{"meta":{"query_hash":"6ab4977ec0b1","filters":{"venue":"Synthesis lectures on data management"},"cohort_total":52,"direct_labels_cover":0,"predictions_cover":52,"exported":52,"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/6ab4977ec0b1","api":"https://metacan.xera.ac/api/v1/cohort?venue=Synthesis+lectures+on+data+management"},"results":[{"id":"W2000922801","doi":"10.1007/978-3-031-01850-3","title":"Information and Influence Propagation in Social Networks","year":2014,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":252,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Social media; Social network (sociolinguistics); Internet privacy; Sociology; Economic geography; Computer science; World Wide Web; Geography","score_opus":0.01122099561622583,"score_gpt":0.24802535413489965,"score_spread":0.23680435851867382,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2000922801","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013405475,0.042022727,0.592636,0.009660764,0.0038884825,0.00008335997,0.00088948815,0.0007695701,0.3366442],"genre_scores_gemma":[0.35961258,0.06427753,0.13163194,0.0016638966,0.012766937,0.00045381158,0.0013476533,0.0006633847,0.4275823],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99958676,0.00009389664,0.000015409449,0.00008195163,0.00019700294,0.000025113659],"domain_scores_gemma":[0.9981111,0.0014843796,0.00006915717,0.00011652618,0.0001587344,0.000060127437],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00067990477,0.00096072035,0.0006758683,0.0016942144,0.00061052747,0.0020855782,0.00061636674,0.00089511095,0.009895643],"category_scores_gemma":[0.003912513,0.00045811286,0.0005081468,0.0026127794,0.0014829476,0.0040422603,0.0008152945,0.001665541,0.002079458],"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":[0.000015567935,0.000033107324,0.00035448745,0.00022358457,0.000044176606,0.00007687766,0.00026204318,0.026686395,0.00047169704,0.77346414,0.062202472,0.13616547],"study_design_scores_gemma":[0.0000060566763,0.000008770323,0.00044635413,0.000051516417,0.000017600309,0.00006327636,0.00004452961,0.042442847,0.00022676111,0.8961101,0.060567286,0.0000149279595],"about_ca_topic_score_codex":0.002021702,"about_ca_topic_score_gemma":0.0023777876,"teacher_disagreement_score":0.009895643,"about_ca_system_score_codex":0.0010628242,"about_ca_system_score_gemma":0.0004741148,"threshold_uncertainty_score":0.03310424},"labels":[],"label_agreement":null},{"id":"W2004506767","doi":"10.2200/s00396ed1v01y201111dtm022","title":"Full-Text (Substring) Indexes in External Memory","year":2011,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Substring; Computer science; Natural language processing; Arithmetic; Mathematics; Programming language; Data structure","score_opus":0.054580631353493975,"score_gpt":0.2540086173953766,"score_spread":0.19942798604188264,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004506767","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044141956,0.007681959,0.8732348,0.0012727603,0.0022670904,0.00017416399,0.0019385581,0.020339468,0.048949193],"genre_scores_gemma":[0.30306774,0.0048066927,0.56861186,0.001127719,0.0019215288,0.0004020889,0.0055055297,0.004872982,0.10968387],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9989114,0.00016366565,0.00016124861,0.00020025119,0.0004143188,0.0001492037],"domain_scores_gemma":[0.9966133,0.000969128,0.00015380346,0.0015358265,0.0005909343,0.00013687588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009798981,0.0011892908,0.0011501685,0.0018133793,0.0010574693,0.0039757844,0.0021154578,0.00088920904,0.026787817],"category_scores_gemma":[0.0056106406,0.0006165078,0.0005626616,0.0044616186,0.0010238595,0.0076170205,0.0027678472,0.00110703,0.011824231],"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":[0.0009275756,0.0001655891,0.000978164,0.0005358892,0.00007913186,0.000418408,0.00034593252,0.0114324605,0.030603457,0.15961409,0.076472975,0.71842635],"study_design_scores_gemma":[0.00021777653,0.00044165342,0.0011883976,0.0004050471,0.00019535272,0.0010124531,0.00033499426,0.14403072,0.18752272,0.42372197,0.2407826,0.00014631326],"about_ca_topic_score_codex":0.0007524105,"about_ca_topic_score_gemma":0.001187725,"teacher_disagreement_score":0.026787817,"about_ca_system_score_codex":0.0009257085,"about_ca_system_score_gemma":0.0009769027,"threshold_uncertainty_score":0.08961415},"labels":[],"label_agreement":null},{"id":"W2019933064","doi":"10.2200/s00363ed1v01y201105dtm018","title":"Fundamentals of Physical Design and Query Compilation","year":2011,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute of Informatics","keywords":"Computer science; Query optimization; Information retrieval","score_opus":0.09803155566453071,"score_gpt":0.27762418553492446,"score_spread":0.17959262987039376,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2019933064","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001517181,0.0071446104,0.96026534,0.0012173298,0.0007855681,0.00006543807,0.0001561899,0.0016818217,0.027166663],"genre_scores_gemma":[0.15326074,0.018807538,0.7662249,0.00087311346,0.0020389147,0.00071342534,0.0007022375,0.00139086,0.05598826],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983286,0.00024213073,0.00011884876,0.00029690366,0.0008739234,0.00013950236],"domain_scores_gemma":[0.9986564,0.0005766683,0.00006952296,0.0003407844,0.0003090922,0.000047555404],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014221119,0.0014966854,0.000982614,0.0010875543,0.0007895465,0.004520024,0.0021538988,0.0011579577,0.016083961],"category_scores_gemma":[0.0045547644,0.001253599,0.001059538,0.0017894785,0.002445603,0.0043200874,0.0013432526,0.0025772576,0.0059283986],"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":[0.00007274147,0.00006783795,0.0002506881,0.00063141866,0.000043435746,0.000061200975,0.0002261512,0.017193269,0.0048631844,0.6680764,0.030258711,0.27825493],"study_design_scores_gemma":[0.000053202675,0.000062492465,0.00042743707,0.00024119983,0.000056898578,0.00023916822,0.00008014586,0.067639105,0.007044692,0.6951364,0.22896962,0.000049719198],"about_ca_topic_score_codex":0.0017753113,"about_ca_topic_score_gemma":0.0014514888,"teacher_disagreement_score":0.016083961,"about_ca_system_score_codex":0.0018801553,"about_ca_system_score_gemma":0.0018834146,"threshold_uncertainty_score":0.053806186},"labels":[],"label_agreement":null},{"id":"W2063388849","doi":"10.1007/978-3-031-01837-4","title":"Data Stream Management","year":2010,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":59,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Stream processing; Data warehouse; Data stream mining; Data stream; Database; Data management; Transaction log; Process (computing); Paging; Information retrieval; Database transaction; Data mining; Distributed computing; Computer network; Operating system","score_opus":0.04680600989006113,"score_gpt":0.2827983911401374,"score_spread":0.23599238125007627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2063388849","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030056762,0.015972566,0.6615009,0.0029180006,0.0060194475,0.00041326176,0.0025269787,0.01972231,0.28792083],"genre_scores_gemma":[0.021607606,0.014894012,0.15234637,0.0013955983,0.003255905,0.0002761109,0.0065506115,0.0027468058,0.79692703],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995552,0.000026009482,0.000023803654,0.000089686044,0.0002787305,0.000026576456],"domain_scores_gemma":[0.9994686,0.000092400194,0.000020101199,0.00013301085,0.00023022405,0.00005564349],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006113336,0.0012125203,0.0006938309,0.0014899877,0.000522672,0.002395288,0.0010785959,0.0006376836,0.056108274],"category_scores_gemma":[0.0013724784,0.0005126059,0.0005039844,0.0023094467,0.00037030326,0.0026473415,0.0012242869,0.0016204624,0.039110214],"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":[0.00007331933,0.000059961174,0.00014174744,0.00019314529,0.000017867373,0.000057450405,0.0000634342,0.0014593144,0.005504433,0.027660113,0.34783038,0.61693877],"study_design_scores_gemma":[0.000020161306,0.00003635513,0.00033525997,0.000093837465,0.000022042384,0.00024304652,0.00003169681,0.014899889,0.006560386,0.030505084,0.94723296,0.000019271345],"about_ca_topic_score_codex":0.0008741265,"about_ca_topic_score_gemma":0.0009704965,"teacher_disagreement_score":0.056108274,"about_ca_system_score_codex":0.0006130569,"about_ca_system_score_gemma":0.0007413116,"threshold_uncertainty_score":0.18770087},"labels":[],"label_agreement":null},{"id":"W2953072692","doi":"10.1007/978-3-031-01873-2_1","title":"On Transactional Concurrency Control","year":2019,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Lock (firearm); Optimistic concurrency control; Multiversion concurrency control; Concurrency control; Non-lock concurrency control; Serializability; Transactional memory; Isolation (microbiology); Timestamp-based concurrency control; Software transactional memory; Concurrency; Distributed concurrency control; Atomicity; Programming language; Distributed computing; Database transaction; Parallel computing; Operating system; Transaction processing; Distributed