{"meta":{"query_hash":"9a3c2b27b4d5","filters":{"venue":"Mining Software Repositories"},"cohort_total":17,"direct_labels_cover":0,"predictions_cover":17,"exported":17,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/9a3c2b27b4d5","api":"https://metacan.xera.ac/api/v1/cohort?venue=Mining+Software+Repositories"},"results":[{"id":"W1968137197","doi":"10.5555/2664446.2664466","title":"The build dependency perspective of Android's concrete architecture","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Advanced Software Engineering Methodologies","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Android (operating system); Architecture; Computer science; Android application; Applications architecture; Embedded system; Operating system; Software architecture; Software","score_opus":0.016041832489007275,"score_gpt":0.276711630110583,"score_spread":0.2606697976215757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1968137197","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027724724,0.00667564,0.9621118,0.00013865672,0.0024367224,0.00010965747,0.0000022507104,0.00060684513,0.00019373106],"genre_scores_gemma":[0.307973,0.000024838624,0.69149923,0.0000115579005,0.00028614508,0.000019079976,4.3248224e-7,0.00001691991,0.0001687944],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9984037,0.00017641108,0.00029051473,0.00029053373,0.0003615701,0.00047726612],"domain_scores_gemma":[0.9950589,0.003616302,0.00020133935,0.0007815024,0.000248743,0.000093197596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00064616936,0.00020680355,0.00025271354,0.00006288103,0.00035836658,0.00009593152,0.00085035345,0.00009664332,6.2029915e-7],"category_scores_gemma":[0.004930605,0.00014927646,0.000096337804,0.00033040525,0.00020151303,0.00045943537,0.00035113745,0.00024486225,0.0000014978702],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022199942,0.00008556497,0.19599801,0.0003723425,0.0009429073,0.00013637551,0.21334507,0.017843748,0.018933207,0.30379173,0.0014952054,0.24683383],"study_design_scores_gemma":[0.0026068098,0.0022592188,0.086415224,0.0009750745,0.00033408377,0.0034688064,0.027503192,0.000641655,0.60449,0.21418644,0.051814105,0.00530543],"about_ca_topic_score_codex":0.000025788902,"about_ca_topic_score_gemma":0.0000016398061,"teacher_disagreement_score":0.58555675,"about_ca_system_score_codex":0.000090336405,"about_ca_system_score_gemma":0.000075627,"threshold_uncertainty_score":0.6087317},"labels":[],"label_agreement":null},{"id":"W1978832807","doi":"10.5555/2820518.2820554","title":"An empirical study of end-user programmers in the computer music community","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Metadata; Software; World Wide Web; Software development; Population; Source code; Computer programming; Musical; End user; Empirical research; Software engineering; Programming language","score_opus":0.0705591971718184,"score_gpt":0.3386892397372468,"score_spread":0.2681300425654284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1978832807","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9357346,0.00004595808,0.0628045,0.000053401858,0.00067601446,0.00033334948,6.204563e-7,0.00033516454,0.000016348271],"genre_scores_gemma":[0.9329204,1.9114994e-7,0.06680507,0.00003230637,0.00015033018,0.00006404658,0.0000029404562,0.000015646057,0.000009080638],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99711907,0.0009843529,0.00036151873,0.00030565602,0.0008676887,0.00036172933],"domain_scores_gemma":[0.9961051,0.0019583188,0.00008774961,0.0014566613,0.00026401327,0.0001281741],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019969363,0.00017895395,0.0002550118,0.00016246756,0.00019111346,0.0002903593,0.0019994392,0.000076840275,6.224639e-7],"category_scores_gemma":[0.0008361543,0.000135782,0.000039988186,0.0009716244,0.00011136864,0.00043114557,0.00044760556,0.00051547156,0.0000017254714],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010768196,0.00087790063,0.8415418,0.000019260237,0.00001950754,0.00007069204,0.14534669,0.00078205875,0.0000060351117,0.000021735304,0.0006729896,0.010630538],"study_design_scores_gemma":[0.001116289,0.0041001914,0.9603761,0.000063596766,0.000014717424,0.00010604374,0.029742897,0.0027026036,0.00019293756,0.000078960555,0.0011430192,0.00036264176],"about_ca_topic_score_codex":0.0009389728,"about_ca_topic_score_gemma":0.00018193157,"teacher_disagreement_score":0.11883427,"about_ca_system_score_codex":0.0000870366,"about_ca_system_score_gemma":0.00016936315,"threshold_uncertainty_score":0.55370295},"labels":[],"label_agreement":null},{"id":"W1983866039","doi":"10.5555/2820518.2820590","title":"A