{"meta":{"query_hash":"39236c7140f9","filters":{"venue":"Balisage series on markup technologies"},"cohort_total":30,"direct_labels_cover":1,"predictions_cover":30,"exported":30,"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/39236c7140f9","api":"https://metacan.xera.ac/api/v1/cohort?venue=Balisage+series+on+markup+technologies"},"results":[{"id":"W2183772626","doi":"10.4242/balisagevol8.huitfeldt02","title":"The MLCD Overlap Corpus (MOC)","year":2012,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Natural language processing; Artificial intelligence","score_opus":0.033983899520821315,"score_gpt":0.21915949484324737,"score_spread":0.18517559532242606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2183772626","genre_codex":"other","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":[0.15130594,0.0037647274,0.0000024725005,0.0037185957,0.002623054,0.00037092296,0.0001269038,0.0024926043,0.8355948],"genre_scores_gemma":[0.9182095,0.0005262588,0.000036260033,0.00036006526,0.00039611803,0.00008243206,0.0000149402385,0.00003844469,0.08033597],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983544,0.000044839086,0.00029233596,0.00022807861,0.00033479158,0.00074556126],"domain_scores_gemma":[0.9987678,0.0002554292,0.00012875254,0.00071307906,0.000083087936,0.000051856954],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0003808168,0.00030203728,0.00022456785,0.00010004268,0.001420404,0.0013797682,0.00075953215,0.00012522888,0.0005780157],"category_scores_gemma":[0.0003554338,0.00019233733,0.00013660725,0.00006692422,0.0015325535,0.0015459837,0.00029830335,0.00039534018,0.00035184066],"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":[0.00007284719,0.000056132332,0.000506908,0.00001350265,0.0000321963,0.000006135852,0.0014718769,1.7161034e-7,0.000014027501,0.9319479,0.028699718,0.037178572],"study_design_scores_gemma":[0.00011879573,0.00019728756,0.00044084684,0.000027037744,0.000012677885,0.00001041753,0.04229631,6.233385e-7,0.0005729192,0.024651527,0.9313875,0.00028409247],"about_ca_topic_score_codex":0.000034344383,"about_ca_topic_score_gemma":0.00038867258,"teacher_disagreement_score":0.90729636,"about_ca_system_score_codex":0.00006483554,"about_ca_system_score_gemma":0.000016135627,"threshold_uncertainty_score":0.9998796},"labels":[],"label_agreement":null},{"id":"W2273502271","doi":"10.4242/balisagevol8.jordanous01","title":"Contemporary transformation of ancient documents for recording and retrieving maximum information: when one form of markup is not enough","year":2012,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Prince Edward Island","funders":"","keywords":"Markup language; Arabic; Computer science; Annotation; RDF; Transformation (genetics); Publication; Information retrieval; World Wide Web; Natural language processing; Linguistics; Artificial intelligence; XML; Semantic Web","score_opus":0.03342159134163366,"score_gpt":0.2468725924931488,"score_spread":0.21345100115151514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2273502271","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":[0.5952048,0.005203496,0.35429418,0.021000996,0.0015782624,0.003496925,0.00029343276,0.0020919784,0.01683597],"genre_scores_gemma":[0.9412752,0.0003447703,0.058007885,0.0002453901,0.000011176466,0.000055423978,0.000010166853,0.000007052747,0.00004293726],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99841356,0.000027420874,0.00066815666,0.00019528883,0.00032860364,0.00036699686],"domain_scores_gemma":[0.99861366,0.0002042842,0.0004506506,0.0005411745,0.00015555431,0.00003470028],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007361134,0.00020825895,0.00040753066,0.0003118312,0.00013596386,0.000091072565,0.00064476504,0.00020450848,0.000007979414],"category_scores_gemma":[0.00042606686,0.00018405734,0.00008792336,0.00032031586,0.00023076936,0.005047609,0.0002835873,0.00012044573,0.0000019581366],"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":[0.0012139303,0.0002794318,0.011611876,0.0027349477,0.00024233198,0.0000019701008,0.026927976,0.0000030252722,0.0073814043,0.23731096,0.017260894,0.6950313],"study_design_scores_gemma":[0.0029830933,0.0022352282,0.018639531,0.0008994604,0.00006865572,0.000041716536,0.025915096,0.0009805926,0.7926175,0.057156853,0.097407416,0.0010548853],"about_ca_topic_score_codex":0.00003671659,"about_ca_topic_score_gemma":0.0000027999838,"teacher_disagreement_score":0.78523606,"about_ca_system_score_codex":0.000048996102,"about_ca_system_score_gemma":0.00004021739,"threshold_uncertainty_score":0.75056404},"labels":[],"label_agreement":null},{"id":"W2277519993","doi":"10.4242/balisagevol5.briquet01","title":"A Virtualization-Based Retrieval and Update API for XML-Encoded Corpora","year":2010,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Computer science; XML; Information retrieval; Markup language; World Wide Web; Simple API for XML; Well-formed document; XML framework; SGML; Representation (politics); Virtualization; Streaming XML; Document Structure Description; Document type definition; XML Signature; Operating system; Cloud computing","score_opus":0.009566712296790668,"score_gpt":0.2309762155939664,"score_spread":0.2214095032971757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2277519993","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":[0.32795975,0.0006642248,0.6237316,0.030856375,0.0023925886,0.0017793064,0.000111839414,0.010712606,0.0017917079],"genre_scores_gemma":[0.8510806,0.00010264954,0.14777644,0.00063500105,0.000034015844,0.000079875215,0.00001690406,0.000023095585,0.0002514478],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99835736,0.000036302783,0.00030281465,0.0006351418,0.0002527969,0.0004155801],"domain_scores_gemma":[0.998151,0.00038954322,0.0001865095,0.0010846498,0.00014237336,0.000045935114],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003981579,0.00028806072,0.00032107328,0.00026093388,0.00030086344,0.00034797846,0.0012463261,0.00035357746,0.000012079687],"category_scores_gemma":[0.0016781552,0.00024287237,0.00007559519,0.00050621846,0.00059767725,0.00057262933,0.00036366258,0.0002850568,0.00001153552],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043393022,0.000103811646,0.0016594111,0.00010439143,0.00003731521,0.00004092025,0.00016217984,0.000023751674,0.01748318,0.9360729,0.00584818,0.038030013],"study_design_scores_gemma":[0.0035142594,0.002218613,0.007143871,0.00014614267,0.00006829375,0.00011817523,0.0026579425,0.039305586,0.4659447,0.3812163,0.09558094,0.0020851798],"about_ca_topic_score_codex":0.000009271496,"about_ca_topic_score_gemma":0.000069862566,"teacher_disagreement_score":0.55485666,"about_ca_system_score_codex":0.00001944208,"about_ca_system_score_gemma":0.000084056126,"threshold_uncertainty_score":0.9904048},"labels":[],"label_agreement":null},{"id":"W2397784846","doi":"10.4242/balisagevol15.gotti01","title":"XSDGuide - Automated Generation of Web Interfaces from XML Schemas: A Case Study for Suspicious Activity Reporting","year":2015,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Validator; XML Schema (W3C); XML; XML validation; Document Structure Description; World Wide Web; Streaming XML; XML Schema Editor; Schema (genetic algorithms); Efficient XML Interchange; JavaScript; XML framework; User interface; Information retrieval; Document type definition; Programming language","score_opus":0.13638877290988888,"score_gpt":0.34546901583523076,"score_spread":0.20908024292534189,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2397784846","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":[0.98060054,0.00028266571,0.0120129185,0.0007998496,0.00046099655,0.0007894548,0.00002522162,0.004869046,0.00015933033],"genre_scores_gemma":[0.9525188,0.000008876692,0.04720482,0.00002264718,0.000034989727,0.0001471543,0.000005488139,0.000010988231,0.000046246456],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9977835,0.00010855733,0.0008156343,0.00064591924,0.00030854746,0.000337884],"domain_scores_gemma":[0.9973204,0.00028354183,0.001049141,0.0010317344,0.00027664733,0.000038503415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009230004,0.00028748045,0.0005360008,0.00018071565,0.0001610399,0.00016677765,0.000784091,0.00022300954,0.0000019260058],"category_scores_gemma":[0.0045209583,0.00021909203,0.000082628176,0.00041983297,0.00017626361,0.0007509858,0.0006600651,0.00016981906,0.0000021502342],"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":[0.0010791542,0.0031583016,0.078646764,0.0003214036,0.0010857037,0.011777381,0.013860974,0.0007229703,0.3905424,0.014194813,0.025570793,0.45903936],"study_design_scores_gemma":[0.0021105204,0.0040047183,0.0025387015,0.00012961213,0.00009911097,0.0008821617,0.08921015,0.11748589,0.7774275,0.0043443316,0.00080648786,0.00096080796],"about_ca_topic_score_codex":0.00093535625,"about_ca_topic_score_gemma":0.0018620351,"teacher_disagreement_score":0.45807853,"about_ca_system_score_codex":0.000087925524,"about_ca_system_score_gemma":0.00010372295,"threshold_uncertainty_score":0.8934314},"labels":[],"label_agreement":null},{"id":"W2405269923","doi":"10.4242/balisagevol1.muldner01","title":"Secure