transaction; Engineering","score_opus":0.025036753098219146,"score_gpt":0.24090283150141617,"score_spread":0.21586607840319702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953072692","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026992226,0.07148541,0.35787183,0.0056309155,0.008963264,0.00010501982,0.0003567324,0.0012677192,0.5516199],"genre_scores_gemma":[0.12450145,0.049935978,0.08376299,0.002809014,0.00670695,0.00034193945,0.0009576635,0.0010457755,0.7299382],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9991059,0.00013923227,0.00004439589,0.00014395411,0.00049289974,0.000073596406],"domain_scores_gemma":[0.99940073,0.0002515365,0.000022877539,0.0001657544,0.0001253051,0.000033766923],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000878337,0.0011891956,0.00084632495,0.0013380641,0.0008822915,0.0029197743,0.0011883214,0.000925064,0.024736725],"category_scores_gemma":[0.0017803056,0.0006705422,0.0005132914,0.0024981475,0.0026248342,0.0038895574,0.0014413896,0.0033209615,0.0076619107],"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":[0.000026913473,0.000027086247,0.000051203693,0.00017431127,0.000014169276,0.000033753484,0.00022438324,0.0026678932,0.0011193535,0.7699822,0.077271275,0.14840747],"study_design_scores_gemma":[0.000012784355,0.000026126187,0.00014275973,0.00015994538,0.0000145445465,0.00009841582,0.000045675864,0.0074704695,0.0012078471,0.5900534,0.40075174,0.000016376205],"about_ca_topic_score_codex":0.001352517,"about_ca_topic_score_gemma":0.0013861432,"teacher_disagreement_score":0.024736725,"about_ca_system_score_codex":0.0015243625,"about_ca_system_score_gemma":0.0008436229,"threshold_uncertainty_score":0.082752585},"labels":[],"label_agreement":null},{"id":"W2966153325","doi":"10.1007/978-3-031-01874-9","title":"Community Search over Big Graphs","year":2019,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":39,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Data science; Big data; World Wide Web; Theoretical computer science; Data mining","score_opus":0.05929447088745657,"score_gpt":0.3019880674766987,"score_spread":0.24269359658924214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2966153325","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024274537,0.020510541,0.7794326,0.010736644,0.0026417824,0.00020507109,0.0040006894,0.0035311058,0.15466711],"genre_scores_gemma":[0.3022747,0.015585934,0.38187075,0.0023056858,0.0056250216,0.00042469223,0.010747294,0.00238056,0.27878538],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9994456,0.00010214541,0.000019792236,0.00016096355,0.00022457926,0.000046901892],"domain_scores_gemma":[0.9978078,0.0013235132,0.00009671941,0.00036520883,0.00024783766,0.00015893513],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008177236,0.00080514397,0.0009880193,0.0021777228,0.0008811346,0.002023683,0.0011099159,0.000955021,0.01540293],"category_scores_gemma":[0.0050553214,0.0004588733,0.0006723044,0.00421582,0.0008560103,0.0053112474,0.0019156985,0.0019906445,0.004302898],"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":[0.00012512121,0.0000881092,0.00047791412,0.00060483295,0.00010960975,0.00008433473,0.00020622413,0.035499368,0.002115466,0.2905491,0.28082085,0.3893191],"study_design_scores_gemma":[0.000036610705,0.00003122369,0.00056261744,0.00008516751,0.000030639283,0.00012839751,0.00009240649,0.15330967,0.0010586777,0.75879997,0.08583966,0.000025098276],"about_ca_topic_score_codex":0.0020324287,"about_ca_topic_score_gemma":0.0034453785,"teacher_disagreement_score":0.01540293,"about_ca_system_score_codex":0.00095373153,"about_ca_system_score_gemma":0.00062875974,"threshold_uncertainty_score":0.051527977},"labels":[],"label_agreement":null},{"id":"W3008612852","doi":"10.2200/s00986ed1v01y202001dtm062","title":"Cloud-Based RDF Data Management","year":2020,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"RDF; Computer science; Linked data; SPARQL; RDF Schema; Cloud computing; World Wide Web; Web resource; Schema (genetic algorithms); Data science; Semantic Web; Knowledge management; Information retrieval","score_opus":0.12178589594641079,"score_gpt":0.29252881383523566,"score_spread":0.17074291788882487,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008612852","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009598897,0.0018591362,0.92205286,0.0060287192,0.001867481,0.0005298065,0.0038042078,0.013678219,0.040580694],"genre_scores_gemma":[0.2606931,0.005248264,0.6606011,0.002619891,0.0013727881,0.00052760274,0.026768204,0.005221243,0.036947716],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99639064,0.0004833825,0.0003529497,0.0006099849,0.0017196613,0.00044335506],"domain_scores_gemma":[0.99574935,0.0005975355,0.00016001567,0.001994623,0.0011240522,0.000374451],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0053257677,0.00080266315,0.0012668673,0.002140528,0.002119976,0.007750207,0.0037628927,0.0013403172,0.010463337],"category_scores_gemma":[0.0057293284,0.0007704227,0.0014594175,0.005414398,0.0014051145,0.011607916,0.004340082,0.0029297962,0.0050030276],"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":[0.0004903302,0.0005021388,0.0021636542,0.00038164295,0.00016420364,0.00051543594,0.00041675763,0.028199695,0.0141257,0.43763703,0.18787566,0.32752773],"study_design_scores_gemma":[0.00012217589,0.000050511353,0.0013473156,0.00026986765,0.00010047201,0.00039659595,0.00047142344,0.36618114,0.02666477,0.24216726,0.3621117,0.00011672366],"about_ca_topic_score_codex":0.02268906,"about_ca_topic_score_gemma":0.01575776,"teacher_disagreement_score":0.02268906,"about_ca_system_score_codex":0.0035074742,"about_ca_system_score_gemma":0.004600543,"threshold_uncertainty_score":0.04511398},"labels":[],"label_agreement":null},{"id":"W3107439925","doi":"10.2200/s01048ed1v01y202009dtm063","title":"Skylines and Other Dominance-Based Queries","year":2020,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Dominance (genetics); Computer science; Information retrieval; Theoretical computer science; Chemistry","score_opus":0.05543170520960824,"score_gpt":0.2604830773332338,"score_spread":0.20505137212362554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3107439925","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04190424,0.016582396,0.78070104,0.017538253,0.003194265,0.00019270893,0.0022965034,0.0022385411,0.13535199],"genre_scores_gemma":[0.56451046,0.013278852,0.29341248,0.006279993,0.0076991646,0.00046863666,0.0039225896,0.0015706451,0.10885724],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99638426,0.0009330902,0.00017389531,0.00048233592,0.0017186371,0.00030783948],"domain_scores_gemma":[0.9936273,0.0038966779,0.0003463528,0.0009474622,0.0008042061,0.00037810652],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034318748,0.0008691516,0.0016725479,0.0018239417,0.001828411,0.0051535885,0.0015440844,0.0017450284,0.015391811],"category_scores_gemma":[0.015014394,0.00054405036,0.001025828,0.00499672,0.0024007645,0.014219666,0.0023863635,0.0036094876,0.0021933164],"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":[0.00020184519,0.000057698675,0.0003961718,0.00020648507,0.000029316298,0.00006128201,0.00020892243,0.0053900485,0.0014173354,0.87283087,0.053282246,0.06591776],"study_design_scores_gemma":[0.000052364005,0.000048482485,0.00026163433,0.000041960575,0.000020550871,0.00021735409,0.00012804534,0.032751314,0.001082687,0.90570664,0.059662767,0.000026195854],"about_ca_topic_score_codex":0.0022140015,"about_ca_topic_score_gemma":0.0023646804,"teacher_disagreement_score":0.015391811,"about_ca_system_score_codex":0.0023659507,"about_ca_system_score_gemma":0.0011073533,"threshold_uncertainty_score":0.051490664},"labels":[],"label_agreement":null},{"id":"W4210956166","doi":"10.2200/s01068ed1v01y202012dtm065","title":"Fault-Tolerant Distributed Transactions on Blockchain","year":2021,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Blockchain Technology Applications and Security","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Blockchain; Computer security; Business; Identity management; Cryptocurrency; Identity (music); Computer science; Authentication (law)","score_opus":0.021027213882897775,"score_gpt":0.25114822558313266,"score_spread":0.23012101170023488,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210956166","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20835328,0.0017169213,0.7530158,0.0018216673,0.000852468,0.00015557188,0.00023392445,0.0012339461,0.03261643],"genre_scores_gemma":[0.95748323,0.0008659437,0.027686832,0.00008311984,0.00018588307,0.00008716303,0.00017620617,0.00009515388,0.013336484],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990601,0.00023316863,0.000061282626,0.00012251074,0.00041193012,0.00011090036],"domain_scores_gemma":[0.9980476,0.0009788825,0.00009206868,0.00044580965,0.000314532,0.00012109472],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00092600443,0.00044769788,0.00071246945,0.000449562,0.0007781142,0.0014641136,0.00084296754,0.0008200985,0.0043253903],"category_scores_gemma":[0.003984906,0.00037034476,0.0002938787,0.0009781464,0.0013077903,0.002590972,0.0012176683,0.0012682104,0.00068187],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006607346,0.000111244684,0.00072814984,0.00020827718,0.000042865628,0.00045854403,0.00035788957,0.3922285,0.022302372,0.47891772,0.0055391924,0.09844453],"study_design_scores_gemma":[0.00008674315,0.000111851004,0.00019382453,0.000023692603,0.00001279924,0.000064042484,0.000040421994,0.58371335,0.0075389137,0.4018772,0.0063187806,0.000018367387],"about_ca_topic_score_codex":0.0013622118,"about_ca_topic_score_gemma":0.0010126025,"teacher_disagreement_score":0.0043253903,"about_ca_system_score_codex":0.0010470005,"about_ca_system_score_gemma":0.0007230627,"threshold_uncertainty_score":0.014469862},"labels":[],"label_agreement":null},{"id":"W4210979382","doi":"10.2200/s00878ed1v01y201810dtm052","title":"Data