dataset of the activity of the git super-repository of Linux in 2012","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; Polytechnique Montréal; University of Victoria","funders":"","keywords":"Linux kernel; Computer science; World Wide Web; Kernel (algebra); Process (computing); Operating system","score_opus":0.026511254488833996,"score_gpt":0.26521439057043333,"score_spread":0.23870313608159935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1983866039","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99035674,0.00037944416,0.007115302,0.00008781177,0.0017599923,0.00016786621,0.000057942154,0.00004935622,0.00002553008],"genre_scores_gemma":[0.9885177,0.0000012467026,0.011257017,0.000004049124,0.00006888721,0.000012432115,0.0000027895328,0.000011690213,0.00012418018],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99811447,0.0001927933,0.00041162712,0.00026106928,0.00078680326,0.00023326007],"domain_scores_gemma":[0.9968115,0.0010959334,0.00025133856,0.0015070575,0.0002735609,0.000060594368],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061802036,0.00013652143,0.00027417677,0.00008674009,0.000070714625,0.00003564982,0.0016805888,0.000092678834,5.070943e-7],"category_scores_gemma":[0.0025085905,0.00008954186,0.000087304245,0.00078675343,0.00027190582,0.00039848773,0.0010117627,0.00022369048,3.9845656e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024038138,0.00011367142,0.98099744,0.00014333498,0.00003313475,0.000007479053,0.0061944798,0.0007170844,0.0074112588,0.000061261635,0.0026509436,0.0016458892],"study_design_scores_gemma":[0.0004597313,0.00018229324,0.5020541,0.00039427378,0.00001888915,0.00005790492,0.00017369626,0.00056950987,0.49398395,0.00007439814,0.0018291863,0.00020206675],"about_ca_topic_score_codex":0.0005927915,"about_ca_topic_score_gemma":0.000033624,"teacher_disagreement_score":0.4865727,"about_ca_system_score_codex":0.00008209187,"about_ca_system_score_gemma":0.0004589184,"threshold_uncertainty_score":0.36514112},"labels":[],"label_agreement":null},{"id":"W1991705476","doi":"10.5555/2820518.2820577","title":"Mining StackOverflow to filter out off-topic IRC discussion","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Spam and Phishing Detection","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Exploit; Filter (signal processing); Set (abstract data type); The Internet; World Wide Web; Online discussion; Stack (abstract data type); Information retrieval; Open source; Data science; Programming language; Computer security","score_opus":0.030339466299692388,"score_gpt":0.26051056375486775,"score_spread":0.23017109745517536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1991705476","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4689188,0.0002594494,0.512066,0.0014090694,0.014958548,0.00020862273,0.0000031184388,0.0011574336,0.0010189323],"genre_scores_gemma":[0.72410333,0.0000019401787,0.26727578,0.00029310308,0.0015051147,0.00003983141,0.000005936,0.00003281481,0.0067421612],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982518,0.000081083504,0.00028562717,0.0005153838,0.00052130583,0.0003447824],"domain_scores_gemma":[0.9986305,0.00013806287,0.000109260975,0.0006727214,0.00018671293,0.0002627298],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003327503,0.00020177147,0.00020404393,0.00011449253,0.00026703475,0.0005623002,0.0005618529,0.000109060435,0.0000031166744],"category_scores_gemma":[0.000700165,0.00015074815,0.00006757408,0.00030882348,0.000026809248,0.00073226116,0.00035871274,0.00011942032,0.00004150812],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012642296,0.00012804683,0.26394346,0.000084793974,0.00009208696,0.00030695522,0.17736746,0.00084805995,0.0016111959,0.00044345055,0.095918216,0.4591298],"study_design_scores_gemma":[0.0028648218,0.0034249648,0.08736383,0.00143079,0.00013221275,0.0004913708,0.0077301054,0.0087107085,0.060628857,0.0030239709,0.8206048,0.0035935356],"about_ca_topic_score_codex":0.000042165706,"about_ca_topic_score_gemma":0.000038760852,"teacher_disagreement_score":0.7246866,"about_ca_system_score_codex":0.00013442877,"about_ca_system_score_gemma":0.00012325138,"threshold_uncertainty_score":0.6147331},"labels":[],"label_agreement":null},{"id":"W1996488581","doi":"10.5555/2820518.2820567","title":"Do onboarding programs work","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Open Source Software Innovations","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Onboarding; Staffing; Computer science; Work (physics); Open source; Software engineering; Software bug; Software; Knowledge management; Data science; Engineering; Management","score_opus":0.04522810760757105,"score_gpt":0.2741533000331867,"score_spread":0.22892519242561565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1996488581","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22893286,0.0004878665,0.7581974,0.000567816,0.005307992,0.00032619064,0.0000010527989,0.0033380592,0.0028407224],"genre_scores_gemma":[0.43643248,5.037354e-7,0.562086,0.00006476889,0.00043032216,0.000059309958,0.00000555882,0.000025706373,0.00089536497],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99781466,0.00007490753,0.00041602054,0.00056625257,0.00066941814,0.00045873658],"domain_scores_gemma":[0.99777097,0.00021930889,0.00019796901,0.0010187295,0.0005827373,0.0002102867],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00058890256,0.00023211591,0.00024766568,0.00013481342,0.00035098364,0.001346816,0.001063233,0.00012052985,0.0000017387136],"category_scores_gemma":[0.0010109718,0.00022620348,0.00007735364,0.0017821512,0.00009977613,0.00096046546,0.00055193447,0.00019753759,0.00005663175],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015081178,0.00011995225,0.7239131,0.000024132145,0.00008170298,0.00015846881,0.028002564,0.00017321602,0.000028349423,0.008188082,0.010812796,0.22848257],"study_design_scores_gemma":[0.004420163,0.0023038485,0.07837059,0.0025262549,0.00021055825,0.0017556557,0.01641955,0.0009344741,0.007737425,0.023493044,0.85504574,0.0067826803],"about_ca_topic_score_codex":0.000028301873,"about_ca_topic_score_gemma":0.0000017432656,"teacher_disagreement_score":0.844233,"about_ca_system_score_codex":0.00016205401,"about_ca_system_score_gemma":0.00026419794,"threshold_uncertainty_score":0.9996899},"labels":[],"label_agreement":null},{"id":"W1998211793","doi":"10.5555/2820518.2820556","title":"What is the Gist?: understanding the use of public Gists on GitHub","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Open Source Software Innovations","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Computer science; GiST; Metadata; World Wide Web; Source code; Coding (social sciences); Variety (cybernetics); Information retrieval; Code (set theory); Medicine","score_opus":0.20485367274747782,"score_gpt":0.30147486018912495,"score_spread":0.09662118744164713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1998211793","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05009427,0.00048617637,0.92181134,0.01965715,0.0064663324,0.0004145946,0.000007792491,0.00060349336,0.0004588584],"genre_scores_gemma":[0.9461852,0.000011998311,0.05060353,0.001261147,0.0003091655,0.00003858772,0.000004414832,0.00003345079,0.0015525606],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99771017,0.00021276552,0.00045108993,0.0004024954,0.0008730926,0.0003503861],"domain_scores_gemma":[0.9954076,0.001973582,0.00037096674,0.0015828449,0.00056908064,0.00009591654],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0009046989,0.00020948604,0.00020757665,0.00011588557,0.00070819445,0.002954628,0.0014090952,0.0000895824,0.0000023762225],"category_scores_gemma":[0.001895076,0.00012774888,0.00009020361,0.0012827213,0.00039652162,0.0023003763,0.000492987,0.00023120036,0.000010418094],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052334173,0.0003049952,0.20024668,0.000084072424,0.0005189867,0.000090016976,0.27641192,0.0013663754,0.00007860761,0.29852328,0.1633053,0.059017446],"study_design_scores_gemma":[0.0022211384,0.001799712,0.019768078,0.0019254754,0.00020771875,0.00076541945,0.11929168,0.0065231873,0.0077882675,0.047661047,0.7892522,0.0027960443],"about_ca_topic_score_codex":0.000061688115,"about_ca_topic_score_gemma":0.000018025468,"teacher_disagreement_score":0.89609087,"about_ca_system_score_codex":0.00025479132,"about_ca_system_score_gemma":0.0003133925,"threshold_uncertainty_score":0.9980804},"labels":[],"label_agreement":null},{"id":"W2041589121","doi":"10.5555/2820518.2820597","title":"The Firefox temporal defect dataset","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Software bug; Process (computing); Software; Plan (archaeology); Data mining; Data