.002029299,0.001217198,0.001604607,0.01464415,0.005582395],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00006734631,0.00002808459,0.00005417341,0.0002330924,0.000009677364,0.000009274662,0.000007197609,0.0001004531,0.0005116216,0.0001541133,0.9872155,0.01160951],"study_design_scores_gemma":[0.0002123264,0.0002000443,0.002180599,0.0003101336,0.00007259021,0.00008007586,0.00007686051,0.001422332,0.002901505,0.001346979,0.9911097,0.00008689659],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"other","genre_scores_codex":[0.004341382,0.02467368,0.02427346,0.01499473,0.0485831,0.00419523,0.6383135,0.04834348,0.1922814],"genre_scores_gemma":[0.006311336,0.007253632,0.01853197,0.003239702,0.008082471,0.002195654,0.6154247,0.005038921,0.3339216],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.1602651,"threshold_uncertainty_score":0.5361401,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2135220386","doi":"","title":"Query-Structure Based Web Page Indexing","year":2012,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Search engine indexing; Computer science; Information retrieval; Web page; Index (typography); World Wide Web; Web search query; Set (abstract data type); Static web page; Search engine; Web navigation","authors":[{"name":"Falah Al-akashi","is_ca":true},{"name":"Diana Inkpen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02983501488973424,"gpt":0.2575284238493121,"spread":0.2276934089595779,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002506004,0.001028791,0.002246284,0.007955873,0.001578396,0.003510144,0.00289301,0.001244792,0.004643549],"category_scores_gemma":[0.01018545,0.0006456714,0.001204089,0.01168967,0.001344521,0.006677349,0.002901894,0.001256402,0.004782162],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001485415,"about_ca_system_score_gemma":0.002561274,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005409706,"about_ca_topic_score_gemma":0.007399084,"domain_scores_codex":[0.9959713,0.0006110158,0.0005147148,0.0005409865,0.002056753,0.0003052768],"domain_scores_gemma":[0.9926155,0.0017394,0.0004942094,0.002872358,0.002026619,0.0002518631],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003724601,0.0007020903,0.004344872,0.0008646477,0.0001511424,0.0002225863,0.0007940956,0.01569455,0.07158908,0.03974602,0.0710656,0.7944528],"study_design_scores_gemma":[0.0002465493,0.0007809799,0.006111533,0.0001276525,0.0002389544,0.001667197,0.0007133445,0.596708,0.1288933,0.1096794,0.1545393,0.0002937441],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02283905,0.002787762,0.9333307,0.0006165609,0.0002303407,0.0011977,0.004581975,0.02681948,0.007596481],"genre_scores_gemma":[0.2010451,0.001797412,0.7693122,0.0004087705,0.0004316098,0.0008591648,0.01593281,0.001597148,0.008615846],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007955873,"threshold_uncertainty_score":0.01553428,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W26866788","doi":"10.1088/0957-4484/27/10/105204","title":"Experiments for HARD and Enterprise Tracks.","year":2005,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"Fundamental Research Funds for the Central Universities; National Natural Science Foundation of China","keywords":"Cohesion (chemistry); Computer science; Natural language processing; Lexical database; Social connectedness; Artificial intelligence; Information retrieval; Selection (genetic algorithm); Lexical item; Schema (genetic algorithms); Linguistics; Psychology; WordNet","authors":[{"name":"Olga Vechtomova","is_ca":true},{"name":"Maheedhar Kolla","is_ca":false},{"name":"Murat Karamuftuoglu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0348368139127696,"gpt":0.3130115114656667,"spread":0.2781746975528971,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001096493,0.0004604264,0.0003213707,0.0007364128,0.001892975,0.002027692,0.0008902908,0.00130525,0.1002617],"category_scores_gemma":[0.002042559,0.0002672195,0.0003159272,0.001238925,0.0004859813,0.002079929,0.001443445,0.001045955,0.03216751],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009416335,"about_ca_system_score_gemma":0.001009401,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00330704,"about_ca_topic_score_gemma":0.005223391,"domain_scores_codex":[0.9986019,0.0001546768,0.00006078235,0.0004987445,0.0004767011,0.0002072439],"domain_scores_gemma":[0.9985348,0.0001845122,0.00009367152,0.0004250526,0.0004840334,0.000278014],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003806325,0.0019848,0.016645,0.00122593,0.0001470808,0.001392152,0.003500403,0.001471254,0.28721,0.1077391,0.201867,0.3730111],"study_design_scores_gemma":[0.0002561547,0.001270421,0.01044754,0.0001274692,0.00006686395,0.0008813381,0.001902383,0.002560584,0.1589284,0.01485297,0.8086368,0.00006919235],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2483395,0.002658606,0.071541,0.004110382,0.002863324,0.00173705,0.02025715,0.004861486,0.6436316],"genre_scores_gemma":[0.4012441,0.00140186,0.05444549,0.001816571,0.0001357762,0.001133048,0.01378868,0.0005466238,0.5254878],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1002617,"threshold_uncertainty_score":0.335409,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3175228675","doi":"","title":"Abstract Podcast Summarization using BART with Longformer Attention.