Profiling","year":2018,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Profiling (computer programming); Metadata; Computer science; Data science; Research data; Information retrieval; World Wide Web; Data curation; Operating system","score_opus":0.484101918142008,"score_gpt":0.4815667505532124,"score_spread":0.0025351675887955727,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210979382","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029024365,0.002289245,0.8911287,0.0073628025,0.0028030123,0.0007016171,0.009618568,0.022889469,0.060304135],"genre_scores_gemma":[0.103436224,0.005104286,0.7193628,0.0056439573,0.0027004262,0.001014613,0.03162248,0.010361431,0.12075387],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9899368,0.0021179598,0.0011463797,0.0023055272,0.003883036,0.00061031774],"domain_scores_gemma":[0.9756566,0.00520239,0.00091430167,0.01209819,0.005087613,0.0010408014],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009461921,0.0017408535,0.0015151376,0.0053087114,0.00206126,0.010401932,0.0024981603,0.0012296219,0.04375417],"category_scores_gemma":[0.03965746,0.0013666918,0.0016061134,0.0060984953,0.0012000346,0.015124093,0.0073317895,0.0035954828,0.042071953],"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":[0.00030634954,0.00009195699,0.0020900732,0.00053839467,0.000055521796,0.000117542004,0.00036781566,0.002311708,0.0052925902,0.15511604,0.20579448,0.6279175],"study_design_scores_gemma":[0.000023572515,0.00004593556,0.0009990203,0.00034167108,0.00004371092,0.0004281549,0.00029459127,0.01445712,0.016094366,0.13452704,0.83267146,0.00007339473],"about_ca_topic_score_codex":0.0013909492,"about_ca_topic_score_gemma":0.0011879932,"teacher_disagreement_score":0.04375417,"about_ca_system_score_codex":0.0018004255,"about_ca_system_score_gemma":0.0035821814,"threshold_uncertainty_score":0.14637226},"labels":[],"label_agreement":null},{"id":"W4211029301","doi":"10.2200/s00439ed1v01y201207dtm030","title":"Foundations of Data Quality Management","year":2012,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":131,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Key Research and Development Program of China; Engineering and Physical Sciences Research Council; Universiteit Antwerpen; University of Waterloo; National Natural Science Foundation of China","keywords":"Data quality; Data management; Computer science; Quality (philosophy); Data governance; Data science; Data administration; Database; Database design; Operations management; Engineering; Database model","score_opus":0.6066300196494542,"score_gpt":0.5319929733621165,"score_spread":0.0746370462873377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211029301","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003627056,0.06882723,0.587062,0.16989338,0.005736002,0.00018102673,0.00058875437,0.00055973325,0.16352478],"genre_scores_gemma":[0.49033952,0.09489041,0.314238,0.02216477,0.02477989,0.0011261245,0.0011173257,0.0005817027,0.050762285],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9851488,0.0055410783,0.0011520652,0.0017333094,0.005690407,0.0007343169],"domain_scores_gemma":[0.9497948,0.030658731,0.002525893,0.008178499,0.0078449175,0.0009970932],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02355294,0.00094347046,0.0013835897,0.005842323,0.002762041,0.012444524,0.002807948,0.0043925783,0.013396668],"category_scores_gemma":[0.05046068,0.00087344885,0.0014298169,0.0065630837,0.01929618,0.015769605,0.004724659,0.0072127613,0.002494189],"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":[0.0000032437338,0.000008911575,0.00007655019,0.00007013923,0.0000065196114,0.0000064320147,0.00007708491,0.00059753563,0.000020378537,0.97910607,0.0054529454,0.01457427],"study_design_scores_gemma":[0.000005063985,0.000004365136,0.0000883085,0.00014279499,0.000004903649,0.000010462334,0.000047460195,0.0014872388,0.000070014576,0.96853536,0.029597063,0.0000068279905],"about_ca_topic_score_codex":0.007413502,"about_ca_topic_score_gemma":0.0026400567,"teacher_disagreement_score":0.02355294,"about_ca_system_score_codex":0.009723727,"about_ca_system_score_gemma":0.009183184,"threshold_uncertainty_score":0.12456131},"labels":[],"label_agreement":null},{"id":"W4211053346","doi":"10.2200/s00338ed1v01y201104dtm015","title":"Peer-to-Peer Data Management","year":2011,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Peer-to-Peer Network Technologies","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Information retrieval; Peer-to-peer; Data management; Schema (genetic algorithms); Data retrieval; Overlay; Data science; World Wide Web; Data mining","score_opus":0.11587906955672425,"score_gpt":0.2975658530360445,"score_spread":0.18168678347932024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211053346","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036922595,0.07985804,0.7367525,0.017407961,0.021128712,0.0006775863,0.0010308424,0.005281375,0.13417077],"genre_scores_gemma":[0.14816616,0.14448507,0.28043213,0.0030252119,0.033479948,0.0010441213,0.0033991418,0.0017292321,0.38423887],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9985707,0.00014329831,0.00007486447,0.00031388036,0.000770796,0.00012651291],"domain_scores_gemma":[0.9981323,0.00054657354,0.000059119717,0.00036220258,0.00072536094,0.0001745373],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00197863,0.0010759373,0.0013135299,0.0010678351,0.000991864,0.0037223944,0.0025369339,0.001609152,0.036080457],"category_scores_gemma":[0.004470595,0.0004727519,0.00044380483,0.0022360107,0.0011289374,0.007301945,0.0017781816,0.0018553848,0.01870651],"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":[0.00007371855,0.000082688035,0.0001245536,0.0011632675,0.000029397806,0.000070349706,0.00015125357,0.0057547325,0.004808784,0.15320782,0.18403013,0.6505032],"study_design_scores_gemma":[0.000026482146,0.00009759018,0.00024905152,0.00030046163,0.000028897879,0.00017459026,0.00012603564,0.01826391,0.0056030266,0.14085668,0.8342334,0.000039904222],"about_ca_topic_score_codex":0.001156327,"about_ca_topic_score_gemma":0.0010137642,"teacher_disagreement_score":0.036080457,"about_ca_system_score_codex":0.0018521951,"about_ca_system_score_gemma":0.0016064331,"threshold_uncertainty_score":0.120701134},"labels":[],"label_agreement":null},{"id":"W4211169459","doi":"10.2200/s00928ed1v01y201906dtm061","title":"Community Search over Big Graphs","year":2019,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":22,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Computer science; Simple (philosophy); Big data; Data science; Theoretical computer science; Data mining; Epistemology","score_opus":0.04453751509950703,"score_gpt":0.3008058352079312,"score_spread":0.25626832010842415,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211169459","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06535112,0.00645669,0.8911329,0.008644567,0.00094830187,0.00021659161,0.00277734,0.0020552247,0.022417372],"genre_scores_gemma":[0.60408604,0.005189649,0.33581582,0.0018548849,0.0031756377,0.0004391272,0.0069317212,0.0012128114,0.04129436],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983215,0.0005389431,0.00006337298,0.0005030247,0.0004345228,0.00013864205],"domain_scores_gemma":[0.9880262,0.0082703605,0.000550386,0.001604025,0.0008435425,0.0007054639],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023564524,0.00090718357,0.0016971938,0.0034450348,0.0013747888,0.0029298225,0.0017398786,0.0018069372,0.008038589],"category_scores_gemma":[0.019730661,0.0007849588,0.0012188798,0.0050774757,0.001421359,0.008668558,0.0031490957,0.0031769623,0.0014186112],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004920357,0.00022246396,0.002484794,0.0009638497,0.00029324903,0.00020135507,0.0004885645,0.15594298,0.004145476,0.48891512,0.10746172,0.23838834],"study_design_scores_gemma":[0.000046135567,0.00003072943,0.0004915157,0.000040160685,0.0000307967,0.000064170046,0.00008787089,0.4466218,0.0008106881,0.54290384,0.008854225,0.000017963266],"about_ca_topic_score_codex":0.003662,"about_ca_topic_score_gemma":0.005049164,"teacher_disagreement_score":0.008038589,"about_ca_system_score_codex":0.0015512331,"about_ca_system_score_gemma":0.00087599515,"threshold_uncertainty_score":0.026891708},"labels":[],"label_agreement":null},{"id":"W4211256942","doi":"10.2200/s00866ed1v01y201807dtm049","title":"Natural Language Data Management and Interfaces","year":2018,"lang":"en","type":"article","venue":"Synthesis lectures on data