science; Geography","score_opus":0.030207281263329028,"score_gpt":0.28240052547346595,"score_spread":0.2521932442101369,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2041589121","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2842198,0.008953754,0.6801862,0.0019955852,0.017313715,0.00073226006,0.00012930152,0.0059412066,0.0005281675],"genre_scores_gemma":[0.8308272,0.000012224466,0.16670018,0.00007236691,0.0008975061,0.000104277926,0.00012634129,0.000052353986,0.0012075506],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99802107,0.00010262143,0.0002501245,0.0004089438,0.00073722453,0.00048002688],"domain_scores_gemma":[0.99618995,0.0020015521,0.00007508188,0.001303402,0.00020289376,0.00022711],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009266607,0.00017876852,0.0001518369,0.000059853508,0.00044875272,0.00079459406,0.001472538,0.00007260815,6.525179e-7],"category_scores_gemma":[0.0051936875,0.00012773158,0.00006234013,0.00046164705,0.00011279787,0.000498538,0.00066630065,0.00020521987,0.000044952536],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028829822,0.00004541553,0.62435985,0.000038075592,0.000112811416,0.0003455057,0.0044049467,0.00031719892,0.000032910888,0.00072148035,0.3374398,0.032153174],"study_design_scores_gemma":[0.0006957459,0.0005039898,0.036073804,0.00009639013,0.00002252178,0.00050719804,0.00047121878,0.0015915463,0.0024927678,0.0009219865,0.9557997,0.00082313106],"about_ca_topic_score_codex":0.00012204632,"about_ca_topic_score_gemma":0.000018545114,"teacher_disagreement_score":0.61835986,"about_ca_system_score_codex":0.00012917382,"about_ca_system_score_gemma":0.00031177932,"threshold_uncertainty_score":0.76622885},"labels":[],"label_agreement":null},{"id":"W2064653370","doi":"10.5555/2820518.2820527","title":"Co-evolution of infrastructure and source code: an empirical study","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Computer science; Server; Source code; Empirical research; Critical infrastructure; Database; Process (computing); Software engineering; Operating system; Programming language; Computer security","score_opus":0.02434610871695176,"score_gpt":0.31298362237117433,"score_spread":0.2886375136542226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064653370","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71269083,0.00018997924,0.28623444,0.000018161374,0.00036915392,0.00010923463,0.0000016689628,0.00037673648,0.000009795094],"genre_scores_gemma":[0.9307715,5.335308e-7,0.06896124,0.0000051573948,0.00015123775,0.000012278814,0.0000018716297,0.000016551096,0.000079583886],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99831,0.0001269417,0.00026006493,0.00041310905,0.00063654507,0.00025333336],"domain_scores_gemma":[0.9982734,0.00049714156,0.00008727577,0.0006173836,0.00030520005,0.00021963706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050133537,0.00015664396,0.00022839094,0.00014248592,0.00012245317,0.00017193695,0.00051125797,0.000096195385,6.1292326e-7],"category_scores_gemma":[0.0018129473,0.00014690828,0.00002349833,0.00041320384,0.00010750988,0.00047469107,0.00022380377,0.00017574402,0.0000011356751],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012486065,0.00006195934,0.97741216,0.00001913795,0.000019833695,0.000020019846,0.017441038,0.00051512354,0.00009589074,0.000024842717,0.0007480675,0.0036294397],"study_design_scores_gemma":[0.00064501906,0.0013589722,0.9875875,0.000035966896,0.000015318163,0.0001616968,0.003945869,0.0029206823,0.0013043991,0.00021904398,0.0015107377,0.000294804],"about_ca_topic_score_codex":0.0000621034,"about_ca_topic_score_gemma":0.0000049120476,"teacher_disagreement_score":0.21808071,"about_ca_system_score_codex":0.00013452797,"about_ca_system_score_gemma":0.0002499771,"threshold_uncertainty_score":0.59907454},"labels":[],"label_agreement":null},{"id":"W2067250880","doi":"10.5555/2820518.2820569","title":"Organizational volatility and post-release defects: a replication case study using data from Google Chrome","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Outsourcing; Popularity; Volatility (finance); Business; Computer science; Directory; Replicate; World Wide Web; Knowledge management; Marketing; Finance; Operating system","score_opus":0.060518256411597054,"score_gpt":0.3147072335686175,"score_spread":0.2541889771570205,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2067250880","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.855729,0.0004243317,0.14252567,0.000026251784,0.00045231322,0.00024494715,0.00003785398,0.00055763376,0.0000020390362],"genre_scores_gemma":[0.88365906,6.128573e-7,0.11596732,0.000012731286,0.00023268227,0.000012076699,0.000067808054,0.000025373824,0.000022319618],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.997693,0.00014382368,0.00031959813,0.0010150134,0.000574502,0.00025408383],"domain_scores_gemma":[0.99533606,0.0011262954,0.00011590874,0.0025307718,0.00064819626,0.0002