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Automatic summarization; Computer science; Information retrieval; Bartok; Multi-document summarization; Natural language processing; Literature; Art","authors":[{"name":"Hannes Karlbom","is_ca":false},{"name":"Ann Clifton","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05972704396145603,"gpt":0.2576175202927503,"spread":0.1978904763312942,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009880287,0.001188894,0.001556703,0.003079118,0.001131353,0.002189683,0.001266888,0.001174187,0.01786386],"category_scores_gemma":[0.004906649,0.0003176511,0.000725773,0.002821326,0.0004125367,0.002356074,0.001628159,0.001245271,0.006409601],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005375932,"about_ca_system_score_gemma":0.001365966,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005041889,"about_ca_topic_score_gemma":0.01117438,"domain_scores_codex":[0.9992656,0.0001294926,0.00006517542,0.0001782003,0.0002502216,0.0001111992],"domain_scores_gemma":[0.9976566,0.0005678583,0.0001373334,0.0004382987,0.0009879549,0.0002119205],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001663616,0.0002861343,0.001132941,0.0003821042,0.0001319972,0.0002297831,0.0001819279,0.01262534,0.02771937,0.002966675,0.05644362,0.8962366],"study_design_scores_gemma":[0.0002745067,0.0009269057,0.005498744,0.00009988905,0.000450588,0.0003588773,0.0006883023,0.8617005,0.04821687,0.01784012,0.06383117,0.000113458],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04515396,0.004560372,0.9017177,0.001264785,0.003634863,0.0005641843,0.004816881,0.02645124,0.01183602],"genre_scores_gemma":[0.5331414,0.001593088,0.3976619,0.0007625823,0.00377862,0.0004118345,0.01696256,0.001611679,0.04407635],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01786386,"threshold_uncertainty_score":0.05976057,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2546705495","doi":"","title":"WaterlooClarke: TREC 2015 Clinical Decision Support Track","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Information retrieval; Mean reciprocal rank; Search engine; Reciprocal; Clinical decision support system; Test (biology); Rank (graph theory); Decision support system; Data mining","authors":[{"name":"Amira Ghenai","is_ca":true},{"name":"Eldar Khalilov","is_ca":false},{"name":"Pavel Valov","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08993383767107038,"gpt":0.3806690461193441,"spread":0.2907352084482737,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01588101,0.002689178,0.00239111,0.007491503,0.002304787,0.004924522,0.003775383,0.003249478,0.03233977],"category_scores_gemma":[0.03375185,0.0009227378,0.001247411,0.005722338,0.001085069,0.004576976,0.002217749,0.002943837,0.01992835],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009992594,"about_ca_system_score_gemma":0.01698072,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1644995,"about_ca_topic_score_gemma":0.270072,"domain_scores_codex":[0.9933714,0.002027831,0.000941037,0.0008191824,0.002371358,0.0004691753],"domain_scores_gemma":[0.9685702,0.009968861,0.001862135,0.002199364,0.01474278,0.002656629],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003070965,0.0002105154,0.0009338949,0.0009810907,0.0001031484,0.00007845258,0.00005550222,0.0008860623,0.001319415,0.0003740012,0.9654694,0.0292815],"study_design_scores_gemma":[0.003198814,0.001060146,0.02635803,0.001554394,0.0005518914,0.0007390126,0.0005435813,0.04867889,0.01784947,0.006426466,0.8926049,0.0004343838],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.03641379,0.02969932,0.02937449,0.05483222,0.00578544,0.008391572,0.737739,0.03154593,0.0662182],"genre_scores_gemma":[0.03816479,0.004605433,0.04460334,0.004586429,0.0008665811,0.00212677,0.8756398,0.001120256,0.02828665],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1644995,"threshold_uncertainty_score":0.3270841,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3013135062","doi":"","title":"Overview of the TREC 2018 Real-Time Summarization Track.","year":2018,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Automatic summarization; Computer science; Track (disk drive); Multi-document summarization; Information retrieval; World Wide Web; Operating system","authors":[{"name":"Royal Sequiera","is_ca":true},{"name":"Luchen Tan","is_ca":true},{"name":"Jimmy Lin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06766606094889346,"gpt":0.2918718006947948,"spread":0.2242057397459013,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008136649,0.002860575,0.002196843,0.01001973,0.002099106,0.004659898,0.003570239,0.00180876,0.03192693],"category_scores_gemma":[0.008990457,0.001098371,0.001579884,0.008820096,0.0004307086,0.005370573,0.00187987,0.002632866,0.04008277],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002617656,"about_ca_system_score_gemma":0.006636704,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0356416,"about_ca_topic_score_gemma":0.06005153,"domain_scores_codex":[0.9964895,0.000830996,0.0004042947,0.0006320597,0.001254418,0.0003887237],"domain_scores_gemma":[0.9880441,0.001488361,0.0006590751,0.001243619,0.007501266,0.001063544],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003056645,0.0002478429,0.0007240107,0.00182338,0.0001552266,0.00008812881,0.0001241238,0.001766142,0.013844,0.0009307282,0.7821152,0.1978756],"study_design_scores_gemma":[0.0002462525,0.0008998708,0.008732851,0.0006598209,0.0004332885,0.0003638337,0.0002658882,0.01651647,0.02930773,0.003205846,0.9391001,0.0002680014],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"review","genre_scores_codex":[0.02192868,0.10356,0.2530893,0.01607168,0.0154391,0.007026718,0.3362373,0.1493195,0.09732769],"genre_scores_gemma":[0.02725723,0.01996502,0.1900789,0.002338879,0.003606979,0.003251988,0.6682023,0.006404411,0.07889424],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.0356416,"threshold_uncertainty_score":0.1068062,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2946940318","doi":"","title":"UWaterlooMDS