management","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Natural (archaeology); Computer science; Geology","score_opus":0.028126999865088882,"score_gpt":0.30841588384572977,"score_spread":0.2802888839806409,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211256942","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023828656,0.010639064,0.9052719,0.010806255,0.004439154,0.000109282664,0.0012642234,0.006083323,0.059003938],"genre_scores_gemma":[0.074900486,0.019029535,0.5486487,0.0029079353,0.010841737,0.00054232107,0.0075524887,0.0067137554,0.328863],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981651,0.00032906805,0.00015358804,0.0005968392,0.0006409629,0.0001144119],"domain_scores_gemma":[0.99680257,0.0015083613,0.00007959246,0.0006382401,0.00077643746,0.0001948007],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003480668,0.0012214605,0.0012208226,0.0020009323,0.0010055241,0.007911374,0.0017367641,0.0013306024,0.062164698],"category_scores_gemma":[0.0071204784,0.0009107218,0.00129067,0.0032392272,0.0019279536,0.013728419,0.0022537163,0.0039437097,0.023803007],"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":[0.00007184661,0.000088651905,0.00020547057,0.00042120923,0.000026514077,0.000055285593,0.0006145036,0.0013942295,0.0052758567,0.31824687,0.2652693,0.40833026],"study_design_scores_gemma":[0.000022852128,0.00003362029,0.00036460094,0.00019694364,0.000030357243,0.00018224597,0.0002060351,0.011776441,0.0052649267,0.31788933,0.6639935,0.000039216615],"about_ca_topic_score_codex":0.0019522504,"about_ca_topic_score_gemma":0.0017173233,"teacher_disagreement_score":0.062164698,"about_ca_system_score_codex":0.0029396901,"about_ca_system_score_gemma":0.0016492312,"threshold_uncertainty_score":0.20796162},"labels":[],"label_agreement":null},{"id":"W4288870185","doi":"10.1007/978-3-031-01865-7_7","title":"Profiling Non-Relational Data","year":2019,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Profiling (computer programming); Computer science; Relational database; XML; Big data; Data type; Data science; Data mining; World Wide Web; Programming language","score_opus":0.38729458902149144,"score_gpt":0.40962350864797265,"score_spread":0.022328919626481203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288870185","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063062957,0.00558653,0.90279704,0.0027023123,0.0008379847,0.00012588048,0.00227132,0.0051701064,0.07420251],"genre_scores_gemma":[0.107387826,0.008717021,0.6827199,0.0014144671,0.00072993507,0.00023243483,0.009139308,0.0033187824,0.18634035],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9975483,0.0003479526,0.00013058561,0.0005075237,0.0013455538,0.00012007621],"domain_scores_gemma":[0.99654776,0.0014225207,0.0001275467,0.0012813051,0.0005111482,0.00010963696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027292129,0.00087807793,0.0007222851,0.0025989162,0.0007024418,0.006059555,0.0018190658,0.0007325292,0.018374551],"category_scores_gemma":[0.008541511,0.00081448036,0.00061101554,0.004484198,0.00088209356,0.010766768,0.0028118507,0.0017585761,0.011679797],"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":[0.000059479724,0.000059219932,0.0012933076,0.00036400568,0.000034466007,0.00012365279,0.00052547397,0.0038804337,0.009500325,0.21993534,0.06769543,0.69652885],"study_design_scores_gemma":[0.0000074729314,0.000037122885,0.0017600163,0.00033764186,0.00003137909,0.0006459926,0.00040213682,0.040969193,0.02413295,0.39971173,0.5319034,0.00006098714],"about_ca_topic_score_codex":0.0008043583,"about_ca_topic_score_gemma":0.0016494179,"teacher_disagreement_score":0.018374551,"about_ca_system_score_codex":0.0010596799,"about_ca_system_score_gemma":0.00083838834,"threshold_uncertainty_score":0.06146902},"labels":[],"label_agreement":null},{"id":"W4288917651","doi":"10.1007/978-3-031-01865-7_8","title":"Data Profiling Tools","year":2019,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Profiling (computer programming); Computer science; Visualization; Data science; Data mining; Operating system","score_opus":0.4881953596693671,"score_gpt":0.4276409553592674,"score_spread":0.06055440431009973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288917651","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012519144,0.0021737602,0.7588921,0.002243657,0.0012214902,0.00038224063,0.013113522,0.06314607,0.1575753],"genre_scores_gemma":[0.018824525,0.0040334994,0.5919562,0.0025969243,0.00066157116,0.000679554,0.03641801,0.019039378,0.32579035],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981504,0.0002586443,0.00019342119,0.00037598534,0.0009087806,0.00011260734],"domain_scores_gemma":[0.99657,0.0012084905,0.00013665133,0.0011613199,0.00071984413,0.00020374898],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026942792,0.00186947,0.00090748485,0.0054147243,0.0010826658,0.006754798,0.0022904626,0.0010414514,0.08913841],"category_scores_gemma":[0.008739182,0.0013527245,0.0012296746,0.005773783,0.0006760074,0.008732699,0.0044622687,0.002941071,0.084361866],"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":[0.00006058798,0.000054110533,0.00045169948,0.00040124942,0.000026078154,0.000110407585,0.00031117076,0.0009394155,0.0030475578,0.10063206,0.37175524,0.5222105],"study_design_scores_gemma":[0.000008365812,0.000009193096,0.000252378,0.0001932956,0.000013051677,0.00021683844,0.000078447745,0.0034961912,0.004085098,0.051969863,0.9396511,0.000026334204],"about_ca_topic_score_codex":0.0012196318,"about_ca_topic_score_gemma":0.0015880112,"teacher_disagreement_score":0.08913841,"about_ca_system_score_codex":0.00097611727,"about_ca_system_score_gemma":0.001477637,"threshold_uncertainty_score":0.29819775},"labels":[],"label_agreement":null},{"id":"W4293162521","doi":"10.1007/978-3-031-01865-7","title":"Data Profiling","year":2019,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Profiling (computer programming); Metadata; Computer science; Data science; Research data; World Wide Web; Data curation; Operating system","score_opus":0.4345880521667448,"score_gpt":0.4454418407113966,"score_spread":0.010853788544651799,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293162521","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016718387,0.0042149797,0.54383636,0.005721062,0.0041552936,0.0005056391,0.00816022,0.020313011,0.4114216],"genre_scores_gemma":[0.021372955,0.0046740347,0.20286994,0.003094672,0.0014435337,0.00036206032,0.013134953,0.0064076157,0.74664026],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99833256,0.00022982007,0.00012126361,0.00041040327,0.0008071041,0.00009879679],"domain_scores_gemma":[0.99736017,0.0005840572,0.000091507354,0.0011097747,0.0006927178,0.00016180122],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019771568,0.0013939556,0.00087479356,0.0032153302,0.0012920315,0.005779337,0.0014327876,0.0008742923,0.10738497],"category_scores_gemma":[0.007005809,0.00093418884,0.0009055509,0.0041174833,0.0006849014,0.007572649,0.003806139,0.0022221177,0.10393819],"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":[0.00006758508,0.00002960635,0.0002584284,0.00020188527,0.000012433924,0.000057720994,0.0001284623,0.0008207094,0.0025139384,0.07984988,0.38066104,0.5353983],"study_design_scores_gemma":[0.0000043483,0.000010216455,0.00020970234,0.00010463435,0.000007887666,0.00019130817,0.000053747444,0.00239657,0.002767274,0.038976494,0.9552615,0.000016326703],"about_ca_topic_score_codex":0.0011866377,"about_ca_topic_score_gemma":0.0014982105,"teacher_disagreement_score":0.10738497,"about_ca_system_score_codex":0.0011474584,"about_ca_system_score_gemma":0.0015442468,"threshold_uncertainty_score":0.35923856},"labels":[],"label_agreement":null},{"id":"W4298212960","doi":"10.1007/978-3-031-01839-8_8","title":"Lazy Replication","year":2010,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Commit; Computer science; Protocol (science); Replication (statistics); Distributed computing; Database; Mathematics","score_opus":0.039418800825405415,"score_gpt":0.2657403954992836,"score_spread":0.22632159467387816,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298212960","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0052669006,0.011896284,0.44242162,0.0034121103,0.003424397,0.00015835262,0.0006759893,0.0072480002,0.5254963],"genre_scores_gemma":[0.11753099,0.00831719,0.13009866,0.0013514264,0.0019229049,0.0002292923,0.0015259972,0.0024390623,0.7365845],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99936813,0.000092261565,0.00003080277,0.0001744986,0.00028349616,0.00005070799],"domain_scores_gemma":[0.99928063,0.00012905641,0.000026704653,0.00037673733,0.00014719082,0.000039704377],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005889859,0.00074976287,0.00060768053,0.00076831965,0.0010917063,0.0023881872,0.0011920768,0.0007205197,0.038657933],"category_scores_gemma":[0.0018177493,0.000557504,0.0004713242,0.0013687628,0.0010933164,0.0046057277,0.0016921439,0.0020065964,0.021003561],"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":[0.00007264926,0.000047663907,0.00012388677,0.00021683292,0.000021653874,0.000051387426,0.00027048367,0.0027329242,0.0038016513,0.523608,0.13235657,0.33669633],"study_design_scores_gemma":[0.00002164122,0.000031570573,0.00015886273,0.0000898191,0.000020316655,0.0002110434,0.000057329835,0.007273402,0.0036109337,0.4311052,0.55739653,0.000023386467],"about_ca_topic_score_codex":0.0008008659,"about_ca_topic_score_gemma":0.0011500589,"teacher_disagreement_score":0.038657933,"about_ca_system_score_codex":0.00095452747,"about_ca_system_score_gemma":0.0009389861,"threshold_uncertainty_score":0.12932366},"labels":[],"label_agreement":null},{"id":"W4298392132","doi":"10.1007/978-3-031-01885-5_2","title":"External