Publishing using Schema-level Role-based Access Control Policies for Fragments of XML Documents","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Access Control and Trust","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Acadia University","funders":"","keywords":"Computer science; Document Structure Description; XML Encryption; XML Schema Editor; XML; Schema (genetic algorithms); Access control; Efficient XML Interchange; Parsing; Encryption; XML validation; Information retrieval; Database; Computer security; World Wide Web; Programming language","score_opus":0.05453315838117753,"score_gpt":0.3539137275040181,"score_spread":0.2993805691228406,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2405269923","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":[0.83827597,0.0023835988,0.020731905,0.07508408,0.0012122624,0.005714152,0.0015613686,0.0044300924,0.05060654],"genre_scores_gemma":[0.99548376,0.00007745362,0.0031730353,0.00080068485,0.00010164715,0.00008667574,0.000019104424,0.000021736501,0.00023590647],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.997585,0.000108373766,0.000472201,0.00043253356,0.00065826956,0.0007436157],"domain_scores_gemma":[0.99836457,0.00028485496,0.0004661477,0.00048160768,0.00033237116,0.000070439506],"candidate_categories":["metaepi_narrow","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0007198096,0.0003165448,0.0005248498,0.0003875064,0.0007761156,0.0010992327,0.0016791922,0.00049697806,0.00004411417],"category_scores_gemma":[0.0024253991,0.00028425254,0.00018523617,0.0006969627,0.000656103,0.0034340043,0.0001495762,0.0002899482,0.0000019113604],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004005722,0.0012770576,0.07100216,0.00040461624,0.00067282235,0.000040500625,0.0041831224,0.0014187878,0.016271321,0.56323916,0.007858248,0.32962644],"study_design_scores_gemma":[0.025605762,0.0030856626,0.05366618,0.0018926443,0.00080671406,0.000008621634,0.11354364,0.0031400372,0.15800321,0.4356367,0.19984297,0.004767846],"about_ca_topic_score_codex":0.0011379337,"about_ca_topic_score_gemma":0.00044260526,"teacher_disagreement_score":0.3248586,"about_ca_system_score_codex":0.00017163072,"about_ca_system_score_gemma":0.00020886239,"threshold_uncertainty_score":0.99996096},"labels":[],"label_agreement":null},{"id":"W2406907785","doi":"10.4242/balisagevol4.cameron01","title":"Parallel Bit Stream Technology as a Foundation for XML Parsing Performance","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Parsing; SIMD; Byte; Compiler; XML; Bitstream; Automatic summarization; Programming language; Simple API for XML; Computer architecture; Stream processing; Parallel computing; Operating system; Efficient XML Interchange; Artificial intelligence; XML Signature","score_opus":0.013928012532668961,"score_gpt":0.25957456622759684,"score_spread":0.24564655369492788,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2406907785","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":[0.076266415,0.0006896112,0.8877546,0.020107932,0.00061445683,0.0014611441,0.000046113,0.008610544,0.0044491803],"genre_scores_gemma":[0.57556325,0.0006270467,0.4222359,0.00030592657,0.00004916838,0.00028031907,0.000043443295,0.000023152277,0.0008717495],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99804336,0.000023962873,0.00039392806,0.00070058176,0.00024827762,0.0005898874],"domain_scores_gemma":[0.9982107,0.000081907645,0.00023151614,0.0013072371,0.00013276588,0.00003581983],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00023647898,0.00034056214,0.00036206914,0.00045917105,0.00048529473,0.00014250446,0.0009805913,0.00028736636,0.0000067203796],"category_scores_gemma":[0.00042477125,0.00030061504,0.000080441445,0.00086254935,0.00029989312,0.0018240401,0.00028254834,0.0002458699,0.000046906505],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000827059,0.000047010068,0.00011995165,0.000034427412,0.000011547934,0.0000111118725,0.0000532651,0.0000746724,0.0011955856,0.57949734,0.00055621535,0.4183162],"study_design_scores_gemma":[0.0017349113,0.005963606,0.0013075243,0.00059192406,0.000027606544,0.00031201867,0.0038470624,0.0050096884,0.11044591,0.20983577,0.65922683,0.0016971411],"about_ca_topic_score_codex":0.0000067517753,"about_ca_topic_score_gemma":0.000012496385,"teacher_disagreement_score":0.6586706,"about_ca_system_score_codex":0.00009727587,"about_ca_system_score_gemma":0.00006591865,"threshold_uncertainty_score":0.99994457},"labels":[],"label_agreement":null},{"id":"W2406990242","doi":"10.4242/balisagevol3.muldner01","title":"XSAQCT: XML Queryable Compressor","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Acadia University","funders":"","keywords":"Computer science; XML; Database; Gas compressor; Accounting; Software engineering; World Wide Web; Engineering; Business; Mechanical engineering","score_opus":0.009228316138441564,"score_gpt":0.22979839607798255,"score_spread":0.22057007993954098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2406990242","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":[0.007854982,0.004135051,0.8594417,0.03569281,0.00155353,0.0012708735,0.00019918055,0.020832,0.069019884],"genre_scores_gemma":[0.67662585,0.00086078316,0.31550273,0.0015271184,0.00010946504,0.00013250973,0.000039772265,0.00004016031,0.005161624],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9979995,0.00005407866,0.00035511228,0.00066786766,0.00034996064,0.00057350274],"domain_scores_gemma":[0.99779075,0.000086998836,0.00016384434,0.0018238875,0.00008073426,0.00005376842],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00022012027,0.00034241416,0.000385192,0.00021995483,0.00032291992,0.00016920286,0.0013419419,0.00019535384,0.000021042597],"category_scores_gemma":[0.0002624443,0.00028604016,0.00009227566,0.0006097569,0.00025413244,0.0017796458,0.0003749282,0.00030196278,0.00009541577],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040969306,0.00007348177,0.0000634099,0.00003080993,0.000013012948,0.00010128628,0.00009165632,0.000052570405,0.003329331,0.8984647,0.011632847,0.08610597],"study_design_scores_gemma":[0.00043144752,0.0007849945,0.00097202923,0.00016115914,0.000006476218,0.0001020093,0.0010815774,0.00036647666,0.08438282,0.03885411,0.8720862,0.0007706521],"about_ca_topic_score_codex":0.000023436625,"about_ca_topic_score_gemma":0.000015486685,"teacher_disagreement_score":0.8604534,"about_ca_system_score_codex":0.000059438786,"about_ca_system_score_gemma":0.000035499805,"threshold_uncertainty_score":0.9999592},"labels":[],"label_agreement":null},{"id":"W2413440334","doi":"10.4242/balisagevol7.mercier01","title":"Including XSLT stylesheets testing in continuous integration process","year":2011,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"XSLT; Code refactoring; Computer science; Programming language; Java; Process (computing); Software engineering; XML; Software; Operating