at the TREC 2018 Common Core Track.","year":2018,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Track (disk drive); Computer science; Core (optical fiber); Information retrieval; Artificial intelligence; Telecommunications; Operating system","authors":[{"name":"Mustafa Abualsaud","is_ca":true},{"name":"Gordon V. Cormack","is_ca":true},{"name":"Nimesh Ghelani","is_ca":true},{"name":"Amira Ghenai","is_ca":true},{"name":"Maura R. Grossman","is_ca":true},{"name":"Shahin Rahbariasl","is_ca":true},{"name":"Haotian Zhang","is_ca":true},{"name":"Mark D. Smucker","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03971195414912118,"gpt":0.2932357582956169,"spread":0.2535238041464958,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002314279,0.001460442,0.002659148,0.004353568,0.002615551,0.006280616,0.002161644,0.001624942,0.7170138],"category_scores_gemma":[0.005034408,0.0007373211,0.0006103997,0.006676141,0.0006332745,0.00531549,0.003123145,0.001950382,0.5120502],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003598168,"about_ca_system_score_gemma":0.00492426,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.07756723,"about_ca_topic_score_gemma":0.2305425,"domain_scores_codex":[0.9985167,0.0001796031,0.00007155792,0.0002897819,0.0007339402,0.0002084211],"domain_scores_gemma":[0.9953995,0.0003472846,0.0001054306,0.0004909902,0.002269926,0.001386722],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0000225653,0.00002420315,0.00004463292,0.0000343797,0.000001959254,0.000008275456,0.000008047572,0.00002680463,0.0001687156,0.0003989428,0.9888188,0.01044266],"study_design_scores_gemma":[0.00002923078,0.0000182904,0.0007514722,0.00005282773,0.000004314703,0.00001615334,0.00006021081,0.0003416156,0.0003624454,0.001705245,0.9966434,0.00001471628],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.0024558,0.004608874,0.007300077,0.0155866,0.01267632,0.0007554541,0.2860174,0.0164507,0.6541489],"genre_scores_gemma":[0.00412772,0.001656484,0.004688639,0.001521651,0.001274119,0.0001773921,0.108339,0.002701893,0.8755131],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7170138,"threshold_uncertainty_score":0.4036454,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2572043595","doi":"","title":"WaterlooClarke: TREC 2015 Total Recall Track","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Recall; Track (disk drive); Cluster analysis; Selection (genetic algorithm); Process (computing); Precision and recall; Information retrieval; Data mining; Artificial intelligence","authors":[{"name":"Haotian Zhang","is_ca":true},{"name":"Lin Wu","is_ca":true},{"name":"Yipeng Wang","is_ca":true},{"name":"Charles L. A. Clarke","is_ca":true},{"name":"Mark D. Smucker","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0752741043559129,"gpt":0.2872297986682088,"spread":0.2119556943122959,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01448561,0.002032948,0.002095816,0.007356036,0.003616372,0.006920555,0.003400273,0.002403941,0.02369987],"category_scores_gemma":[0.0205852,0.0007930917,0.0009092933,0.004083272,0.001361721,0.004411845,0.002564009,0.003380972,0.01586404],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01200396,"about_ca_system_score_gemma":0.01634326,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.271963,"about_ca_topic_score_gemma":0.5217234,"domain_scores_codex":[0.9910861,0.001628319,0.0004408884,0.0007808627,0.005269765,0.0007940609],"domain_scores_gemma":[0.9713995,0.002266387,0.000967263,0.002253056,0.02098079,0.002132892],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001845378,0.0001371379,0.0007162642,0.0003116088,0.00006144543,0.00003390732,0.00004166469,0.0009562999,0.002010691,0.0009522534,0.9629249,0.03166927],"study_design_scores_gemma":[0.0005675825,0.0007032862,0.02162011,0.0003693745,0.0002089898,0.0002502717,0.0002613504,0.02540704,0.01967499,0.004582406,0.9260593,0.0002954387],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.04729506,0.03558841,0.0710213,0.04083132,0.02050482,0.007762236,0.4767047,0.04615949,0.2541327],"genre_scores_gemma":[0.09328292,0.005436054,0.04583829,0.004940753,0.00193638,0.002325764,0.5402797,0.002435003,0.3035253],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.271963,"threshold_uncertainty_score":0.54076,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1503829033","doi":"","title":"Experiments with the Negotiated Boolean Queries of the TREC 2008 Legal Track","year":2008,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Open Text (Canada)","funders":"","keywords":"Relevance (law); Task (project management); Computer science; Relevance feedback; Information retrieval; Recall; Standard Boolean model; Rank (graph theory); Boolean expression; Learning to rank; Boolean network; And-inverter graph; Theoretical computer science; Boolean function; Data mining; Artificial intelligence; Algorithm; Mathematics; Ranking (information retrieval); Combinatorics; Psychology; Cognitive psychology","authors":[{"name":"Stephen