Construction of Suffix Trees","year":2012,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Generalized suffix tree; Suffix; Computer science; Suffix tree; Section (typography); Scalability; Compressed suffix array; Auxiliary memory; Algorithm; Theoretical computer science; Brute force; Data structure; Programming language; Database","score_opus":0.03614112577618683,"score_gpt":0.24932231780882308,"score_spread":0.21318119203263625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298392132","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009582135,0.00071680895,0.8823618,0.00029316926,0.00054816424,0.00010022669,0.00074005034,0.004182188,0.10147533],"genre_scores_gemma":[0.14152685,0.0018372973,0.73454624,0.0003630179,0.0004909295,0.00033161437,0.0069815363,0.005323603,0.10859893],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9991743,0.00014116101,0.00007443097,0.00020974939,0.0003272228,0.00007312721],"domain_scores_gemma":[0.99873036,0.0003239103,0.000039980936,0.00055048027,0.00030111647,0.000054189517],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00070822693,0.0008323288,0.00073673006,0.0012370788,0.0011758041,0.0029815473,0.0014196715,0.0007986272,0.02735121],"category_scores_gemma":[0.0028698563,0.00073329505,0.0012562877,0.00244109,0.0013622891,0.003784735,0.003328614,0.0018678842,0.016416878],"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":[0.000094400246,0.000057692065,0.0003782185,0.0004605101,0.000034169556,0.0002218609,0.0004885879,0.004535416,0.017075047,0.5496728,0.026303954,0.40067738],"study_design_scores_gemma":[0.000027015212,0.0000676712,0.0003273591,0.0001980143,0.00005784687,0.0007298411,0.00016073571,0.030900916,0.040983252,0.5515384,0.3749656,0.000043392753],"about_ca_topic_score_codex":0.00021942407,"about_ca_topic_score_gemma":0.0003370958,"teacher_disagreement_score":0.02735121,"about_ca_system_score_codex":0.00060519803,"about_ca_system_score_gemma":0.0005420767,"threshold_uncertainty_score":0.09149885},"labels":[],"label_agreement":null},{"id":"W4298412919","doi":"10.1007/978-3-031-01839-8_4","title":"Replication Architecture","year":2010,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Maintainability; Replication (statistics); Computer science; Scalability; Architecture; Task (project management); Distributed computing; Interface (matter); Software engineering; Computer architecture; Embedded system; Engineering; Systems engineering; Operating system","score_opus":0.03188859320682832,"score_gpt":0.2583578644803138,"score_spread":0.22646927127348546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298412919","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0056321467,0.002919528,0.51928794,0.002112468,0.001438756,0.00038829763,0.0009890067,0.011279813,0.45595214],"genre_scores_gemma":[0.13243574,0.004433622,0.32380888,0.00094997196,0.0007850349,0.0006528838,0.0036466627,0.0023204081,0.53096676],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9995276,0.00007520648,0.0000294057,0.00013215636,0.00019208713,0.000043536882],"domain_scores_gemma":[0.9994349,0.00006365723,0.000017108,0.00028930412,0.00015744861,0.000037508693],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00053337525,0.00070310524,0.00043373235,0.00069772755,0.0010435804,0.002793339,0.0016208283,0.00096620456,0.04231833],"category_scores_gemma":[0.0012088795,0.0005797968,0.00051487144,0.0009854396,0.0006357354,0.0031205437,0.0012901581,0.0013636657,0.023901151],"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":[0.0000764223,0.000060371007,0.00020601222,0.00024517626,0.000023061484,0.00008707564,0.00024750968,0.0038353864,0.01175291,0.60414106,0.0942498,0.28507522],"study_design_scores_gemma":[0.000032475375,0.000057329904,0.00025303056,0.0000845365,0.000028163882,0.0003835685,0.00008758089,0.018338736,0.010207016,0.23066226,0.7398355,0.00002976918],"about_ca_topic_score_codex":0.0012528773,"about_ca_topic_score_gemma":0.0014535749,"teacher_disagreement_score":0.04231833,"about_ca_system_score_codex":0.0009517645,"about_ca_system_score_gemma":0.0011916811,"threshold_uncertainty_score":0.14156884},"labels":[],"label_agreement":null},{"id":"W4299149419","doi":"10.1007/978-3-031-01880-0_2","title":"Background: Corpora and Evaluation Methods","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Automatic summarization; Computer science; Natural language processing; Terminology; Information retrieval; Artificial intelligence; Linguistics","score_opus":0.12909839802379802,"score_gpt":0.36666621950415146,"score_spread":0.23756782148035344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299149419","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043393527,0.1135256,0.33462548,0.014791128,0.008545862,0.0034299989,0.12605299,0.021383734,0.37330586],"genre_scores_gemma":[0.032341428,0.07594899,0.4653009,0.005752387,0.011406948,0.013130109,0.2213294,0.013852574,0.16093734],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9916618,0.002889308,0.0011054365,0.0012211222,0.002829095,0.00029315014],"domain_scores_gemma":[0.97779787,0.009660398,0.00078547833,0.0046747276,0.0063739773,0.00070756284],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011861021,0.001924264,0.0019076095,0.01141107,0.0022308126,0.007908425,0.0037325465,0.0022427153,0.106478915],"category_scores_gemma":[0.036325045,0.0018157485,0.00097290165,0.016130548,0.0020571973,0.008550759,0.003450925,0.003625573,0.09412237],"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":[0.00013155509,0.00019881378,0.00034058097,0.0032555896,0.00004388778,0.00009901797,0.00026241867,0.0017861358,0.0019061938,0.034946904,0.542401,0.414628],"study_design_scores_gemma":[0.00005256507,0.00006729339,0.0010953067,0.0015468545,0.000049359787,0.0002828196,0.00011539412,0.0016383545,0.0039502005,0.043340296,0.9478043,0.000057329326],"about_ca_topic_score_codex":0.003631196,"about_ca_topic_score_gemma":0.003764516,"teacher_disagreement_score":0.106478915,"about_ca_system_score_codex":0.002276164,"about_ca_system_score_gemma":0.004100743,"threshold_uncertainty_score":0.35620743},"labels":[],"label_agreement":null},{"id":"W4299335947","doi":"10.1007/978-3-031-01880-0_3","title":"Mining Text Conversations","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Natural language processing; Linguistics; Philosophy","score_opus":0.060711542777487804,"score_gpt":0.2734736238599594,"score_spread":0.21276208108247158,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299335947","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10855678,0.011801819,0.6825737,0.007726374,0.0022389917,0.0012558886,0.07636155,0.024415907,0.085068956],"genre_scores_gemma":[0.36014363,0.0052816723,0.38664845,0.00079733884,0.0017084646,0.001232661,0.16482066,0.002167135,0.07720003],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.997063,0.0007827566,0.00024273088,0.00085770205,0.0008469111,0.00020701649],"domain_scores_gemma":[0.995748,0.002223713,0.00036150767,0.000608586,0.0008220085,0.00023633418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013977888,0.0013411427,0.0008849389,0.00908979,0.0018657562,0.004453789,0.0014578021,0.0011935123,0.017118758],"category_scores_gemma":[0.010551365,0.00060604105,0.0012798383,0.0059401486,0.0006387776,0.0069671962,0.0026175217,0.0017602175,0.021757407],"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":[0.0002453155,0.00025204784,0.008763437,0.0006819373,0.0001549539,0.00042272426,0.0015368538,0.0021877224,0.011064535,0.018906828,0.106993556,0.8487901],"study_design_scores_gemma":[0.00007612995,0.00028905168,0.024029497,0.0012250774,0.00041167715,0.0022181405,0.009530745,0.1902312,0.050107535,0.19104557,0.5306367,0.00019868182],"about_ca_topic_score_codex":0.0016593933,"about_ca_topic_score_gemma":0.0024737325,"teacher_disagreement_score":0.017118758,"about_ca_system_score_codex":0.00080688053,"about_ca_system_score_gemma":0.0013238739,"threshold_uncertainty_score":0.057267904},"labels":[],"label_agreement":null},{"id":"W4299475857","doi":"10.1007/978-3-031-01885-5_1","title":"Structures for Indexing Substrings","year":2012,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Substring; Search engine indexing; Computer science; Information retrieval; Programming language; Data