system","score_opus":0.12505217525480483,"score_gpt":0.25729339954486585,"score_spread":0.13224122429006102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2413440334","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":[0.5103761,0.00010827689,0.0000051820407,0.00014065378,0.00021780426,0.00025513594,0.00002624617,0.0012146548,0.48765597],"genre_scores_gemma":[0.9956062,0.00002114609,0.00035488515,0.000116075666,0.00006426761,0.00010426795,0.000014298713,0.000034193778,0.0036846886],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9985477,0.0000356118,0.00040460363,0.00036169696,0.00022237282,0.00042805434],"domain_scores_gemma":[0.99920505,0.00010443564,0.00016139398,0.00033745734,0.00016630038,0.000025363304],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00029783213,0.00029437526,0.00030607684,0.00035701846,0.00035956706,0.00053714454,0.0005220834,0.00013851021,0.0004261201],"category_scores_gemma":[0.0008704011,0.00024816877,0.000056441062,0.00016069229,0.0006198692,0.0016380025,0.00017966032,0.00042596943,0.00004016075],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020332377,0.00022797953,0.005558615,0.000103649225,0.000024096253,0.00012551501,0.030133404,0.0000017498917,0.00013073038,0.850168,0.0006085164,0.112714425],"study_design_scores_gemma":[0.0010238644,0.0021183935,0.006195171,0.001386244,0.000037069716,0.000032619435,0.6389791,0.000056487253,0.015328651,0.2912559,0.041893072,0.0016934358],"about_ca_topic_score_codex":0.00018814689,"about_ca_topic_score_gemma":0.00234173,"teacher_disagreement_score":0.6088457,"about_ca_system_score_codex":0.000072213625,"about_ca_system_score_gemma":0.000025414605,"threshold_uncertainty_score":0.9999971},"labels":[],"label_agreement":null},{"id":"W2414491991","doi":"10.4242/balisagevol3.sperberg-mcqueen01","title":"Formal and informal meaning from documents through skeleton sentences","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Meaning (existential); Sentence; Set (abstract data type); Natural language; Natural (archaeology); Semantics (computer science); Linguistics; Natural language processing; Complement (music); Natural language generation; Artificial intelligence; Epistemology; Programming language; Philosophy; History","score_opus":0.009119940111175638,"score_gpt":0.2504557326011053,"score_spread":0.24133579248992967,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2414491991","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":[0.48336208,0.016060883,0.38473114,0.031399712,0.0010870065,0.0015894623,0.000068617956,0.042267594,0.0394335],"genre_scores_gemma":[0.60991925,0.0004514674,0.3888568,0.0006106302,0.000020229063,0.000013354371,0.000006898541,0.000006836551,0.00011453869],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9983379,0.000026998847,0.0002948861,0.00047130312,0.0003646019,0.0005042737],"domain_scores_gemma":[0.99901366,0.00006695507,0.00017270699,0.000656649,0.000058149166,0.000031903084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018639446,0.00030017504,0.0002650393,0.00015668491,0.00033268586,0.0005332311,0.0014243547,0.00024190555,0.000007456245],"category_scores_gemma":[0.00022914805,0.00023846379,0.0000502421,0.00042035565,0.0002273206,0.0054186983,0.0007766104,0.00037374577,0.000007132189],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052386924,0.000028189235,0.0003894941,0.000017645909,0.000017620654,0.000062558094,0.0010836405,0.0000012460971,0.002123934,0.23674232,0.0008852439,0.7585957],"study_design_scores_gemma":[0.0005968145,0.00127133,0.0024372293,0.00034527326,0.000017789642,0.00010154544,0.0037255029,0.00050280784,0.26710102,0.7102609,0.01263782,0.0010019371],"about_ca_topic_score_codex":0.00008978221,"about_ca_topic_score_gemma":0.000026285594,"teacher_disagreement_score":0.75759375,"about_ca_system_score_codex":0.00003779725,"about_ca_system_score_gemma":0.000023263057,"threshold_uncertainty_score":0.97242707},"labels":[],"label_agreement":null},{"id":"W2414577476","doi":"10.4242/balisagevol3.huitfeldt01","title":"Markup Meaning and Mereology","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Philosophy and Theoretical Science","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Mereology; Markup language; Meaning (existential); Computer science; Information retrieval; Epistemology; Philosophy; World Wide Web; XML","score_opus":0.013978655467464386,"score_gpt":0.27426816607449706,"score_spread":0.26028951060703265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2414577476","genre_codex":"other","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":[0.11187103,0.0008361607,0.00020663167,0.025111306,0.00044561585,0.00025443014,0.00001639057,0.0018744587,0.859384],"genre_scores_gemma":[0.9964378,0.00015629585,0.0012508405,0.0010218953,0.000039849245,0.000023311308,0.00000260041,0.000010805791,0.0010566305],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9984075,0.000101851845,0.00022815475,0.00055887824,0.00016693867,0.0005367139],"domain_scores_gemma":[0.99904346,0.00017911504,0.00008003641,0.0006106049,0.000027067344,0.00005968711],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000334443,0.00024287232,0.00027936278,0.00016544863,0.0002127459,0.000050426326,0.0005504826,0.00033144647,0.00048462927],"category_scores_gemma":[0.00044308195,0.00019755567,0.000052719366,0.00037473804,0.0013947585,0.00019129127,0.00017150333,0.00036467137,0.00012106105],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018603503,0.00005209512,0.00023576093,0.000003527366,0.000014303038,0.000082089326,0.0002008599,2.800578e-7,0.000709241,0.89438444,0.00095257815,0.10317881],"study_design_scores_gemma":[0.00031937243,0.0009775286,0.008527313,0.000025322426,0.000014578439,0.00012528643,0.0020269612,0.0000050394847,0.0023930713,0.971748,0.013526677,0.00031087734],"about_ca_topic_score_codex":0.000008580575,"about_ca_topic_score_gemma":0.000005648221,"teacher_disagreement_score":0.8845667,"about_ca_system_score_codex":0.000023327695,"about_ca_system_score_gemma":0.00000789668,"threshold_uncertainty_score":0.80560863},"labels":[],"label_agreement":null},{"id":"W2415938545","doi":"10.4242/balisagevol10.nordin01","title":"Markup and Canada's National Model Building Codes","year":2013,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Optics and Image Analysis","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Markup language; Political science; World Wide Web; XML","score_opus":0.008138877558754322,"score_gpt":0.1904909705946607,"score_spread":0.18235209303590638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2415938545","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":[0.8849738,0.00037867337,0.0011174375,0.03074681,0.00021815853,0.0005327565,0.00004970684,0.0013624033,0.08062025],"genre_scores_gemma":[0.99190724,0.00013371007,0.004740984,0.0015403454,0.00008616495,0.000059361126,0.000021593492,0.000024568757,0.0014860106],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987417,0.0000045544125,0.00022882402,0.00033470834,0.00037234958,0.0003178569],"domain_scores_gemma":[0.9992144,0.00004533279,0.00013491733,0.00026373702,0.00033160252,0.000009973503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00016690374,0.00022921943,0.00023279824,0.00029561363,0.00027881475,0.0007218507,0.00034061735,0.00010619565,0.00013540385],"category_scores_gemma":[0.00059715897,0.0002002084,0.00004708555,0.00032256378,0.0001768423,0.0015311551,0.00039053784,0.00016321898,0.000026728672],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009190689,0.00014239212,0.020082306,0.00042264722,0.00032543833,0.000052541513,0.000024047538,0.0022375898,0.008151945,0.5326008,0.34106573,0.09480263],"study_design_scores_gemma":[0.0013560619,0.000085146006,0.025455698,0.00026750087,0.00028735795,0.000021108865,0.009749011,0.35769227,0.004782252,0.3310983,0.26633224,0.002873085],"about_ca_topic_score_codex":0.09430878,"about_ca_topic_score_gemma":0.07843253,"teacher_disagreement_score":0.35545468,"about_ca_system_score_codex":0.000068721776,"about_ca_system_score_gemma":0.00006082823,"threshold_uncertainty_score":0.9383837},"labels":[],"label_agreement":null},{"id":"W2416130262","doi":"10.4242/balisagevol5.huitfeldt01","title":"Extension of the type/token distinction to document structure","year":2010,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","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":"Université de Montréal","funders":"","keywords":"Security token; Computer science; SGML; Document type definition; Extension (predicate logic); Type (biology); sort; Markup language; XML; Word (group theory); Information retrieval; Document Structure Description; Natural language processing; Linguistics; Programming language; World Wide