Tomlinson","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04573418768938414,"gpt":0.2495933093840824,"spread":0.2038591216946983,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0119051,0.001926317,0.002209743,0.001337522,0.002068636,0.001995619,0.0022435,0.002416651,0.003809337],"category_scores_gemma":[0.04002263,0.000735289,0.0009949096,0.002422002,0.001084087,0.002851512,0.001763914,0.002137366,0.001631231],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001887333,"about_ca_system_score_gemma":0.00156927,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01432741,"about_ca_topic_score_gemma":0.01474274,"domain_scores_codex":[0.9843163,0.008304994,0.001367103,0.002150874,0.003229439,0.0006313056],"domain_scores_gemma":[0.9626464,0.02851802,0.001482786,0.002916577,0.003274839,0.001161329],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.03236903,0.02388874,0.03836549,0.008505081,0.00380754,0.002405461,0.00586862,0.1216327,0.1625288,0.003208821,0.1005139,0.4969058],"study_design_scores_gemma":[0.008120317,0.03329992,0.117037,0.0003721862,0.001906527,0.003156762,0.003761257,0.5885445,0.1803715,0.004405428,0.05804318,0.0009815376],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9726173,0.002249887,0.009712175,0.0007292391,0.0003888877,0.0008673828,0.003192299,0.004077038,0.006165873],"genre_scores_gemma":[0.9373167,0.0004664112,0.04033968,0.0004498585,0.0004117449,0.0007778126,0.01403035,0.0006252815,0.00558226],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01432741,"threshold_uncertainty_score":0.06296098,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2182378606","doi":"","title":"Universite de Montreal at TREC 2013: Experiments with Quantum Language Models in the Web Track.","year":2013,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Université de Montréal","funders":"","keywords":"Robustness (evolution); Computer science; Language model; Artificial intelligence; Focus (optics); Information retrieval; Data mining","authors":[{"name":"Alessandro Sordoni","is_ca":true},{"name":"Wei Yuan","is_ca":true},{"name":"Jian‐Yun Nie","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03172284299808632,"gpt":0.2476529462624022,"spread":0.2159301032643159,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01196033,0.002630054,0.001896585,0.001385499,0.002720298,0.002266593,0.003127241,0.00256919,0.01105106],"category_scores_gemma":[0.02092661,0.0007766219,0.001473482,0.002080713,0.0009778687,0.004066245,0.001840319,0.003653599,0.005702792],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006045502,"about_ca_system_score_gemma":0.005000384,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.291977,"about_ca_topic_score_gemma":0.3823018,"domain_scores_codex":[0.9944775,0.002832887,0.0002292022,0.0009008114,0.001057597,0.0005021073],"domain_scores_gemma":[0.9900803,0.005462933,0.0003278703,0.001480162,0.001578812,0.001069828],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.005400076,0.01121858,0.009376397,0.001905683,0.001277835,0.0004338104,0.000850266,0.06879713,0.008614266,0.00544503,0.646936,0.239745],"study_design_scores_gemma":[0.005223336,0.007532037,0.04993889,0.0004095506,0.0009850633,0.0004228171,0.001764853,0.7205356,0.02867185,0.01795079,0.1657602,0.0008050893],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6364254,0.02166127,0.06466748,0.009958145,0.006254591,0.006907433,0.1079313,0.04300958,0.1031847],"genre_scores_gemma":[0.6806932,0.002540616,0.08554547,0.002875383,0.0009148879,0.002669645,0.1702147,0.002097773,0.05244831],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.291977,"threshold_uncertainty_score":0.580555,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2604796591","doi":"","title":"\"When to Stop\" Waterloo (Cormack) Participation in the TREC 2016 Total Recall Track.","year":2016,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Border Security and International Relations","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Recall; Track (disk drive); Computer science; Information retrieval; Natural language processing; Psychology; Operating system; Cognitive psychology","authors":[{"name":"Gordon V. Cormack","is_ca":true},{"name":"Maura R. Grossman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06282087771289732,"gpt":0.3540823752187992,"spread":0.2912614975059018,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009419212,0.0009098294,0.001006043,0.001826579,0.004479566,0.004405573,0.002445646,0.002407358,0.05124534],"category_scores_gemma":[0.01492743,0.0003065591,0.0004994993,0.001696255,0.001238661,0.003396224,0.002431605,0.002290394,0.02080099],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005832062,"about_ca_system_score_gemma":0.01519569,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.3191621,"about_ca_topic_score_gemma":0.6500387,"domain_scores_codex":[0.9953619,0.0015889,0.0002902941,0.00062732,0.00153724,0.0005944304],"domain_scores_gemma":[0.989551,0.001486498,0.0003312332,0.0006562002,0.006639223,0.001335793],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001501241,0.0000522453,0.0007755391,0.0001597927,0.000009814786,0.00002236861,0.000354374,0.00009595444,0.001251499,0.0004834821,0.9594476,0.0371972],"study_design_scores_gemma":[0.0000803477,0.0001111776,0.006370639,0.0002661249,0.00004032345,0.00006289206,0.001904452,0.001144448,0.005807392,0.001558496,0.9825912,0.00006249005],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.08449607,0.02020852,0.03304755,0.2044785,0.02302779,0.004432297,0.1968069,0.01356998,0.4199323],"genre_scores_gemma":[0.1030924,0.004435987,0.02578557,0.02108015,0.001622139,0.0009755804,0.126893,0.002611313,0.713504],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.3191621,"threshold_uncertainty_score":0.6346087,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W125929235","doi":"","title":"In Enterprise Search: Methods to Identify Argumentative Discussions and to Find Topical Experts.","year":2006,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Expert