427966],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.00068839715,0.00019124683,0.00020785684,0.00010523663,0.0003001281,0.00049867405,0.00080743263,0.00007968177,0.000002649698],"category_scores_gemma":[0.009970362,0.00019727221,0.000016035623,0.0006227545,0.00006337526,0.0010179459,0.0013983833,0.00015654242,0.0000040365294],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011385031,0.00012895928,0.9880184,0.000019811168,0.000051553976,0.0010564342,0.008906617,0.00008613446,0.00044629144,0.0000097209395,0.00017272876,0.0010919778],"study_design_scores_gemma":[0.0029920987,0.0012742744,0.6512264,0.00022618334,0.00024249763,0.011568218,0.011941355,0.31504607,0.002561061,0.00061662146,0.0004907244,0.0018144868],"about_ca_topic_score_codex":0.002615226,"about_ca_topic_score_gemma":0.00006906933,"teacher_disagreement_score":0.336792,"about_ca_system_score_codex":0.00018549628,"about_ca_system_score_gemma":0.0004519704,"threshold_uncertainty_score":0.9983691},"labels":[],"label_agreement":null},{"id":"W2074549053","doi":"10.5555/2820518.2820578","title":"An insight into the unresolved questions at stack overflow","year":2015,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Expert finding and Q&A systems","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Stack (abstract data type); Computer science; Popularity; Recall; Data science; Exploratory research; Information retrieval; Theoretical computer science; Artificial intelligence; Cognitive psychology; Programming language; Psychology; Social psychology","score_opus":0.025449479710167613,"score_gpt":0.2739315046489646,"score_spread":0.24848202493879698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2074549053","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6915136,0.0045506996,0.2854517,0.0012505855,0.012039574,0.00032327042,0.0000035827593,0.0022412133,0.0026257667],"genre_scores_gemma":[0.92285013,0.00000617337,0.07339271,0.000114222006,0.00086072064,0.000047497022,0.000007567262,0.00002290916,0.0026980964],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982501,0.00022461326,0.00029674207,0.00043913015,0.0004961813,0.00029321807],"domain_scores_gemma":[0.997941,0.00028711007,0.00013728229,0.0011783635,0.00024885437,0.00020739637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004526268,0.0001902963,0.00019116588,0.00006412197,0.00079997745,0.0005848537,0.0009101943,0.00010842243,0.0000011750496],"category_scores_gemma":[0.0004112383,0.00013384815,0.00005697398,0.00035746256,0.00010583114,0.00062136794,0.00030145224,0.00013040357,0.000029832785],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004190273,0.00016391193,0.35960948,0.00004413115,0.00010299422,0.00017697967,0.5825591,0.00090939115,0.0023987724,0.009927065,0.03247862,0.011587662],"study_design_scores_gemma":[0.0014552312,0.0017270392,0.02251848,0.0005877636,0.00006817612,0.0006645915,0.009140545,0.015763683,0.026475258,0.0019997796,0.9176403,0.0019591178],"about_ca_topic_score_codex":0.0007079636,"about_ca_topic_score_gemma":0.00016725934,"teacher_disagreement_score":0.8851617,"about_ca_system_score_codex":0.00022792612,"about_ca_system_score_gemma":0.00020798424,"threshold_uncertainty_score":0.61528605},"labels":[],"label_agreement":null},{"id":"W2079972013","doi":"10.5555/2664446.2664462","title":"Mining challenge 2012: the Android platform","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Android (operating system); Computer science; Data science; World Wide Web; Android application; Software; Android app; Operating system","score_opus":0.027878367104243282,"score_gpt":0.26460729060075805,"score_spread":0.23672892349651475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2079972013","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6554719,0.018094253,0.30567014,0.0008892875,0.0137130665,0.00049112935,0.0000034424359,0.0035203102,0.0021464773],"genre_scores_gemma":[0.90320504,0.00003198394,0.09394171,0.000046054094,0.0013780737,0.000078803845,0.0000027245221,0.00004143134,0.0012741756],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9976545,0.000055183405,0.00030604674,0.00037580956,0.0007183258,0.0008901617],"domain_scores_gemma":[0.9963207,0.0021232858,0.00010333585,0.0010920875,0.00013904902,0.00022153593],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008334334,0.0002558976,0.00022273758,0.000117940406,0.000549649,0.00038316852,0.0012464271,0.00013136365,0.000008475669],"category_scores_gemma":[0.0015059329,0.00019179728,0.00009172895,0.00049167423,0.000104042985,0.0016013405,0.00060618715,0.00029299545,0.000042851