Web","score_opus":0.016206641282411113,"score_gpt":0.21752586626501513,"score_spread":0.20131922498260402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2416130262","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":[0.92742115,0.000034574467,0.0000011804129,0.0019654385,0.0018991191,0.0002885141,0.00009990594,0.00039734895,0.06789279],"genre_scores_gemma":[0.9896127,0.000009476288,0.00015375175,0.00020963758,0.00010877559,0.00001203777,0.000009933551,0.000017351298,0.009866346],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990902,0.00001893506,0.00021273158,0.00022319063,0.00024664207,0.00020833219],"domain_scores_gemma":[0.9990404,0.000039999835,0.00010121951,0.00065053976,0.00014553587,0.000022277485],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000106699124,0.00017697107,0.0001677762,0.00009222528,0.00035149322,0.00029448819,0.000510954,0.00010418813,0.0010949008],"category_scores_gemma":[0.00035081027,0.000109007786,0.00007174325,0.000087980254,0.0005642004,0.00039619455,0.00029495923,0.00035997713,0.000023946135],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014031946,0.00003553838,0.00034206314,0.000034092063,0.0000185354,0.000003113723,0.0016399728,0.0000027881763,0.0052960655,0.9648731,0.008526277,0.019088151],"study_design_scores_gemma":[0.00016170152,0.00049412734,0.005232934,0.00008349543,0.00002204371,0.00000934184,0.014526421,0.0000010559401,0.019710025,0.1048801,0.85456085,0.0003179001],"about_ca_topic_score_codex":0.00004986874,"about_ca_topic_score_gemma":0.0019319875,"teacher_disagreement_score":0.859993,"about_ca_system_score_codex":0.000021676315,"about_ca_system_score_gemma":0.00001520227,"threshold_uncertainty_score":0.9998182},"labels":[],"label_agreement":null},{"id":"W2418003691","doi":"10.4242/balisagevol7.david01","title":"Challenges and Potential of Local Loading of XML Ebooks","year":2011,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Mathematics, Computing, and Information Processing","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Council of University Libraries; University of Toronto","funders":"","keywords":"Computer science; XML; World Wide Web; XML validation; Efficient XML Interchange; XSLT; XML Schema Editor; Presentation (obstetrics); XML framework; Upload; Streaming XML; Simple API for XML; Multimedia; XML Signature","score_opus":0.03435968104011919,"score_gpt":0.22018607522563133,"score_spread":0.18582639418551214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2418003691","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":[0.08585947,0.001989849,0.7638897,0.0003988255,0.00030632145,0.00028861634,0.000004079618,0.0015552044,0.14570798],"genre_scores_gemma":[0.94076145,0.0002779537,0.058873296,0.000015436273,0.000004311678,0.000003946326,2.2111121e-7,0.0000052826126,0.000058127065],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9990777,0.000018275703,0.00037026106,0.00018374163,0.00017919188,0.00017083342],"domain_scores_gemma":[0.99909616,0.000038454393,0.000318885,0.0004380838,0.000090661066,0.00001777991],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002768659,0.00013270642,0.000240034,0.00020314091,0.00006645243,0.000028582908,0.0006456922,0.00011502732,0.0000047244375],"category_scores_gemma":[0.000095422925,0.00011295531,0.00003944737,0.00008653528,0.00043611944,0.00046152738,0.0004892894,0.00010188385,0.0000015301601],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013647875,0.000026186919,0.000013367963,0.0003866748,0.00001544022,0.0000032654978,0.0038232857,0.0000033922338,0.0005051013,0.44821918,0.000026772466,0.5469637],"study_design_scores_gemma":[0.00058028114,0.0009144872,0.0016897087,0.00074632565,0.000022587876,0.00011100718,0.03247121,0.009516918,0.68270653,0.26984626,0.0008162089,0.00057847623],"about_ca_topic_score_codex":0.0000034993532,"about_ca_topic_score_gemma":0.0000016765799,"teacher_disagreement_score":0.85490197,"about_ca_system_score_codex":0.000008526799,"about_ca_system_score_gemma":0.000019994452,"threshold_uncertainty_score":0.4606184},"labels":[],"label_agreement":null},{"id":"W2471310903","doi":"10.4242/balisagevol9.zhao01","title":"Quality Control Practice for Scholars Portal, an XML-based E-journals Repository","year":2012,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Mathematics, Computing, and Information Processing","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":true,"ca_institutions":"Ontario Council of University Libraries","funders":"","keywords":"Computer science; Workflow; XML; World Wide Web; Information retrieval; Set (abstract data type); Quality (philosophy); Control (management); Database; Programming language","score_opus":0.03797933385401914,"score_gpt":0.33522051458186375,"score_spread":0.2972411807278446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2471310903","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":[0.014549036,0.0008169658,0.96480495,0.0039115953,0.000917256,0.00062431913,0.00001446547,0.0030581935,0.011303236],"genre_scores_gemma":[0.7532617,0.000025540472,0.24515039,0.0012444865,0.00007944224,0.000071060334,0.000004710709,0.000016516116,0.0001461499],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976797,0.0001679998,0.00080236036,0.00031842486,0.00048908003,0.00054243696],"domain_scores_gemma":[0.99667233,0.00067207374,0.0010150276,0.0010995462,0.0004442485,0.00009674957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030953446,0.00026367395,0.00035160908,0.000248176,0.00062980276,0.00089898315,0.0012918685,0.0002375152,0.0000103934735],"category_scores_gemma":[0.004914612,0.00022956618,0.00011562289,0.0003342869,0.00017460439,0.008154406,0.00016574455,0.00034743195,0.000011266064],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042873647,0.0015794537,0.0039903633,0.0010078275,0.00020560643,0.000030852985,0.003707982,0.0005002764,0.0036959364,0.7398884,0.011733911,0.23323065],"study_design_scores_gemma":[0.009491127,0.004509995,0.0106126685,0.0015734893,0.00028117842,0.001375259,0.04893577,0.1094335,0.26137584,0.099243425,0.44708633,0.0060813995],"about_ca_topic_score_codex":0.00000438059,"about_ca_topic_score_gemma":8.914322e-7,"teacher_disagreement_score":0.73871267,"about_ca_system_score_codex":0.00006258672,"about_ca_system_score_gemma":0.00010736553,"threshold_uncertainty_score":0.9361437},"labels":[],"label_agreement":null},{"id":"W2474486350","doi":"10.4242/balisagevol1.marcoux01","title":"Graph characterization of overlap-only TexMECS and other overlapping markup formalisms","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Markup language; Computer science; Rotation formalisms in three dimensions; Graph; Programming language; Serialization; Serializability; Document Structure Description; Theoretical computer science; XML; Information retrieval; World Wide Web; Database transaction; Mathematics","score_opus":0.009800205548426531,"score_gpt":0.21770388659977685,"score_spread":0.20790368105135032,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2474486350","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":[0.9517115,0.000621292,0.031843387,0.0049148314,0.0004035293,0.00052859815,0.00004151483,0.0028341813,0.007101216],"genre_scores_gemma":[0.9860347,0.00053435576,0.012414212,0.0007140414,0.000021061544,0.000014211163,0.0000059307886,0.000013928622,0.00024756594],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99832124,0.00004955258,0.00040422162,0.0004923248,0.00030998688,0.00042266506],"domain_scores_gemma":[0.9986502,0.00008447533,0.00027822447,0.00088438165,0.0000690906,0.000033622353],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0002646961,0.0003063099,0.000406155,0.0003706075,0.00015676687,0.00017753131,0.0009746651,0.0002718904,0.000008621354],"category_scores_gemma":[0.00021533993,0.00025984074,0.00008909647,0.00062152866,0.00032230536,0.001126951,0.00032481784,0.00019189868,0.0000044407457],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015762374,0.00014071503,0.011757055,0.00010731616,0.0000555309,0.000040012128,0.0007126403,0.000004334709,0.1036245,0.6224313,0.0004224891,0.2605465],"study_design_scores_gemma":[0.001893607,0.003138113,0.53776693,0.00068068533,0.00006469654,0.0003962133,0.0050506447,0.0010410125,0.2559381,0.15622044,0.03575196,0.002057586],"about_ca_topic_score_codex":0.000015548523,"about_ca_topic_score_gemma":0.0000102662225,"teacher_disagreement_score":0.52600986,"about_ca_system_score_codex":0.000025665222,"about_ca_system_score_gemma":0.000029153913,"threshold_uncertainty_score":0.9999854},"labels":[],"label_agreement":null},{"id":"W2499820415","doi":"10.4242/balisagevol6.ruus01","title":"A brief history of markup of social science data: from punched cards to \"the life cycle\" approach","year":2010,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Metadata; Computer science; Markup language; Raw data; Documentation; Software; Metadata repository; Process (computing); Generator (circuit theory); World Wide Web; Data science; Information