finding and Q&A systems","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Argumentative; Computer science; Ranking (information retrieval); Argument (complex analysis); Information retrieval; Rank (graph theory); World Wide Web; Topic model; Data science; Mathematics","authors":[{"name":"Maheedhar Kolla","is_ca":true},{"name":"Olga Vechtomova","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06126711204444565,"gpt":0.4111753007816831,"spread":0.3499081887372375,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01469083,0.00184322,0.002918023,0.01552042,0.002641454,0.007460475,0.003660403,0.005022048,0.02362129],"category_scores_gemma":[0.05330307,0.001121327,0.001968276,0.01238165,0.002892332,0.01310514,0.004800465,0.002623906,0.008235662],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001788444,"about_ca_system_score_gemma":0.002172649,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003082418,"about_ca_topic_score_gemma":0.006926318,"domain_scores_codex":[0.9836028,0.01126954,0.0007411248,0.001653018,0.002338675,0.0003948114],"domain_scores_gemma":[0.9389367,0.05308066,0.001979776,0.003076413,0.002121798,0.000804661],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004123701,0.0007340079,0.01013625,0.003203636,0.0006105514,0.0002881846,0.002804995,0.006101186,0.002048191,0.1016246,0.05684399,0.815192],"study_design_scores_gemma":[0.0004084918,0.0003256188,0.01207472,0.001281313,0.0003631162,0.001491957,0.005076489,0.1403735,0.004318241,0.6257686,0.2082094,0.0003085685],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01871308,0.03735979,0.8923295,0.006852638,0.001010235,0.001774134,0.003550988,0.002838128,0.0355716],"genre_scores_gemma":[0.1044998,0.007301779,0.8545703,0.002043601,0.001036504,0.002152125,0.004221282,0.0005117679,0.02366283],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02362129,"threshold_uncertainty_score":0.0790211,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2572785564","doi":"","title":"WaterlooClarke: TREC 2015 Temporal Summarization Track.","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Automatic summarization; Computer science; Track (disk drive); Multi-document summarization; Information retrieval","authors":[{"name":"Ahsan Raza","is_ca":false},{"name":"Devin M. Rotondo","is_ca":false},{"name":"Charles L. A. Clarke","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04169239849904795,"gpt":0.2977835804427729,"spread":0.256091181943725,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004229312,0.002339002,0.001903537,0.005931581,0.002466084,0.003956087,0.00328958,0.001533863,0.05565622],"category_scores_gemma":[0.008653516,0.0006817976,0.0009297352,0.005481855,0.0008557058,0.005237808,0.001852506,0.002205516,0.03810326],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004214403,"about_ca_system_score_gemma":0.01020843,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.2007812,"about_ca_topic_score_gemma":0.394282,"domain_scores_codex":[0.9978922,0.0005455954,0.0002219027,0.0003373482,0.00079961,0.0002033503],"domain_scores_gemma":[0.9914004,0.001073123,0.0003888904,0.0008902984,0.00565314,0.000594245],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0000897623,0.00004223484,0.00009397145,0.0003392551,0.00003131018,0.00002195524,0.00003091272,0.0002553123,0.002122566,0.0004503889,0.9723853,0.02413686],"study_design_scores_gemma":[0.0002836052,0.0001630016,0.0043053,0.0002951408,0.0001946333,0.0001356502,0.0003338899,0.008324075,0.0143377,0.003592112,0.9679184,0.0001164128],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.009614935,0.01873,0.06440945,0.01281799,0.007839403,0.002585698,0.7714569,0.04199877,0.07054693],"genre_scores_gemma":[0.009979734,0.003177332,0.05017483,0.001409443,0.0007833684,0.0008625587,0.8469881,0.002317729,0.08430685],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.2007812,"threshold_uncertainty_score":0.3992251,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2576853603","doi":"","title":"Laval University and Lakehead University at TREC Dynamic Domain 2015: Combination of Techniques for Subtopics Coverage.","year":2015,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Lakehead University; Université Laval","funders":"","keywords":"Computer science; Domain (mathematical analysis); Information retrieval; Library science; Mathematics","authors":[{"name":"Robin Joganah","is_ca":false},{"name":"Luc Lamontagne","is_ca":true},{"name":"Richard Khoury","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02569653499890819,"gpt":0.2502806054467507,"spread":0.2245840704478425,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005580177,0.001935767,0.001460173,0.01268018,0.001648923,0.003161165,0.002305296,0.001736423,0.01719768],"category_scores_gemma":[0.02187644,0.0008030303,0.001218858,0.007227711,0.0006933728,0.005870084,0.003520823,0.00220693,0.01303868],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002072274,"about_ca_system_score_gemma":0.004106578,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04595241,"about_ca_topic_score_gemma":0.1115282,"domain_scores_codex":[0.9922167,0.001932863,0.000408374,0.001437282,0.003235384,0.0007693003],"domain_scores_gemma":[0.9856039,0.003813663,0.000518431,0.002591351,0.005955839,0.001516885],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0008685585,0.0005761759,0.008367204,0.0004823333,0.000164672,0.0001213156,0.0003947157,0.003370088,0.02119266,0.002397073,0.3867164,0.5753489],"study_design_scores_gemma":[0.0007435952,0.001196217,0.06814034,0.0004058972,0.0004594416,0.000935797,0.002049868,0.2730742,0.1226745,0.01606289,0.5137635,0.0004937129],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1729242,0.02245936,0.3759293,0.01173189,0.003851753,0.002292669,0.1798733,0.1352432,0.09569445],"genre_scores_gemma":[0.2300602,0.003026803,0.445011,0.001010407,0.0009927576,0.001262241,0.2438617,0.007998213,0.06677672],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.04595241,"threshold_uncertainty_score":0.09136987,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3176527752","doi":"","title":"Spotify at the TREC 2020 Podcasts Track: Segment Retrieval.