076],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022711029,0.00021949325,0.75481784,0.00017705957,0.00019310026,0.000118720855,0.07469465,0.00013814523,0.00027294864,0.0058926884,0.028730014,0.13472262],"study_design_scores_gemma":[0.0016378714,0.0010046475,0.53405356,0.00067394134,0.000104606734,0.002122976,0.0048396527,0.0023244887,0.018137177,0.000726249,0.43122473,0.0031500868],"about_ca_topic_score_codex":0.000017958595,"about_ca_topic_score_gemma":0.000003901926,"teacher_disagreement_score":0.40249473,"about_ca_system_score_codex":0.00010971735,"about_ca_system_score_gemma":0.00009510924,"threshold_uncertainty_score":0.7821266},"labels":[],"label_agreement":null},{"id":"W2081647265","doi":"10.5555/2664446.2664469","title":"The evolution of the social programmer","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Open Source Software Innovations","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Insourcing; Outsourcing; Crowdsourcing; Crowdsourcing software development; Programmer; Software development; Software; Knowledge management; Computer science; World Wide Web; Social software engineering; Social media; Social software; Social computing; Software engineering; Data science; Business; Software development process; Software construction; Marketing","score_opus":0.013594462858546503,"score_gpt":0.26177028209367786,"score_spread":0.24817581923513135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2081647265","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4802229,0.0013570039,0.503334,0.0027418253,0.009692071,0.0005534236,0.0000040529767,0.00075561396,0.0013391144],"genre_scores_gemma":[0.9608056,4.122696e-7,0.03802035,0.000024509818,0.00050755346,0.000040167353,5.204978e-7,0.000010146691,0.00059073593],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986986,0.000104874314,0.00029279385,0.00014858063,0.00042036732,0.00033481934],"domain_scores_gemma":[0.99850804,0.00034558933,0.00025830275,0.0005674407,0.0002893581,0.00003126381],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0005126092,0.0001055471,0.00010270435,0.000028069304,0.0014693992,0.00020161983,0.0009324281,0.00006686944,6.66772e-7],"category_scores_gemma":[0.00059774745,0.00006293148,0.00008586629,0.00086073985,0.00024112142,0.00047530024,0.00040726355,0.00012967896,0.0000038663097],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000044775275,0.00007398436,0.8333093,0.000016954315,0.000053076084,4.7162393e-7,0.023376273,0.000005665119,0.00022646121,0.0900276,0.003941751,0.048964016],"study_design_scores_gemma":[0.0003199839,0.00009340793,0.8446714,0.000094126524,0.00006106864,0.00009748742,0.0062969956,0.00010745981,0.0071650892,0.0029425307,0.13763516,0.00051524246],"about_ca_topic_score_codex":0.000070481554,"about_ca_topic_score_gemma":0.000007389766,"teacher_disagreement_score":0.48058268,"about_ca_system_score_codex":0.00013810511,"about_ca_system_score_gemma":0.00015949253,"threshold_uncertainty_score":0.99983054},"labels":[],"label_agreement":null},{"id":"W2117757207","doi":"10.5555/2664446.2664448","title":"Towards improving bug tracking systems with game mechanisms","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Agile software development; Tracking system; Quality (philosophy); Tracking (education); Reputation; Software engineering; Computer security; Human–computer interaction; Artificial intelligence","score_opus":0.016023266705648975,"score_gpt":0.2417133949857941,"score_spread":0.22569012828014515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117757207","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12012558,0.0012242072,0.87322944,0.000031256375,0.0033100406,0.00019992857,0.0000016828726,0.0018032527,0.00007461563],"genre_scores_gemma":[0.76841664,0.0000019509573,0.23061411,0.0000109933135,0.0005828506,0.00007445643,0.0000019674148,0.00004864166,0.0002483862],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9971562,0.00008001145,0.00033662346,0.0005129077,0.0009501185,0.00096414314],"domain_scores_gemma":[0.9976283,0.00073581,0.00014867347,0.0009051953,0.00028671566,0.00029535193],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00072699843,0.00031414878,0.00032051932,0.00018747947,0.00027929392,0.00087862945,0.00089969515,0.00014219567,0.0000020334496],"category_scores_gemma":[0.0010998328,0.00026897123,0.00006606511,0.0006197745,0.000057532914,0.0015625285,0.00036731997,0.00029287196,0.000013498029],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007753482,0.00034893092,0.77885026,0.0015614576,0.0004869544,0.00068995263,0.05105057,0.0044261883,0.0118435,0.021892091,0.00084079383,0.12793177],"study_design_scores_gemma":[0.0053996053,0.0048137223,0.5034528,0.00505538,0.00047109902,0.016485304,0.009694713,0.039253246,0.3782895,0.0011590971,0.024366006,0.011559557],"about_ca_topic_score_codex":0.00017489928,"about_ca_topic_score_gemma":0.0000015419542,"teacher_disagreement_score":0.64829105,"about_ca_system_score_codex":0.00022765697,"about_ca_system_score_gemma":0.00020154269,"threshold_uncertainty_score":0.9999763},"labels":[],"label_agreement":null},{"id":"W2131704937","doi":"10.5555/2664446.2664458","title":"Green