retrieval; XML","score_opus":0.06953390648258413,"score_gpt":0.3141411563154728,"score_spread":0.24460724983288867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2499820415","genre_codex":"other","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":[0.22358994,0.0031197798,0.22530976,0.123185664,0.009272612,0.007339443,0.003556907,0.006920869,0.39770502],"genre_scores_gemma":[0.8828949,0.00024994725,0.11424727,0.0003990428,0.00011106128,0.00012795925,0.000071958566,0.000028998265,0.0018688627],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99617034,0.0001584931,0.00046957383,0.0011418787,0.0015050697,0.00055465003],"domain_scores_gemma":[0.9932579,0.00035429915,0.00042433102,0.005622626,0.0002601824,0.00008071033],"candidate_categories":["metaresearch","sts","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0034955386,0.00025458177,0.00040191785,0.00053578,0.0002413153,0.0006275294,0.018123822,0.00016957753,0.00003246575],"category_scores_gemma":[0.009287612,0.00019657012,0.00006761273,0.0015996193,0.0028299522,0.013280578,0.011688736,0.00066206174,0.000011952679],"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":[0.00030983353,0.00045590536,0.0005243242,0.00018487882,0.00019578733,0.000023614279,0.0014775645,0.000022904087,0.039336964,0.7279038,0.13794199,0.09162244],"study_design_scores_gemma":[0.00053116714,0.00035459123,0.017486408,0.000035516776,0.000055036573,0.000006431896,0.004859687,0.0029895217,0.009755416,0.003068004,0.9602378,0.0006204281],"about_ca_topic_score_codex":0.00044523497,"about_ca_topic_score_gemma":0.00013002586,"teacher_disagreement_score":0.8222958,"about_ca_system_score_codex":0.00013474324,"about_ca_system_score_gemma":0.000590628,"threshold_uncertainty_score":0.9998838},"labels":[],"label_agreement":null},{"id":"W2500534248","doi":"10.4242/balisagevol15.marcoux01","title":"Applying intertextual semantics to Cyberjustice","year":2015,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Artificial Intelligence in Law","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; nobody; Semantics (computer science); Process (computing); Forcing (mathematics); Project management; Engineering; Systems engineering; Programming language","score_opus":0.09485323916849173,"score_gpt":0.3652355035328645,"score_spread":0.27038226436437274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2500534248","genre_codex":"other","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":[0.26060674,0.0005157031,0.015168922,0.07684808,0.0053240634,0.0043544103,0.00006823777,0.013601259,0.62351257],"genre_scores_gemma":[0.9883033,0.000112202615,0.006968561,0.0006870067,0.00016952735,0.00027111685,0.0000028914567,0.000030196667,0.003455171],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978731,0.00012689992,0.0003227349,0.00042029915,0.0006171045,0.00063985714],"domain_scores_gemma":[0.9986931,0.00024358243,0.000098484605,0.00056551356,0.00024164739,0.00015766198],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009607525,0.00023559644,0.000264828,0.00022317667,0.00050533155,0.00025135896,0.0010946316,0.00031002867,0.00009375107],"category_scores_gemma":[0.0071456125,0.00021962455,0.00006574685,0.00082367886,0.0011446113,0.00049236766,0.00044357797,0.00031079614,0.000810117],"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":[0.00019948027,0.00009930884,0.0012754451,0.000018347688,0.000029946326,0.00007886726,0.012987953,0.00012174949,0.000478642,0.70243007,0.038195934,0.24408424],"study_design_scores_gemma":[0.000053107782,0.00026378257,0.000032678392,0.00006389346,0.000016573085,0.000003901947,0.25883237,0.000021482214,0.010065481,0.02409785,0.7061858,0.00036304866],"about_ca_topic_score_codex":0.0009468342,"about_ca_topic_score_gemma":0.004012994,"teacher_disagreement_score":0.7276966,"about_ca_system_score_codex":0.00025895386,"about_ca_system_score_gemma":0.00013148304,"threshold_uncertainty_score":0.9999679},"labels":[],"label_agreement":null},{"id":"W2516342496","doi":"10.4242/balisagevol10.marcoux01","title":"Modeling overlapping structures","year":2013,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science","score_opus":0.03674893105993818,"score_gpt":0.21381884489443073,"score_spread":0.17706991383449255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2516342496","genre_codex":"other","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":[0.48715132,0.00043728275,0.000039286115,0.0016506964,0.0006072589,0.0003965167,0.00006003013,0.0029566793,0.50670093],"genre_scores_gemma":[0.9855974,0.0000712771,0.0002811879,0.0003761902,0.00015575741,0.000069114554,0.000021498701,0.000039434155,0.013388147],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99864,0.0000204407,0.00029211596,0.00033610352,0.0002601973,0.00045114913],"domain_scores_gemma":[0.99922884,0.0000463827,0.00006650702,0.0005013392,0.00012145848,0.00003548668],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00008681318,0.0002987493,0.000259403,0.00020010454,0.0006042869,0.0018349027,0.0005688477,0.00012872613,0.003544026],"category_scores_gemma":[0.00016220329,0.00023951764,0.00011155356,0.00005222445,0.00052619114,0.0018923713,0.00023307724,0.0003419011,0.0002547734],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021238147,0.000023517936,0.000021638201,0.00003397864,0.000031329553,0.000008584693,0.0015867325,0.000091454946,0.00008269127,0.962309,0.006279959,0.029509865],"study_design_scores_gemma":[0.00030766,0.0003882331,0.00012331086,0.00013645136,0.000019979603,0.000015139882,0.11438871,0.0010126231,0.0005914132,0.5737741,0.30834007,0.00090230285],"about_ca_topic_score_codex":0.00022347516,"about_ca_topic_score_gemma":0.00023231766,"teacher_disagreement_score":0.49844608,"about_ca_system_score_codex":0.000046273977,"about_ca_system_score_gemma":0.000011783641,"threshold_uncertainty_score":0.9992013},"labels":[],"label_agreement":null},{"id":"W2522795258","doi":"10.4242/balisagevol10.cameron01","title":"icXML: Accelerating a Commercial XML Parser Using SIMD and Multicore Technologies","year":2013,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; SIMD; XML; Parsing; Pipeline (software); Simple API for XML; Programming language; XML framework; Multi-core processor; Parallel computing; Software engineering; XML Signature; Operating system; Efficient XML Interchange","score_opus":0.032674047108623594,"score_gpt":0.25233480553892,"score_spread":0.21966075843029642,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2522795258","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":[0.7357542,0.0027636774,0.22564217,0.010392517,0.00092963694,0.002365085,0.00012276394,0.019256227,0.0027737564],"genre_scores_gemma":[0.6936123,0.00032252955,0.3055317,0.00015204107,0.000028339055,0.00019454217,0.000007805891,0.00003228837,0.00011846352],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9976741,0.00006689193,0.00047505152,0.00080139755,0.00030607733,0.00067644526],"domain_scores_gemma":[0.9979612,0.00019358103,0.00025283877,0.0014134713,0.00013281447,0.000046077337],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0002134334,0.0004649675,0.00048928603,0.0003086445,0.00061558594,0.00042704356,0.0010672719,0.00036181588,0.000012688893],"category_scores_gemma":[0.0007485948,0.0003822617,0.000070861875,0.0006136424,0.00080590474,0.0027882122,0.002037815,0.00046761613,0.000028966153],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000038054346,0.00009734194,0.0042987186,0.00016605084,0.00006457816,0.00007529163,0.0005285955,0.00008763808,0.017847937,0.29027447,0.0029440825,0.68357724],"study_design_scores_gemma":[0.0049969875,0.0035822133,0.017825942,0.0026945015,0.00013005934,0.0012844917,0.12760276,0.097652435,0.3372182,0.10986012,0.28780603,0.009346229],"about_ca_topic_score_codex":0.000208564,"about_ca_topic_score_gemma":0.00009722399,"teacher_disagreement_score":0.674231,"about_ca_system_score_codex":0.000067816036,"about_ca_system_score_gemma":0.000037279984,"threshold_uncertainty_score":0.9998629},"labels":[],"label_agreement":null},{"id":"W2756031164","doi":"10.4242/balisagevol19.lee01","title":"The Secret Life of Schema in Web Protocols, API's and Software Type Systems","year":2017,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Service-Oriented Architecture and Web Services","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nexen (Canada)","funders":"","keywords":"Computer science; JSON; Document Structure Description; Software engineering; Programming language; XML; Software; Markup