structure","score_opus":0.06175335427616284,"score_gpt":0.2722009832694912,"score_spread":0.21044762899332836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299475857","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007619075,0.015987044,0.83124256,0.001994725,0.00258208,0.00031859247,0.0037446816,0.009142503,0.12736876],"genre_scores_gemma":[0.068348244,0.014546532,0.753335,0.0011456806,0.002011529,0.0007406557,0.011505706,0.0031283745,0.14523832],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989396,0.00010018779,0.00014729236,0.0002188914,0.00051400816,0.0000800329],"domain_scores_gemma":[0.99857545,0.00039322212,0.00008198279,0.0006301564,0.00026097446,0.00005820279],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00074930827,0.0013334595,0.0013318374,0.0045606378,0.0014048236,0.0046005603,0.0026335735,0.0011959221,0.03267192],"category_scores_gemma":[0.003680271,0.0009760016,0.0010967092,0.008771432,0.0021271533,0.009691102,0.0032520567,0.0020364637,0.019947438],"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":[0.000052237025,0.000034113807,0.00012908423,0.0005501024,0.000021336191,0.00006787382,0.00034470702,0.0011644473,0.0043274146,0.40945768,0.05557659,0.5282744],"study_design_scores_gemma":[0.00002445009,0.000055510092,0.00019584961,0.00029728765,0.000043086384,0.00047285,0.00014509544,0.0075480985,0.007721409,0.60713416,0.3763176,0.00004459451],"about_ca_topic_score_codex":0.00093658577,"about_ca_topic_score_gemma":0.0011854442,"teacher_disagreement_score":0.03267192,"about_ca_system_score_codex":0.0013577329,"about_ca_system_score_gemma":0.0010501952,"threshold_uncertainty_score":0.10929847},"labels":[],"label_agreement":null},{"id":"W4299610662","doi":"10.1007/978-3-031-01881-7_6","title":"Updating Data","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Tuple; Schema (genetic algorithms); Relation (database); Consistency (knowledge bases); Data integrity; Set (abstract data type); Theoretical computer science; Programming language; Database; Information retrieval; Mathematics; Artificial intelligence","score_opus":0.10257417112385193,"score_gpt":0.2771190980116017,"score_spread":0.1745449268877498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299610662","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00948101,0.0036185596,0.5682369,0.0050800126,0.0054676724,0.00081197335,0.011966841,0.0430012,0.3523358],"genre_scores_gemma":[0.082209975,0.0049704695,0.3021552,0.0035727634,0.0015351697,0.00046783432,0.021577623,0.010435903,0.57307506],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980679,0.0002454557,0.00015250912,0.0004760484,0.0009367537,0.000121316356],"domain_scores_gemma":[0.9957035,0.0005655862,0.000120394885,0.0027779273,0.0007194175,0.00011319181],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021172764,0.0011966496,0.00081758096,0.0021698538,0.0009228508,0.0044118348,0.0024819225,0.0007660369,0.06835495],"category_scores_gemma":[0.008005534,0.0007378707,0.0008567587,0.002785985,0.0009253368,0.0066843894,0.0032261317,0.0019700965,0.040168546],"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":[0.00013583033,0.00006817705,0.00082777586,0.00043341692,0.000041678813,0.00013049066,0.00048121077,0.0013364071,0.007005145,0.07833052,0.23447455,0.67673475],"study_design_scores_gemma":[0.000017211221,0.000025939984,0.00031009034,0.000089673675,0.000053534754,0.00025506242,0.00010171956,0.003365596,0.014519581,0.024843771,0.9563912,0.00002656952],"about_ca_topic_score_codex":0.0018256423,"about_ca_topic_score_gemma":0.0014861899,"teacher_disagreement_score":0.06835495,"about_ca_system_score_codex":0.0007909091,"about_ca_system_score_gemma":0.0012853574,"threshold_uncertainty_score":0.22867006},"labels":[],"label_agreement":null},{"id":"W4299697650","doi":"10.1007/978-3-031-01880-0_4","title":"Summarizing Text Conversations","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Natural language processing; Linguistics; Philosophy","score_opus":0.05137841069642054,"score_gpt":0.2638444147118555,"score_spread":0.21246600401543494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299697650","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013901742,0.006654597,0.34041756,0.019152695,0.010791068,0.00058475486,0.01899474,0.019260127,0.57024276],"genre_scores_gemma":[0.20956472,0.0053194007,0.14664835,0.0027893428,0.005523974,0.0008582744,0.04118906,0.010771195,0.5773357],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985386,0.0006035158,0.00008478641,0.00028781075,0.00038301296,0.0001022454],"domain_scores_gemma":[0.9969831,0.0013399965,0.0001498841,0.00047466965,0.00092441327,0.00012798829],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010920168,0.0012313225,0.0005278677,0.0036681949,0.002674967,0.004994449,0.0010009478,0.0009862422,0.10804025],"category_scores_gemma":[0.009821166,0.0005153288,0.00046159176,0.0029648647,0.0009678321,0.0066062366,0.0028458012,0.0019613078,0.055046435],"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":[0.0001926373,0.000042274365,0.00041068057,0.00036280783,0.000022653378,0.00022640519,0.004503747,0.00069153676,0.0041335607,0.1540418,0.4984479,0.33692393],"study_design_scores_gemma":[0.0000128667225,0.000020079748,0.0004348539,0.00028587115,0.000024173933,0.000118844015,0.0021242693,0.0027257083,0.0036997548,0.06723307,0.9232952,0.000025211504],"about_ca_topic_score_codex":0.0019135835,"about_ca_topic_score_gemma":0.002268753,"teacher_disagreement_score":0.10804025,"about_ca_system_score_codex":0.0011588864,"about_ca_system_score_gemma":0.0011351904,"threshold_uncertainty_score":0.36143064},"labels":[],"label_agreement":null},{"id":"W4299904045","doi":"10.1007/978-3-031-01885-5_4","title":"Queries for Disk-based Indexes","year":2012,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Computer science","score_opus":0.05805241371313015,"score_gpt":0.2748976804022351,"score_spread":0.21684526668910498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299904045","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0077458313,0.037228134,0.72412866,0.004670619,0.0031162892,0.00023051482,0.0021425704,0.010575167,0.21016222],"genre_scores_gemma":[0.1313535,0.033076514,0.46088913,0.0023343184,0.005395507,0.0004873328,0.0065401928,0.0062045376,0.35371897],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99768233,0.0002759353,0.00022079897,0.00031680794,0.001367736,0.0001364773],"domain_scores_gemma":[0.99847025,0.00064959883,0.000048041446,0.00048044376,0.00029652638,0.000055162054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012932621,0.0011956181,0.0012721979,0.0014808741,0.0010227457,0.006239801,0.0019527253,0.0013193161,0.02826964],"category_scores_gemma":[0.0054985834,0.0010000007,0.0007289554,0.004655824,0.001772842,0.011563472,0.0019866677,0.0026772371,0.013272845],"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":[0.000096018855,0.000055201443,0.00019447853,0.00056541106,0.000027413042,0.00009812551,0.00043518294,0.0025806688,0.0055506737,0.5705007,0.15460426,0.2652918],"study_design_scores_gemma":[0.000032848115,0.000036446057,0.0002815513,0.00018189815,0.000031641826,0.00037464045,0.00011647499,0.018911727,0.005044387,0.4660245,0.5089237,0.000040088562],"about_ca_topic_score_codex":0.0011988955,"about_ca_topic_score_gemma":0.0011051659,"teacher_disagreement_score":0.02826964,"about_ca_system_score_codex":0.001915467,"about_ca_system_score_gemma":0.0007840243,"threshold_uncertainty_score":0.09457141},"labels":[],"label_agreement":null},{"id":"W4300128205","doi":"10.1007/978-3-031-01885-5","title":"Full-Text (Substring) Indexes in External Memory","year":2012,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Substring; Computer science; Type (biology); Natural language processing; Information retrieval; Data structure; Programming language; Biology","score_opus":0.031759205710675927,"score_gpt":0.25630779246615987,"score_spread":0.22454858675548395,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300128205","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018871753,0.034344524,0.707536,0.0011940738,0.0048183035,0.00022933871,0.003048449,0.023020504,0.20693707],"genre_scores_gemma":[0.09143351,0.016047629,0.41591927,0.00094265275,0.0022174842,0.00026975843,0.007229793,0.0059535513,0.45998645],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9995192,0.000043268217,0.00006361126,0.00008001957,0.0002430983,0.000050820236],"domain_scores_gemma":[0.9991098,0.00024959157,0.000048356884,0.00031266708,0.00023273077,0.00004687209],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00043149496,0.0013571125,0.00095355127,0.0017679544,0.00058328017,0.0026704497,0.0017364946,0.00057708786,0.04549244],"category_scores_gemma":[0.0019189771,0.00048229127,0.00045556834,0.0048606107,0.000648335,0.0047009927,0.0015540133,0.00090183935,0.026771855],"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":[0.00021942442,0.000057657708,0.00019655522,0.0006108915,0.000036146434,0.00020447087,0.00013606974,0.0027389228,0.01994489,0.058789667,0.09621038,0.82085496],"study_design_scores_gemma":[0.000075636104,0.00019461718,0.000685325,0.0004377893,0.00010335913,0.0012452976,0.000121167715,0.0228303,0.09073034,0.14105162,0.74244523,0.00007937625],"about_ca_topic_score_codex":0.0005264462,"about_ca_topic_score_gemma":0.0007794073,"teacher_disagreement_score":0.04549244,"about_ca_system_score_codex":0.00057246257,"about_ca_system_score_gemma":0.00059414573,"threshold_uncertainty_score":0.15218735},"labels":[],"label_agreement":null},{"id":"W4300426065","doi":"10.1007/978-3-031-01873-2","title":"On Transactional Concurrency