mining: a methodology of relating software change to power consumption","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Green IT and Sustainability","field":"Engineering","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Software; Computer science; Consumption (sociology); Power consumption; BitTorrent; Power (physics); Operating system; The Internet","score_opus":0.06653757338892163,"score_gpt":0.2953637917156099,"score_spread":0.22882621832668826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131704937","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9704406,0.0013858826,0.025188211,0.000031026615,0.0018900196,0.00025524152,0.000009682086,0.00065249496,0.0001468421],"genre_scores_gemma":[0.8204187,0.000003049008,0.17888115,0.000022109021,0.00036388508,0.00007355347,0.0000073983074,0.00004880596,0.0001813076],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9983762,0.00013944824,0.00047085877,0.00023490792,0.00023222416,0.0005463501],"domain_scores_gemma":[0.99825597,0.0008777983,0.00010614095,0.00039659164,0.0001828114,0.00018069708],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0006982178,0.00024363026,0.0004128785,0.00015442613,0.0001332659,0.000022358232,0.00015830944,0.00022288888,0.000025924006],"category_scores_gemma":[0.0016626265,0.00025453357,0.000097299235,0.00025893474,0.00008043136,0.00034102023,0.00011077708,0.00016044044,0.000008251431],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000023686023,0.000020845559,0.9393493,0.0003686055,0.00006039938,0.000007600598,0.046499543,0.00007873096,0.00062878506,0.000047881873,0.0003235029,0.012591138],"study_design_scores_gemma":[0.0006061072,0.0005080538,0.9636118,0.00045268264,0.00020168797,0.00017116437,0.010906613,0.00014149112,0.012112321,0.0002114428,0.009863535,0.0012131311],"about_ca_topic_score_codex":0.00014602112,"about_ca_topic_score_gemma":0.000022258642,"teacher_disagreement_score":0.15369295,"about_ca_system_score_codex":0.00014040539,"about_ca_system_score_gemma":0.000023940735,"threshold_uncertainty_score":0.9999907},"labels":[],"label_agreement":null},{"id":"W2154813515","doi":"10.5555/2664446.2664465","title":"Do the stars align?: multidimensional analysis of Android's layered architecture","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Advanced Malware Detection Techniques","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Android (operating system); Computer science; Popularity; Architecture; Locality; Operating system; Geography","score_opus":0.011744882525077764,"score_gpt":0.260679922151485,"score_spread":0.24893503962640723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2154813515","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17520955,0.0005688084,0.8228338,0.00007419843,0.0006300961,0.00011189498,0.000008629584,0.0004931983,0.000069875954],"genre_scores_gemma":[0.65096354,0.0000057743905,0.3488045,0.000038072143,0.00008910287,0.000020149351,0.0000032575383,0.000009394107,0.00006622505],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99845374,0.00011722876,0.00034395015,0.000308097,0.00046339992,0.0003135703],"domain_scores_gemma":[0.99795634,0.00063691096,0.00027505914,0.0008368686,0.0002104085,0.000084430816],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036286327,0.00017991122,0.00029690153,0.0003847638,0.00026738874,0.00006606994,0.0004987528,0.00008609431,0.000005058947],"category_scores_gemma":[0.00033872685,0.00012902572,0.00017543067,0.0013655528,0.00011541631,0.0003878422,0.0002894074,0.00014562623,0.0000010336814],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019748775,0.0005805031,0.59841335,0.00015542045,0.0046347817,0.000071667695,0.09131022,0.024542132,0.065459244,0.017969225,0.003730935,0.19293505],"study_design_scores_gemma":[0.0007432618,0.00065414864,0.22003058,0.00020665777,0.0012411473,0.00038544892,0.0021015669,0.004141114,0.71756315,0.002254951,0.04894118,0.0017367799],"about_ca_topic_score_codex":0.000054343232,"about_ca_topic_score_gemma":0.000009063064,"teacher_disagreement_score":0.6521039,"about_ca_system_score_codex":0.00004855823,"about_ca_system_score_gemma":0.000050385195,"threshold_uncertainty_score":0.52615166},"labels":[],"label_agreement":null},{"id":"W2160841184","doi":"10.5555/2664446.2664451","title":"A