language; Schema (genetic algorithms); World Wide Web; Information retrieval","score_opus":0.016728122882226866,"score_gpt":0.2596243976694916,"score_spread":0.24289627478726472,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2756031164","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":[0.9043379,0.008376988,0.0025360521,0.043775335,0.0019205597,0.021910343,0.000089823385,0.0055183657,0.011534652],"genre_scores_gemma":[0.99229217,0.0005272744,0.0054595754,0.0001351973,0.000031288684,0.0013627227,0.0000023070559,0.000019316592,0.00017014476],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985769,0.00006724679,0.00031263678,0.0004190018,0.00028338554,0.000340848],"domain_scores_gemma":[0.9973025,0.0001933843,0.0003271825,0.002027203,0.000111116,0.0000386508],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041677,0.00022386192,0.0002980135,0.00014392305,0.0005062006,0.0005537462,0.0028239281,0.00019662919,0.0000014738274],"category_scores_gemma":[0.0005453605,0.00014782672,0.000036710593,0.00030199933,0.0004250646,0.0005585179,0.0014342171,0.00028481463,0.000004819091],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010920033,0.00030021137,0.13122903,0.002416962,0.00023703118,0.00018713287,0.002882994,0.00011243731,0.006346341,0.6371471,0.0028418913,0.21520682],"study_design_scores_gemma":[0.0067572542,0.005881267,0.14317006,0.00658867,0.00007157534,0.0002696954,0.04337528,0.010262834,0.106263004,0.10952017,0.56381744,0.004022772],"about_ca_topic_score_codex":0.0001261382,"about_ca_topic_score_gemma":0.0006421828,"teacher_disagreement_score":0.56097555,"about_ca_system_score_codex":0.000017270815,"about_ca_system_score_gemma":0.000087237895,"threshold_uncertainty_score":0.60281986},"labels":[],"label_agreement":null},{"id":"W2941403856","doi":"10.4242/balisagevol22.morrissey01","title":"“Be in the Room Where It Happens”","year":2018,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Diverse academic research themes","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Nuclear Physics; National Institute of Information and Communications Technology; Japan Aerospace Exploration Agency; Canadian Space Agency; Center for Long-Term Cybersecurity, University of California Berkeley; Geo-Informatics and Space Technology Development Agency; Iran Telecommunication Research Center; European Space Agency; Agenzia Spaziale Italiana; European Organization for the Exploitation of Meteorological Satellites; Korea Aerospace Research Institute; National Commission for Science and Technology; Türkiye Bilimsel ve Teknolojik Araştırma Kurumu; National Oceanic and Atmospheric Administration; National Space Organization; China National Space Administration; Belgian Federal Science Policy Office; Chinese Academy of Sciences; U.S. Geological Survey; Indian Space Research Organisation; Centre National d’Etudes Spatiales; Commonwealth Scientific and Industrial Research Organisation; National Aeronautics and Space Administration","keywords":"USable; Verifiable secret sharing; Computer science; Key (lock); Task (project management); Face (sociological concept); Resource (disambiguation); Digital content; Internet privacy; Data science; World Wide Web; Multimedia; Computer security; Engineering; Sociology; Set (abstract data type)","score_opus":0.16508507791108423,"score_gpt":0.41564468531797755,"score_spread":0.2505596074068933,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2941403856","genre_codex":"other","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":[0.26065472,0.00087373273,0.00014079602,0.31637332,0.00065945485,0.0011763835,0.000102180406,0.0013488842,0.41867054],"genre_scores_gemma":[0.9848983,0.0007705086,0.0008970421,0.0019070883,0.00008402181,0.000076568875,0.0000022914999,0.000019671596,0.011344492],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99527174,0.00044302526,0.0005912749,0.00074864994,0.002242977,0.0007023292],"domain_scores_gemma":[0.99563074,0.0018115207,0.00019371965,0.002080884,0.00024220982,0.00004090386],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.005479905,0.00027267574,0.00035302978,0.0006195414,0.0003489074,0.000633538,0.005786615,0.0003600292,0.0013917056],"category_scores_gemma":[0.018070359,0.00015357393,0.00011311568,0.0021449807,0.0018917661,0.00093607884,0.0012577397,0.0007827659,0.0015907896],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028778682,0.00006445975,0.0062812944,0.0000066122893,0.000016953445,0.00012408795,0.0013776488,0.000003194794,0.00051552075,0.08753384,0.7964358,0.1073528],"study_design_scores_gemma":[0.00021300376,0.00041950814,0.0023752828,0.000040818188,0.0000030680271,0.000019806119,0.0738568,0.000031713367,0.003241164,0.110173255,0.8094289,0.00019669512],"about_ca_topic_score_codex":0.000073856456,"about_ca_topic_score_gemma":0.0005902233,"teacher_disagreement_score":0.7242436,"about_ca_system_score_codex":0.00007020237,"about_ca_system_score_gemma":0.00006683407,"threshold_uncertainty_score":0.99959254},"labels":[],"label_agreement":null},{"id":"W2979338130","doi":"10.4242/balisagevol23.boyer01","title":"Multitasking algorithms in XForms","year":2019,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IBM (Canada)","funders":"","keywords":"Human multitasking; Computer science; Algorithm","score_opus":0.025212435694555068,"score_gpt":0.21830197535124637,"score_spread":0.19308953965669132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979338130","genre_codex":"other","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":[0.46429524,0.00024071024,0.0000022957672,0.00072317343,0.0007854171,0.00039848138,0.00007086269,0.0012833998,0.53220046],"genre_scores_gemma":[0.9626119,0.00007889813,0.0002263382,0.00021903092,0.000071936396,0.000050401886,0.000024282148,0.00003396849,0.03668323],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985891,0.000019561963,0.00031102877,0.00036008056,0.00025249607,0.00046774128],"domain_scores_gemma":[0.99925125,0.00008346913,0.000087911,0.0005028549,0.00005202161,0.000022503822],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002166915,0.00026903194,0.00029707977,0.00032449083,0.0001785604,0.00066921575,0.00053625216,0.00013473314,0.001635446],"category_scores_gemma":[0.00011808661,0.00021610966,0.000094253664,0.00009273345,0.0004184026,0.0014452295,0.0002194509,0.00039760367,0.00046537974],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009948913,0.00010892603,0.002349233,0.00007218279,0.000024576228,0.00005965133,0.0036500983,0.00000971838,0.00004914987,0.92708176,0.0018132251,0.06468199],"study_design_scores_gemma":[0.00075700955,0.000659708,0.0015173861,0.000263269,0.000007566381,0.000010383727,0.113938764,0.00005093024,0.0011054748,0.047928654,0.8330082,0.0007526483],"about_ca_topic_score_codex":0.000094510986,"about_ca_topic_score_gemma":0.0010104338,"teacher_disagreement_score":0.87915313,"about_ca_system_score_codex":0.00008104262,"about_ca_system_score_gemma":0.000015496717,"threshold_uncertainty_score":0.9992772},"labels":[],"label_agreement":null},{"id":"W4205261768","doi":"10.4242/balisagevol13.sperberg-mcqueen01","title":"Document lattices","year":2014,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Markup language; XHTML; RuleML; Computer science; Document type definition; Information retrieval; Consistency (knowledge bases); Document Structure Description; XML; Programming language; Natural language processing; World Wide Web; Artificial intelligence","score_opus":0.025126128682454143,"score_gpt":0.22030216916037387,"score_spread":0.19517604047791973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205261768","genre_codex":"other","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":[0.106225535,0.00022524921,0.0000092881055,0.0033722322,0.0007135902,0.00020476493,0.000038352817,0.0025034347,0.88670754],"genre_scores_gemma":[0.9427865,0.00007667384,0.00017261686,0.0005629962,0.00020165888,0.000055491393,0.000017012919,0.000029965751,0.05609705],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986927,0.000041465024,0.0002622905,0.0003258664,0.00027842494,0.0003992471],"domain_scores_gemma":[0.99910104,0.00012783393,0.000103952385,0.0005618175,0.000071517825,0.00003386483],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002697108,0.0002675839,0.0002512776,0.00016508103,0.00048122337,0.00124157,0.0005884609,0.00009943847,0.0017789454],"category_scores_gemma":[0.0002492401,0.00021117587,0.00010164564,0.000048636208,0.00081577955,0.0011252367,0.00020803658,0.00025333453,0.0003626022],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040327912,0.000034914025,0.000060774742,0.000029965458,0.000021216636,0.0000065885215,0.0010711038,0.0000011242512,0.000009942702,0.9454275,0.009703351,0.043593217],"study_design_scores_gemma":[0.00015335258,0.0004230049,0.00008326041,0.000047470992,0.0000110840965,0.0000030914125,0.013994762,0.000002724289,0.0006228818,0.12673812,0.857639,0.00028121297],"about_ca_topic_score_codex":0.00003364335,"about_ca_topic_score_gemma":0.0004265789,"teacher_disagreement_score":0.8479357,"about_ca_system_score_codex":0.000037426264,"about_ca_system_score_gemma":0.000008973674,"threshold_uncertainty_score":0.99979526},"labels":[],"label_agreement":null},{"id":"W4206983357","doi":"10.4242/balisagevol7.usdin01","title":"Serendipity","year":2011,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Serendipity; Luck; Face (sociological concept); Key (lock); Computer science; Psychology; Sociology; Epistemology; Computer security; Social science","score_opus":0.7067267809977053,"score_gpt":0.5319341951220168,"score_spread":0.17479258587568847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206983357","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":[0.5341503,0.00092740456,0.0016969627,0.0037381074,0.0014145311,0.00056763436,0.00013699634,0.002281627,0.45508644],"genre_scores_gemma":[0.9866053,0.00047863074,0.0054707495,0.00011873758,0.000022064964,0.00003275269,0.0000025240577,0.000016110025,0.007253158],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99306124,0.00015352585,0.0005883446,0.000900963,0.0045384355,0.00075750874],"domain_scores_gemma":[0.99565107,0.0010909706,0.00022800219,0.0019284666,0.0009623819,0.00013911491],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0075378492,0.00024618115,0.00036516882,0.028333465,0.00033171245,0.0010183526,0.0046367366,0.00031171463,0.0022414648],"category_scores_gemma":[0.03841117,0.00016239216,0.00016584131,0.08080282,0.0007701659,0.00096270465,0.0016039136,0.00045039898,0.001340389],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019728648,0.00027219238,0.07292542,0.000007680131,0.000029204177,0.00012807082,0.00022117278,0.0000011646825,0.0006546542,0.07505944,0.06567333,0.7848304],"study_design_scores_gemma":[0.00054944685,0.0014151903,0.25541136,0.000020575246,0.000010050658,0.000044737033,0.010743829,0.00009069971,0.07405129,0.35995802,0.29692134,0.0007834618],"about_ca_topic_score_codex":0.000066585984,"about_ca_topic_score_gemma":0.000039166112,"teacher_disagreement_score":0.78404695,"about_ca_system_score_codex":0.00007874801,"about_ca_system_score_gemma":0.000066318215,"threshold_uncertainty_score":0.99943715},"labels":[],"label_agreement":null},{"id":"W4230652012","doi":"10.4242/balisagevol26.jenstad01","title":"Converting an SGML/XML Hybrid to TEI-XML: The Case of the Internet Shakespeare Editions","year":2021,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"SGML; Markup language; XML; Computer science; The Internet; World Wide Web; RuleML; Encoding (memory); Document type definition; Software; Code (set theory); Programming language; Document Structure Description; Artificial intelligence","score_opus":0.03736616915092719,"score_gpt":0.23516744469880627,"score_spread":0.19780127554787907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4230652012","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":[0.8799887,0.00033281633,0.000008087411,0.0069529614,0.0010178956,0.00036242412,0.00071359,0.0006683747,0.10995519],"genre_scores_gemma":[0.97644603,0.00001685651,0.00004316865,0.0009187006,0.00017499953,0.000058522415,0.000030647014,0.000029163459,0.022281885],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986811,0.0000979933,0.00035442002,0.0003223207,0.00020891069,0.0003352086],"domain_scores_gemma":[0.99848634,0.00015587841,0.00013585256,0.00094417116,0.00024233927,0.00003543235],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00023443047,0.00023716678,0.00023800919,0.00010895174,0.0007232538,0.00079421134,0.00077981176,0.00006887723,0.0012715706],"category_scores_gemma":[0.0006514271,0.00014555104,0.00015415205,0.00012688363,0.0010421037,0.0006927105,0.00058437506,0.0003939555,0.00003354145],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006386506,0.0001349868,0.00017875776,0.00007076068,0.00007065343,0.0007608829,0.013336713,0.0000075392345,0.00014231526,0.9324662,0.03436629,0.018401036],"study_design_scores_gemma":[0.00019779238,0.00043451728,0.00017616371,0.00024710858,0.000045930898,0.0005841304,0.3623316,0.000018202476,0.020017609,0.025308605,0.59023136,0.0004069726],"about_ca_topic_score_codex":0.00017517101,"about_ca_topic_score_gemma":0.0064096777,"teacher_disagreement_score":0.9071576,"about_ca_system_score_codex":0.000043082142,"about_ca_system_score_gemma":0.00004196625,"threshold_uncertainty_score":0.9996414},"labels":[],"label_agreement":null},{"id":"W4255233508","doi":"10.4242/balisagevol1.levy01","title":"Beyond the Semantic Web: the Semantic Space","year":2009,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Parsing; Semantics (computer science); Semantic Web; Semantic Web Rule Language; Space (punctuation); Social Semantic Web; Variety (cybernetics); Natural language; Artificial intelligence; World Wide Web; Natural language processing; Semantic analytics; Programming language","score_opus":0.009949442858470403,"score_gpt":0.2247738060855085,"score_spread":0.2148243632270381,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4255233508","genre_codex":"commentary","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":[0.1556397,0.0083355,0.021965439,0.73624474,0.0027463587,0.0021604397,0.000017972283,0.014954352,0.057935525],"genre_scores_gemma":[0.98983216,0.0008244353,0.0048029507,0.002670743,0.00006828637,0.00005009214,0.0000015599888,0.000017894386,0.0017319018],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99731886,0.00017147732,0.00039044782,0.00071916415,0.00058330176,0.00081675063],"domain_scores_gemma":[0.9961824,0.00054832635,0.00021131497,0.002932358,0.00008811353,0.00003752704],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007009835,0.00046302506,0.0004097651,0.00019687749,0.0008443592,0.00067684706,0.0048179007,0.0002730878,0.000016599493],"category_scores_gemma":[0.0006270427,0.00024078599,0.00018863953,0.0011449233,0.00091371493,0.0006706986,0.00085585855,0.0006281627,0.00014303929],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045195375,0.00010018805,0.0005272108,0.000026756634,0.000058958867,0.0001634191,0.0008037503,0.00004162794,0.0017482444,0.86685216,0.04649754,0.08313492],"study_design_scores_gemma":[0.0011096392,0.0019121143,0.033763226,0.0002378893,0.00012815172,0.0007557026,0.02174649,0.005720067,0.046271797,0.7680689,0.11839746,0.0018885809],"about_ca_topic_score_codex":0.00003292282,"about_ca_topic_score_gemma":0.0001225687,"teacher_disagreement_score":0.83419245,"about_ca_system_score_codex":0.00005201318,"about_ca_system_score_gemma":0.00007510344,"threshold_uncertainty_score":0.98189676},"labels":[],"label_agreement":null},{"id":"W4288685011","doi":"10.4242/balisagevol27.takeda01","title":"Serverless Searching with XSLT and JavaScript","year":2022,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Web Data Mining and Analysis","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; Simon Fraser University","funders":"","keywords":"XSLT; JavaScript; Computer science; World Wide Web; Bytecode; Ajax; Presentation (obstetrics); XML; Information retrieval; Programming language; Web application; Virtual