Control","year":2019,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Concurrency control; Concurrency; Computer science; Multiversion concurrency control; Transactional leadership; Transactional analysis; Transactional memory; Optimistic concurrency control; Programming language; Volume (thermodynamics); Control (management); Database; Distributed concurrency control; Database transaction; Artificial intelligence; Psychology; Social psychology","score_opus":0.020988508987970084,"score_gpt":0.2444799830385521,"score_spread":0.223491474050582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300426065","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021111309,0.07668664,0.2880813,0.0042246,0.010286441,0.000102096514,0.00045618106,0.0017240829,0.6163275],"genre_scores_gemma":[0.056096803,0.041551854,0.05679811,0.0016569906,0.005816914,0.00023681785,0.0009421604,0.0010416796,0.83585864],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.999326,0.000073035444,0.0000303349,0.000100777565,0.0004172026,0.000052690157],"domain_scores_gemma":[0.9995173,0.0001938596,0.000018464625,0.00012048811,0.00011553298,0.00003423751],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005940673,0.0012550306,0.00075650634,0.0013011959,0.00069109054,0.0024253607,0.0010083743,0.0007214459,0.034127127],"category_scores_gemma":[0.0013030997,0.0005622923,0.00043416736,0.0025470268,0.0017436933,0.0031298567,0.0012597074,0.0026849231,0.013215613],"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":[0.000045550703,0.000037889116,0.000058914957,0.00029111875,0.000018556013,0.000045035737,0.00019544516,0.0037945765,0.002207248,0.47830975,0.18656132,0.3284346],"study_design_scores_gemma":[0.000016612577,0.000033162905,0.00018337635,0.00018826107,0.000015972184,0.00011853779,0.000035417423,0.008287237,0.0015159638,0.34057912,0.6490079,0.000018363386],"about_ca_topic_score_codex":0.0011939309,"about_ca_topic_score_gemma":0.0014355185,"teacher_disagreement_score":0.034127127,"about_ca_system_score_codex":0.0012635702,"about_ca_system_score_gemma":0.0008212274,"threshold_uncertainty_score":0.11416656},"labels":[],"label_agreement":null},{"id":"W4300637421","doi":"10.1007/978-3-031-01839-8_10","title":"Other Aspects of Replication","year":2010,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Replication (statistics); Psychology; Computer science; Biology; Virology","score_opus":0.03869562300640143,"score_gpt":0.2710489913433356,"score_spread":0.23235336833693415,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300637421","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003199299,0.014857274,0.092976965,0.011989091,0.0057217404,0.00008198911,0.00016191218,0.00044658064,0.8705651],"genre_scores_gemma":[0.21244618,0.013449284,0.04855884,0.007036538,0.010105353,0.00042673486,0.000396832,0.0008556873,0.70672464],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99836534,0.00049630785,0.00006172056,0.00033153617,0.0005998989,0.00014511035],"domain_scores_gemma":[0.99765223,0.00066941546,0.00009326494,0.0011337296,0.0003405653,0.00011081508],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016366489,0.0006542943,0.00074112514,0.0008309903,0.0025834662,0.0047103907,0.00192937,0.0023622024,0.03667115],"category_scores_gemma":[0.006038955,0.00046244884,0.00085550605,0.0018989154,0.004415363,0.009468815,0.0021599764,0.003321408,0.008662139],"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":[0.000009397144,0.0000102977765,0.000031278083,0.00004364869,0.0000038593357,0.00002891448,0.00018848441,0.00036369127,0.00024829738,0.9558835,0.02361037,0.0195783],"study_design_scores_gemma":[0.000007210417,0.0000099566305,0.000056475597,0.000033513632,0.0000046109317,0.00013006016,0.000082014776,0.00074598787,0.00027186502,0.76236224,0.23628896,0.0000070977026],"about_ca_topic_score_codex":0.0019145911,"about_ca_topic_score_gemma":0.0013496887,"teacher_disagreement_score":0.03667115,"about_ca_system_score_codex":0.0019825802,"about_ca_system_score_gemma":0.0013363538,"threshold_uncertainty_score":0.12267727},"labels":[],"label_agreement":null},{"id":"W4300667769","doi":"10.1007/978-3-031-01885-5_3","title":"Scaling Up: When the Input Exceeds the Main Memory","year":2012,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Suffix; String (physics); Computer science; Generalized suffix tree; Suffix tree; Auxiliary memory; Random access; Compressed suffix array; Algorithm; Scaling; Theoretical computer science; Parallel computing; Data structure; Mathematics; Programming language; Computer hardware","score_opus":0.04826054880005584,"score_gpt":0.250437980547949,"score_spread":0.20217743174789318,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300667769","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.26718,0.015297204,0.37441766,0.008270953,0.0056437408,0.0008659441,0.0049199653,0.079094164,0.24431041],"genre_scores_gemma":[0.7630432,0.0042491783,0.11404228,0.0040163537,0.0017896474,0.0006461319,0.0035272457,0.012572868,0.096112974],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988482,0.000090135014,0.000088611254,0.00028307777,0.0004913422,0.0001987013],"domain_scores_gemma":[0.99514455,0.0017569497,0.00020043335,0.0015650213,0.001040085,0.00029290232],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008882851,0.0019689735,0.0015110946,0.0012690744,0.0011148338,0.0035503272,0.001948957,0.0017116397,0.03799129],"category_scores_gemma":[0.009914193,0.0010584374,0.0006763838,0.0017858539,0.0010769694,0.010882688,0.0027337603,0.002797139,0.011903581],"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":[0.0017065962,0.0003154406,0.0045244824,0.0011315435,0.00006529636,0.0022460239,0.0012339439,0.010083857,0.2156731,0.02553831,0.096442595,0.6410387],"study_design_scores_gemma":[0.0001704721,0.0009604744,0.009154959,0.0010575668,0.00025286034,0.0040448285,0.0017825724,0.20116323,0.3984038,0.12581408,0.2569084,0.00028678201],"about_ca_topic_score_codex":0.00094487163,"about_ca_topic_score_gemma":0.0010578719,"teacher_disagreement_score":0.03799129,"about_ca_system_score_codex":0.0007476987,"about_ca_system_score_gemma":0.000787317,"threshold_uncertainty_score":0.1270935},"labels":[],"label_agreement":null},{"id":"W4300907083","doi":"10.1007/978-3-031-01881-7_3","title":"Basic Physical Design and Query Plans","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Teaching and Learning Programming","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Query plan; Focus (optics); Plan (archaeology); Simple (philosophy); Query language; Data structure; Query optimization; Hash function; Theoretical computer science; Information retrieval; Database; Sargable; Programming language; Web search query; Search engine","score_opus":0.0706653485221524,"score_gpt":0.25656319284026274,"score_spread":0.18589784431811035,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300907083","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019334934,0.0011083551,0.85097617,0.0009607124,0.00026037777,0.000490803,0.0016500392,0.0042900294,0.13833003],"genre_scores_gemma":[0.056040548,0.003096788,0.70121443,0.00045604893,0.0002464048,0.0013682818,0.005084841,0.0024403906,0.23005219],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99896204,0.00010059292,0.00009700274,0.00020143014,0.00056319695,0.000075726864],"domain_scores_gemma":[0.99946266,0.00016005794,0.000019822513,0.00018956067,0.00014126896,0.00002658322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010016155,0.001302042,0.0007475206,0.0007861906,0.001093782,0.004202657,0.002254243,0.0011116104,0.063360825],"category_scores_gemma":[0.002635602,0.0015670932,0.0009340695,0.0014164991,0.0018125379,0.004898593,0.0014933406,0.0021360167,0.022473326],"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":[0.00009222525,0.00007275385,0.0001520108,0.0005417009,0.000016168362,0.000086147054,0.00043732335,0.0057968735,0.007980575,0.7433541,0.042842265,0.1986279],"study_design_scores_gemma":[0.000042828953,0.000052702522,0.00026445318,0.00014910361,0.000027224105,0.00029312103,0.00014600006,0.021488035,0.007884061,0.464699,0.50491524,0.000038294067],"about_ca_topic_score_codex":0.002215699,"about_ca_topic_score_gemma":0.0017609583,"teacher_disagreement_score":0.063360825,"about_ca_system_score_codex":0.0010854955,"about_ca_system_score_gemma":0.0016498844,"threshold_uncertainty_score":0.21196306},"labels":[],"label_agreement":null},{"id":"W4300914219","doi":"10.1007/978-3-031-01880-0_5","title":"Conclusions / Final Thoughts","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Digital Communication and Language","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Instant messaging; Variety (cybernetics); Period (music); Social media; World Wide Web; Computer science; Internet privacy; Multimedia; Artificial