linked data platform for mining software repositories","year":2012,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Software versioning; Software; World Wide Web; Cloud computing; Software mining; BitTorrent tracker; Database; Software engineering; Software development; Data science; Software construction; Operating system","score_opus":0.05717894938690845,"score_gpt":0.3065874026592379,"score_spread":0.24940845327232944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2160841184","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13775694,0.0034694842,0.8421883,0.00015090476,0.011837075,0.00067713886,0.000070892034,0.0037846996,0.000064557564],"genre_scores_gemma":[0.25117415,0.000013243335,0.74368906,0.000051033705,0.0034828084,0.00025131964,0.00014907452,0.00012078031,0.0010685571],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9949946,0.00007463923,0.0008570282,0.0012662297,0.0011458369,0.0016616931],"domain_scores_gemma":[0.9884138,0.006415865,0.00033695714,0.0037558693,0.0005582807,0.00051919156],"candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0015093263,0.00058731483,0.0006268432,0.0003021492,0.0010094051,0.0010813418,0.0035802468,0.00037209602,0.0000044249036],"category_scores_gemma":[0.014735435,0.0005876362,0.00017309496,0.0009166464,0.00016681362,0.0039003897,0.0023359128,0.00040755083,0.0000168813],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011531488,0.00027077974,0.8696361,0.0008073658,0.00045499994,0.00013035609,0.025875244,0.00016214185,0.0005960972,0.0012263431,0.035757482,0.06496774],"study_design_scores_gemma":[0.0063572805,0.0027215034,0.25285256,0.003050306,0.00069950806,0.0039891293,0.0060442197,0.013528657,0.041016463,0.0013719195,0.6585188,0.009849642],"about_ca_topic_score_codex":0.00006556195,"about_ca_topic_score_gemma":0.0000072780435,"teacher_disagreement_score":0.6227613,"about_ca_system_score_codex":0.0002806435,"about_ca_system_score_gemma":0.0004399906,"threshold_uncertainty_score":0.99995565},"labels":[],"label_agreement":null},{"id":"W2907438654","doi":"","title":"Revisiting \"Programmers' Build Errors\" in the Visual Studio Context","year":2018,"lang":"en","type":"article","venue":"Mining Software Repositories","topic":"Software Engineering Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Deliverable; Computer science; TRACE (psycholinguistics); Context (archaeology); Replicate; Workspace; Studio; Codebase; Code (set theory); Software engineering; Human–computer interaction; Programming language; Source code; World Wide Web; Artificial intelligence; Systems engineering","score_opus":0.01943302576174965,"score_gpt":0.3028428703419978,"score_spread":0.28340984458024815,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2907438654","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8443751,0.0012301386,0.15016328,0.00069528964,0.002043879,0.00035949002,9.05788e-7,0.0009600643,0.00017186532],"genre_scores_gemma":[0.9439716,0.0000041863505,0.054498915,0.00012978293,0.0011835113,0.0000702981,0.0000016158874,0.000023864422,0.000116276955],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9974315,0.0001963661,0.00039144533,0.0005693722,0.0007605413,0.0006507872],"domain_scores_gemma":[0.99636394,0.0023559039,0.00010532224,0.0008074002,0.0002816131,0.00008584592],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012953958,0.00023313187,0.0002573289,0.00016154059,0.00043873998,0.0006225429,0.0014024483,0.0000989351,0.0000030934864],"category_scores_gemma":[0.003834785,0.00018360848,0.0000776602,0.0011631709,0.00020712122,0.0004258702,0.00043986624,0.00034161503,0.000025816189],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016181633,0.00007526017,0.7769428,0.00007361505,0.00005241577,0.00029258715,0.036026098,0.000021419433,0.00013323374,0.001032993,0.0014814598,0.18385194],"study_design_scores_gemma":[0.0027585237,0.0031016704,0.8176125,0.0021558043,0.00007838792,0.0013937706,0.023799116,0.0062352535,0.013200772,0.0005758749,0.12605423,0.0030340562],"about_ca_topic_score_codex":0.00016073766,"about_ca_topic_score_gemma":0.000025895297,"teacher_disagreement_score":0.18081789,"about_ca_system_score_codex":0.00011702125,"about_ca_system_score_gemma":0.00012923143,"threshold_uncertainty_score":0.74873364},"labels":[],"label_agreement":null}]}