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Track (disk drive); Computer science; Information retrieval; World Wide Web; Operating system","authors":[{"name":"Yongze Yu","is_ca":false},{"name":"Jussi Karlgren","is_ca":false},{"name":"Ann Clifton","is_ca":true},{"name":"Md. Iftekhar Tanveer","is_ca":false},{"name":"Rosie Jones","is_ca":false},{"name":"Hamed Bonab","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05451127183627481,"gpt":0.2668095807724929,"spread":0.2122983089362181,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004697726,0.002605902,0.002200182,0.003562521,0.002617128,0.005394547,0.002425323,0.002778831,0.09111392],"category_scores_gemma":[0.007243303,0.0005641526,0.0009149056,0.003332645,0.0006609594,0.005216287,0.002577624,0.003158809,0.06771435],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002197122,"about_ca_system_score_gemma":0.004630622,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.06878214,"about_ca_topic_score_gemma":0.1561614,"domain_scores_codex":[0.997855,0.0004774475,0.00008110065,0.0003148188,0.0009207783,0.0003507822],"domain_scores_gemma":[0.9924774,0.001319226,0.0001812041,0.0008428376,0.003345078,0.001834272],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000142837,0.00007740664,0.0001454623,0.0001433739,0.00001658794,0.0000214668,0.00003571178,0.0001781407,0.00272046,0.0002875172,0.9815672,0.0146638],"study_design_scores_gemma":[0.0006848253,0.0008269942,0.007826312,0.0001754328,0.0001547846,0.0001969124,0.0006182693,0.01722401,0.02102565,0.00445498,0.9466633,0.0001485302],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.03491908,0.0153802,0.07520244,0.0288929,0.03860525,0.003982918,0.5637702,0.09874078,0.1405062],"genre_scores_gemma":[0.03192368,0.002551854,0.04111877,0.00356169,0.005238068,0.0009285126,0.675215,0.005684714,0.2337777],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.09111392,"threshold_uncertainty_score":0.3048065,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2296709160","doi":"","title":"GUCAS at TREC 2011 Microblog Track.","year":2011,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Computer science; Microblogging; Social media; Context (archaeology); Relevance (law); Track (disk drive); Probabilistic logic; Information retrieval; Baseline (sea); Natural language processing; Query expansion; Artificial intelligence; Language model; World Wide Web","authors":[{"name":"Xin Zhang","is_ca":false},{"name":"Kai Hui","is_ca":false},{"name":"Ben He","is_ca":true},{"name":"Tiejian Luo","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09059576231893875,"gpt":0.2552359157374093,"spread":0.1646401534184706,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007395474,0.003923351,0.002864947,0.006008905,0.00394399,0.003822803,0.002702709,0.002966729,0.03237714],"category_scores_gemma":[0.0148717,0.0007151858,0.00110931,0.00353243,0.000658425,0.006441104,0.002290321,0.003052769,0.01968764],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006484662,"about_ca_system_score_gemma":0.00554504,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1770106,"about_ca_topic_score_gemma":0.247325,"domain_scores_codex":[0.9952023,0.00160817,0.0002430152,0.0006672394,0.001795791,0.0004835608],"domain_scores_gemma":[0.9892335,0.002483814,0.0003684731,0.001511402,0.004745489,0.001657298],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003460523,0.0006948509,0.0006904686,0.0005582514,0.00008969654,0.00008174958,0.00006531863,0.001113289,0.002131268,0.0005342386,0.9588061,0.0348888],"study_design_scores_gemma":[0.001928244,0.001531166,0.02168546,0.0003047897,0.000410389,0.0003830646,0.001043748,0.09617876,0.03209002,0.003328566,0.840717,0.0003989507],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.1056494,0.01416838,0.02120934,0.014621,0.01703408,0.006981912,0.665282,0.07146175,0.08359205],"genre_scores_gemma":[0.08192725,0.002690453,0.04146848,0.002149773,0.00200006,0.003373956,0.7520476,0.002332142,0.1120104],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1770106,"threshold_uncertainty_score":0.3519605,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3173373527","doi":"","title":"TREC 2020 Notebook: CAsT Track.