machine","score_opus":0.008208968830791925,"score_gpt":0.19156553632581513,"score_spread":0.18335656749502321,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288685011","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":[0.9181979,0.0017347338,0.033629898,0.029673465,0.00035621488,0.0004657068,0.00018674166,0.007790302,0.007965051],"genre_scores_gemma":[0.9734228,0.00012268053,0.02493216,0.00023317005,0.000009274937,0.000075506665,0.000014439052,0.000015128671,0.0011747987],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984722,0.000102469414,0.00014578352,0.00054523634,0.00040370694,0.00033060455],"domain_scores_gemma":[0.99881005,0.00007338363,0.00007773416,0.0009780222,0.000027622671,0.00003315789],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041799748,0.00018009654,0.00021266176,0.00027786914,0.0006705372,0.00031676836,0.0014416211,0.000046505167,0.000020853036],"category_scores_gemma":[0.00008570813,0.00014741943,0.000037916056,0.0008341328,0.0002160187,0.0006390156,0.0020919435,0.0003902724,0.000005015529],"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":[0.0002205673,0.00024293472,0.01529735,0.00013092924,0.00030697027,0.0008870717,0.0022326312,0.0006336105,0.0017970147,0.35823196,0.01458821,0.6054307],"study_design_scores_gemma":[0.003861026,0.008607206,0.023492122,0.00041612858,0.00021709659,0.002016628,0.13623641,0.02923293,0.031203756,0.05875278,0.70091486,0.005049034],"about_ca_topic_score_codex":0.000068874244,"about_ca_topic_score_gemma":0.000033741584,"teacher_disagreement_score":0.6863267,"about_ca_system_score_codex":0.00004520649,"about_ca_system_score_gemma":0.00003659524,"threshold_uncertainty_score":0.60115904},"labels":[],"label_agreement":null},{"id":"W4288685812","doi":"10.4242/balisagevol27.holstege01","title":"Metadata for Creators","year":2022,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Nuclear Physics; National Institute of Information and Communications Technology; Japan Aerospace Exploration Agency; Canadian Space Agency; Center for Long-Term Cybersecurity, University of California Berkeley; Geo-Informatics and Space Technology Development Agency; Iran Telecommunication Research Center; European Space Agency; Agenzia Spaziale Italiana; European Organization for the Exploitation of Meteorological Satellites; Korea Aerospace Research Institute; National Commission for Science and Technology; Türkiye Bilimsel ve Teknolojik Araştırma Kurumu; National Oceanic and Atmospheric Administration; National Space Organization; China National Space Administration; Belgian Federal Science Policy Office; Chinese Academy of Sciences; U.S. Geological Survey; Indian Space Research Organisation; Centre National d’Etudes Spatiales; Commonwealth Scientific and Industrial Research Organisation; National Aeronautics and Space Administration","keywords":"Metadata; Computer science; Meta Data Services; World Wide Web; Geospatial metadata; Process (computing); Key (lock); Metadata repository; Data element; Information retrieval; Computer security","score_opus":0.018163097642529862,"score_gpt":0.21971845434499412,"score_spread":0.20155535670246427,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288685812","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":[0.060433943,0.0016470267,0.6774051,0.10650801,0.00861116,0.0055131647,0.001259805,0.02776041,0.11086137],"genre_scores_gemma":[0.92163754,0.00009543727,0.0571841,0.002965388,0.000051503826,0.0014976967,0.00008024494,0.000024665513,0.016463457],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99866366,0.00004433347,0.0002696464,0.00033766244,0.00038089996,0.0003037968],"domain_scores_gemma":[0.9986658,0.00009713878,0.0001393545,0.0010396601,0.000032092972,0.000025941628],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050597766,0.00014156177,0.0001737389,0.0002692334,0.00063845766,0.0005314644,0.0030153801,0.000050562252,0.00004508239],"category_scores_gemma":[0.00015743503,0.00011902025,0.00007852931,0.00085868937,0.000118356154,0.006947652,0.0013970501,0.00014055207,0.000021000245],"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":[0.000033999102,0.00003729816,0.00030268254,0.000023494573,0.000020720774,0.000007661748,0.00061782275,0.000115211864,0.00022428793,0.8509189,0.09758151,0.050116412],"study_design_scores_gemma":[0.00019590034,0.0005050804,0.00011781594,0.000005001042,0.0000014357573,0.000031038053,0.009083283,0.0012943157,0.009035253,0.01315918,0.9663343,0.00023742284],"about_ca_topic_score_codex":0.0000054246907,"about_ca_topic_score_gemma":7.4318035e-7,"teacher_disagreement_score":0.8687528,"about_ca_system_score_codex":0.00004650763,"about_ca_system_score_gemma":0.000061695835,"threshold_uncertainty_score":0.56033766},"labels":[],"label_agreement":null},{"id":"W4288686143","doi":"10.4242/balisagevol27.david01","title":"BITS for Government Information?","year":2022,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Mathematics, Computing, and Information Processing","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Council of University Libraries; University of Toronto","funders":"","keywords":"Metadata; Suite; Computer science; World Wide Web; Schema (genetic algorithms); Government (linguistics); Rewriting; Vocabulary; Information retrieval; Political science","score_opus":0.008548217516537648,"score_gpt":0.20715457783906724,"score_spread":0.1986063603225296,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288686143","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":[0.0036339955,0.000105285355,0.945137,0.007988967,0.00087856036,0.00084103114,0.000092251605,0.003843314,0.03747959],"genre_scores_gemma":[0.6691029,0.00006621347,0.32387012,0.0032534283,0.000042222484,0.001042503,0.000039361727,0.000024892968,0.0025583194],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985868,0.00001958795,0.00040210396,0.00017391844,0.000541147,0.00027645272],"domain_scores_gemma":[0.9989082,0.00010454913,0.00031873214,0.0005788598,0.00006885857,0.000020754178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004797042,0.0001568522,0.00016207532,0.00014133957,0.0007153971,0.00034147382,0.0013477121,0.000053015592,0.00003828531],"category_scores_gemma":[0.00039923762,0.00014719536,0.000064087326,0.0003992894,0.00006453165,0.0020701424,0.001085963,0.0001702997,0.000019211113],"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":[0.000022402533,0.000047905844,0.000033274293,0.00015677478,0.000016154676,0.0000012926964,0.0015824606,0.0005967779,0.000027889711,0.7395393,0.026678821,0.23129693],"study_design_scores_gemma":[0.0005408762,0.00051907654,0.000074818214,0.000033791708,0.000005712232,0.00003932969,0.010319455,0.045879506,0.007470113,0.07072894,0.8639276,0.00046076011],"about_ca_topic_score_codex":0.0000011118569,"about_ca_topic_score_gemma":4.59189e-7,"teacher_disagreement_score":0.8372488,"about_ca_system_score_codex":0.00015023357,"about_ca_system_score_gemma":0.000046406698,"threshold_uncertainty_score":0.60024524},"labels":[],"label_agreement":null},{"id":"W4385386714","doi":"10.4242/balisagevol28.hymers01","title":"Retractions and Corrections at Scholars Portal Journals","year":2023,"lang":"en","type":"article","venue":"Balisage series on markup technologies","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Council of University Libraries","funders":"","keywords":"Metadata; Computer science; Workflow; Identifier; World Wide Web; XML; Process (computing); Digital library; Service (business); Object (grammar); Information retrieval; Library science; Database; Business","score_opus":0.025519186116147733,"score_gpt":0.2562568147893987,"score_spread":0.23073762867325098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385386714","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":[0.8079193,0.0005096484,0.010188575,0.066731155,0.0043768934,0.0007325766,0.00008106969,0.018006455,0.09145432],"genre_scores_gemma":[0.9866352,0.0011729646,0.0023281148,0.00060182565,0.000023059834,0.000047441437,0.000009505964,0.000009100767,0.00917275],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988396,0.00004131466,0.00027427007,0.0002611472,0.00031850603,0.00026512388],"domain_scores_gemma":[0.99914145,0.00009282404,0.00014036194,0.0005242752,0.000053567644,0.00004753911],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005468072,0.00012670994,0.00013434788,0.00051971525,0.00061174494,0.00078654365,0.0006994558,0.00012755228,0.000039802962],"category_scores_gemma":[0.00033878558,0.00010526507,0.000046237285,0.0016493491,0.00018088073,0.008221623,0.0005915297,0.00026750873,0.00028909795],"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":[0.00004168981,0.00007874131,0.042758748,0.00006427655,0.00008447246,0.00021970898,0.0045829066,0.00029761638,0.0070871664,0.37272176,0.3711047,0.2009582],"study_design_scores_gemma":[0.0005641551,0.0006834316,0.19281884,0.0001886409,0.0000064940723,0.0015741396,0.033501595,0.004530536,0.026108041,0.023962691,0.7149915,0.0010699198],"about_ca_topic_score_codex":0.00000504048,"about_ca_topic_score_gemma":0.000009423959,"teacher_disagreement_score":0.34875906,"about_ca_system_score_codex":0.000028566368,"about_ca_system_score_gemma":0.000039685707,"threshold_uncertainty_score":0.7584658},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"}]}