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Track (disk drive); Computer science; Operating system","authors":[{"name":"Sheng-Chieh Lin","is_ca":false},{"name":"Jheng-Hong Yang","is_ca":false},{"name":"Jimmy Lin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06551698452288615,"gpt":0.2644454919931136,"spread":0.1989285074702274,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006658957,0.004504459,0.00308029,0.009631332,0.002431271,0.00719639,0.005421328,0.002355945,0.2679192],"category_scores_gemma":[0.01100251,0.001313659,0.001438333,0.01025905,0.0008779931,0.006125077,0.002536385,0.002956116,0.2820902],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004675356,"about_ca_system_score_gemma":0.01277009,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.167513,"about_ca_topic_score_gemma":0.2149768,"domain_scores_codex":[0.9971201,0.0005336733,0.0001765314,0.0002845172,0.00150092,0.0003842585],"domain_scores_gemma":[0.9829618,0.001595598,0.0007039419,0.001185708,0.01020368,0.003349389],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003026413,0.00001366763,0.00001995832,0.0001392052,0.000005262087,0.000003437343,0.000002090427,0.00008060235,0.000214068,0.0001217166,0.9926087,0.006760996],"study_design_scores_gemma":[0.0002270702,0.0001796208,0.002454943,0.0002885069,0.00006573498,0.00007316947,0.00004782663,0.001636488,0.002319575,0.002072472,0.9905272,0.0001073479],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"methods","genre_scores_codex":[0.002049759,0.01363951,0.009974361,0.007061492,0.02092871,0.001733451,0.7268223,0.02055333,0.1972371],"genre_scores_gemma":[0.005310373,0.005175546,0.01027976,0.002299771,0.004380962,0.001149142,0.6500084,0.004168643,0.3172274],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.2679192,"threshold_uncertainty_score":0.8962792,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W342589544","doi":"","title":"York University at TREC 2012: CrowdSourcing Track.","year":2012,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Crowdsourcing; Computer science; Quality (philosophy); Work (physics); Crowdsourcing software development; Data science; Control (management); Information retrieval; World Wide Web; Artificial intelligence; Engineering","authors":[{"name":"Qinmin Hu","is_ca":true},{"name":"Zhi Xu","is_ca":false},{"name":"Jimmy Xiangji Huang","is_ca":true},{"name":"Zheng Ye","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04766109456374461,"gpt":0.2510287572091137,"spread":0.2033676626453691,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01987685,0.007234418,0.00445744,0.009414261,0.009248908,0.009217993,0.005930183,0.004843805,0.08711207],"category_scores_gemma":[0.02264315,0.001461212,0.001348932,0.008401699,0.002214527,0.008890569,0.004935153,0.006895278,0.08572266],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0125611,"about_ca_system_score_gemma":0.01893056,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.3457804,"about_ca_topic_score_gemma":0.4783251,"domain_scores_codex":[0.9878335,0.003597973,0.0006569964,0.001323473,0.005505962,0.00108215],"domain_scores_gemma":[0.9703686,0.00359933,0.001033894,0.004514141,0.01497625,0.005507737],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004827742,0.00006008073,0.00009386343,0.0001389481,0.00001029155,0.00001022141,0.00002034499,0.0003505086,0.0002113424,0.0002812657,0.9910572,0.007717749],"study_design_scores_gemma":[0.0003784115,0.0001516333,0.004926038,0.0004251365,0.00004860865,0.00006172384,0.0003434896,0.009215571,0.002834974,0.005097433,0.9762915,0.0002255778],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.008757146,0.01700005,0.02163902,0.03626464,0.01773245,0.005122889,0.6697195,0.03786725,0.185897],"genre_scores_gemma":[0.02087327,0.004106049,0.02560998,0.003293392,0.002023505,0.003410425,0.7230422,0.003683373,0.2139579],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.3457804,"threshold_uncertainty_score":0.6875354,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2403585915","doi":"","title":"University of Ottawa at TREC 2010 Web Track Ranking Web Documents Using Meta-Data","year":2010,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Track (disk drive); Ranking (information retrieval); Information retrieval; World Wide Web","authors":[{"name":"Falah Al-akashi","is_ca":true},{"name":"Diana Inkpen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0936111074354567,"gpt":0.2862664339247185,"spread":0.1926553264892618,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009206556,0.001996802,0.002186654,0.007361643,0.004986791,0.008567633,0.002980205,0.001699443,0.04607755],"category_scores_gemma":[0.01557219,0.001387502,0.0009926368,0.007491638,0.001691746,0.005238658,0.001667178,0.002575822,0.0240755],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.02348439,"about_ca_system_score_gemma":0.03517812,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.832477,"about_ca_topic_score_gemma":0.8948792,"domain_scores_codex":[0.9934769,0.0008896525,0.00030149,0.0008778123,0.003447707,0.001006471],"domain_scores_gemma":[0.9607899,0.002942892,0.0004730314,0.002895601,0.02840463,0.004494096],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004684631,0.0002117168,0.002407877,0.0002221993,0.00005549366,0.00006542487,0.0001665179,0.00167241,0.002642578,0.001791217,0.929446,0.06085013],"study_design_scores_gemma":[0.0005124972,0.0003809478,0.02694826,0.0004282427,0.0003136637,0.0001785249,0.001457414,0.02803431,0.02504309,0.004611707,0.9117294,0.0003619226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.1093102,0.03715201,0.04969872,0.08597151,0.0227293,0.002215936,0.4120395,0.03917145,0.2417114],"genre_scores_gemma":[0.1300956,0.01017485,0.06737836,0.00160501,0.001248725,0.0004015886,0.2184267,0.003258476,0.5674107],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.832477,"threshold_uncertainty_score":0.3370191,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2407544020","doi":"","title":"Ranking Web Pages Using Collective Knowledge.","year":2011,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Search engine indexing; Information retrieval; Web page; Ranking (information retrieval); World Wide Web; Static web page; Index (typography); Web navigation","authors":[{"name":"Falah Al-akashi","is_ca":true},{"name":"Diana Inkpen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1401130716483935,"gpt":0.2917074747109926,"spread":0.1515944030625991,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002927383,0.00110925,0.001734135,0.02312804,0.001287911,0.004313236,0.00122542,0.001014984,0.002873558],"category_scores_gemma":[0.01337517,0.0004465196,0.001127966,0.01421608,0.0005735827,0.004948688,0.001989674,0.000974493,0.003321356],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001251378,"about_ca_system_score_gemma":0.001480682,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004978637,"about_ca_topic_score_gemma":0.01435791,"domain_scores_codex":[0.9968514,0.0005423191,0.000228782,0.0004784484,0.001694759,0.0002042334],"domain_scores_gemma":[0.9924893,0.002709763,0.0008930835,0.001777537,0.001677537,0.0004528276],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002112216,0.0005819359,0.02225172,0.0008206472,0.0005794395,0.0002192354,0.0005069761,0.01242577,0.007846484,0.01112836,0.05087525,0.892553],"study_design_scores_gemma":[0.0001952299,0.001065924,0.06435592,0.0005733579,0.001264076,0.001835949,0.001911187,0.5113569,0.03404561,0.2263054,0.1566932,0.0003972744],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.212673,0.01673969,0.6544166,0.002420217,0.0007356385,0.00204992,0.02216678,0.02194862,0.06684953],"genre_scores_gemma":[0.6349729,0.003353688,0.313951,0.0002472528,0.0006964006,0.0005018158,0.03066734,0.0006419564,0.01496762],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02312804,"threshold_uncertainty_score":0.01548165,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3173187978","doi":"","title":"The CLaC System at the TREC 2020 News Track.","year":2020,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Track (disk drive); Computer science; Operating system","authors":[{"name":"Pavel Khloponin","is_ca":true},{"name":"Leila Kosseim","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02671213014052321,"gpt":0.2636659143570772,"spread":0.236953784216554,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004110232,0.002386817,0.001810962,0.005277916,0.002104765,0.004032168,0.003394437,0.0029307,0.07925594],"category_scores_gemma":[0.0080679,0.0006495892,0.0008869894,0.003560626,0.0005143775,0.005159366,0.002965828,0.002016439,0.1071409],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002128896,"about_ca_system_score_gemma":0.00515174,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.042862,"about_ca_topic_score_gemma":0.07113191,"domain_scores_codex":[0.997074,0.001009174,0.0001769966,0.0004274471,0.0009783035,0.0003340032],"domain_scores_gemma":[0.9951335,0.00057109,0.0001316627,0.0008228168,0.00260257,0.0007382924],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001409628,0.00006762381,0.00016709,0.0001480481,0.00001882504,0.00002302685,0.00001529659,0.0001433631,0.00113348,0.0005432597,0.9760424,0.02155667],"study_design_scores_gemma":[0.0008040276,0.0003930037,0.004171336,0.0002820597,0.0001901546,0.0003588063,0.0002200967,0.02114186,0.01384146,0.006335604,0.9520629,0.000198634],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.01451023,0.009369059,0.05614618,0.009746712,0.009237397,0.002905227,0.5695041,0.1769685,0.1516127],"genre_scores_gemma":[0.02011231,0.001146511,0.05871778,0.002079651,0.001061694,0.0009287228,0.8103909,0.004817327,0.1007452],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.07925594,"threshold_uncertainty_score":0.2651376,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W60608539","doi":"","title":"Passage Retrieval by Shrinkage of Language Models.","year":2006,"lang":"en","type":"article","venue":"Text REtrieval Conference","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Guelph","funders":"","keywords":"Computer science; Information retrieval; Scope (computer science); Search engine; World Wide Web; The Internet; Process (computing); Human–computer information retrieval","authors":[{"name":"Fei Song","is_ca":false},{"name":"Joe Vasak","is_ca":false},{"name":"Wei Wang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02283974409388151,"gpt":0.2468398962509053,"spread":0.2240001521570238,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00930017,0.00186047,0.002499024,0.003448332,0.0009514055,0.00240764,0.002700091,0.001855396,0.008803787],"category_scores_gemma":[0.03729002,0.001320247,0.003052324,0.002926619,0.00166884,0.005749535,0.002797563,0.003694678,0.008370481],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001460488,"about_ca_system_score_gemma":0.001591328,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00600573,"about_ca_topic_score_gemma":0.006388693,"domain_scores_codex":[0.9960006,0.002182406,0.0002163176,0.0007439905,0.0006266758,0.0002300116],"domain_scores_gemma":[0.9875132,0.009126886,0.0005420595,0.001558041,0.001005977,0.0002537596],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000777949,0.000336583,0.002021171,0.0009978095,0.0005637226,0.0005002843,0.0007464394,0.2759209,0.006187542,0.2498753,0.0380997,0.4239726],"study_design_scores_gemma":[0.0000723843,0.0001146427,0.0004399695,0.00005797728,0.0001287715,0.0002117534,0.00004386747,0.8614446,0.001981169,0.1234862,0.01195954,0.00005904979],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00502526,0.002362251,0.9868326,0.000453058,0.000335312,0.0002110342,0.0005454954,0.001692207,0.002542799],"genre_scores_gemma":[0.2612721,0.007167402,0.6777792,0.001156633,0.002324821,0.002274149,0.007494954,0.002346441,0.03818431],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00930017,"threshold_uncertainty_score":0.04918462,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}