intelligence; Art","score_opus":0.11919806083317265,"score_gpt":0.2800653757569073,"score_spread":0.16086731492373468,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300914219","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026384755,0.011525955,0.008149169,0.2038212,0.06530159,0.00025615588,0.002697979,0.0008566533,0.7047527],"genre_scores_gemma":[0.043217544,0.009067647,0.007054457,0.05352081,0.011753744,0.00030755773,0.002311303,0.0008348681,0.871932],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974164,0.00054056715,0.00007943045,0.00049491867,0.0009279681,0.00054072076],"domain_scores_gemma":[0.9961372,0.00049973093,0.00012080716,0.00039241053,0.0022033616,0.0006464698],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.003959074,0.0013284765,0.00062532525,0.00148771,0.002976068,0.011377499,0.0026923849,0.0029374463,0.2070402],"category_scores_gemma":[0.008840183,0.0003102459,0.0012152189,0.0010929112,0.00218311,0.0068309614,0.003730143,0.004139958,0.085369855],"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":[0.00017402066,0.000057208938,0.0003055061,0.00045437826,0.000020788562,0.00011909877,0.0007201584,0.0002894662,0.00044398406,0.17662278,0.7474409,0.07335177],"study_design_scores_gemma":[0.000036834947,0.000024683068,0.00044512076,0.0006022145,0.000027118433,0.000059366306,0.0027358313,0.00015926521,0.0008589002,0.08190882,0.9131218,0.00002008839],"about_ca_topic_score_codex":0.010292001,"about_ca_topic_score_gemma":0.010694156,"teacher_disagreement_score":0.7929598,"about_ca_system_score_codex":0.0059235226,"about_ca_system_score_gemma":0.0061060875,"threshold_uncertainty_score":0.6926185},"labels":[],"label_agreement":null},{"id":"W4300916332","doi":"10.1007/978-3-031-01881-7_1","title":"Introduction","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Geology","score_opus":0.042258180871965624,"score_gpt":0.23903430150067678,"score_spread":0.19677612062871117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4300916332","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004159429,0.0030615963,0.0070437505,0.003070222,0.006071924,0.00012689938,0.0023118716,0.0011665846,0.9767313],"genre_scores_gemma":[0.0011682983,0.001399576,0.0015718899,0.0008205437,0.00079204125,0.00006355159,0.0014970055,0.00026688966,0.99242026],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993944,0.00005242615,0.000019199662,0.0001324884,0.00033429652,0.000067162735],"domain_scores_gemma":[0.99926263,0.00008237113,0.000024334888,0.000106934414,0.00035454036,0.0001693016],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0005578091,0.0009499586,0.0005587352,0.0015888738,0.0011550769,0.00374819,0.0014815595,0.001484552,0.5875006],"category_scores_gemma":[0.0019482586,0.00031431988,0.0005171369,0.0014542767,0.00050795457,0.002723928,0.0023601276,0.0016409563,0.51209307],"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":[0.000023444769,0.000031484185,0.000083189974,0.00013985414,0.0000023512714,0.00003651261,0.00007838166,0.00013902673,0.000488466,0.02664315,0.78087413,0.19145998],"study_design_scores_gemma":[0.0000016926325,0.000005837864,0.00006066127,0.000045226898,9.559826e-7,0.000023270137,0.000023211269,0.000026177393,0.00008954944,0.0033186013,0.9964026,0.000002245483],"about_ca_topic_score_codex":0.0018800193,"about_ca_topic_score_gemma":0.0032655413,"teacher_disagreement_score":0.5875006,"about_ca_system_score_codex":0.0012660826,"about_ca_system_score_gemma":0.0017646225,"threshold_uncertainty_score":0.58838034},"labels":[],"label_agreement":null},{"id":"W4301036419","doi":"10.1007/978-3-031-01837-4_2","title":"Data Stream Management Systems","year":2010,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science","score_opus":0.05484729722792187,"score_gpt":0.27223055054017625,"score_spread":0.2173832533122544,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301036419","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064389193,0.008263849,0.6584716,0.0033892023,0.0035698498,0.0005422098,0.0020592695,0.022771375,0.29449373],"genre_scores_gemma":[0.056880355,0.011628128,0.25124797,0.0017997497,0.0024753616,0.00043765915,0.007201429,0.0027897737,0.6655396],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99944085,0.00005372566,0.000036660505,0.00009079933,0.00033860526,0.000039339684],"domain_scores_gemma":[0.9993899,0.00011776856,0.000020663536,0.00017104791,0.00024162004,0.000059008264],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00095688907,0.0008301463,0.0005489772,0.00128983,0.0006446851,0.0030705244,0.0011963408,0.00074675167,0.04164145],"category_scores_gemma":[0.0019261937,0.00051921105,0.00041552563,0.0019264844,0.00042553316,0.0030899316,0.0015027391,0.0013220249,0.02347762],"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":[0.000120807745,0.00010571563,0.00037482663,0.00023430161,0.000026233767,0.00010720206,0.00019898996,0.0022703586,0.008459018,0.08250519,0.25694892,0.64864844],"study_design_scores_gemma":[0.00003251856,0.000048965347,0.00041445906,0.00009036286,0.000024130553,0.00021843793,0.00006170733,0.022757674,0.00829382,0.042374484,0.9256631,0.000020337102],"about_ca_topic_score_codex":0.00091025466,"about_ca_topic_score_gemma":0.0009554838,"teacher_disagreement_score":0.04164145,"about_ca_system_score_codex":0.0006239848,"about_ca_system_score_gemma":0.0008395771,"threshold_uncertainty_score":0.13930452},"labels":[],"label_agreement":null},{"id":"W4301040220","doi":"10.1007/978-3-031-01881-7","title":"Fundamentals of Physical Design and Query Compilation","year":2011,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Executable; Computer science; Domain (mathematical analysis); Query language; Query optimization; RDF query language; Programming language; Software engineering; Information retrieval; Theoretical computer science; Web search query; Web query classification; Mathematics; Search engine","score_opus":0.08689146013064306,"score_gpt":0.2754796345842789,"score_spread":0.18858817445363582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301040220","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009420129,0.00911164,0.91185564,0.001297815,0.0007180635,0.00014045606,0.00045455236,0.0033726592,0.072107024],"genre_scores_gemma":[0.052783653,0.021101361,0.8014955,0.0008213975,0.0010938329,0.00088914763,0.0020086842,0.0028401513,0.11696641],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985045,0.0001887423,0.00013892612,0.00025647026,0.0008129276,0.000098412514],"domain_scores_gemma":[0.9991628,0.0003641639,0.000028956634,0.00023326665,0.00018631866,0.000024599938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012606544,0.0014127736,0.0012868863,0.0014226971,0.00090100017,0.0056155426,0.002597442,0.0012278856,0.030500658],"category_scores_gemma":[0.0035861623,0.0012373922,0.0012781011,0.0027132742,0.0023691931,0.006028117,0.0016742798,0.0022537229,0.01250029],"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":[0.00003776329,0.000044701545,0.00010133833,0.0007848857,0.000030236572,0.000067369154,0.00036707355,0.006568572,0.0031864403,0.61034787,0.052955315,0.32550848],"study_design_scores_gemma":[0.000022224644,0.000019351028,0.000170162,0.00021846525,0.000029939973,0.00024979704,0.00009399336,0.01914725,0.003614917,0.5574811,0.41892263,0.00003014149],"about_ca_topic_score_codex":0.0022873257,"about_ca_topic_score_gemma":0.0017364227,"teacher_disagreement_score":0.030500658,"about_ca_system_score_codex":0.0016770764,"about_ca_system_score_gemma":0.0017203761,"threshold_uncertainty_score":0.10203487},"labels":[],"label_agreement":null},{"id":"W4301284864","doi":"10.1007/978-3-031-01883-1_2","title":"The Notions of Repair and Consistent Answer","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Epistemology, Ethics, and Metaphysics","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Philosophy; History; Psychology; Epistemology","score_opus":0.14527470404633736,"score_gpt":0.26083651861006363,"score_spread":0.11556181456372627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301284864","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0054748813,0.013152496,0.81852347,0.04801436,0.0015689755,0.00015771898,0.0006078884,0.00061413844,0.11188606],"genre_scores_gemma":[0.4796406,0.010574539,0.45615482,0.010685466,0.0053255055,0.0015498416,0.001683487,0.0012188983,0.0331668],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9711719,0.01500589,0.0027063098,0.004396488,0.00566987,0.0010495299],"domain_scores_gemma":[0.90306014,0.07176477,0.0026936068,0.016231373,0.0053993645,0.0008507376],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02909345,0.0015146427,0.0028941913,0.004841668,0.0051102587,0.01373931,0.0056198896,0.010474501,0.017711908],"category_scores_gemma":[0.090966046,0.0021197754,0.0035054954,0.0051607196,0.051631805,0.06131398,0.009271219,0.013802321,0.0032020023],"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":[0.000009813401,0.0000045445627,0.00002015168,0.000038818787,0.0000043852597,0.000009675771,0.00024102548,0.0002046144,0.000024287057,0.99451584,0.0014556311,0.0034712022],"study_design_scores_gemma":[0.00000850354,0.0000027417177,0.000009446211,0.000032426902,0.000004099916,0.000015827756,0.0000801498,0.00046877516,0.00006506868,0.99405247,0.005255245,0.0000052878945],"about_ca_topic_score_codex":0.0023076378,"about_ca_topic_score_gemma":0.0010097813,"teacher_disagreement_score":0.02909345,"about_ca_system_score_codex":0.0055343905,"about_ca_system_score_gemma":0.0041439584,"threshold_uncertainty_score":0.15386266},"labels":[],"label_agreement":null},{"id":"W4301385930","doi":"10.1007/978-3-031-01881-7_5","title":"Query Compilation and Plan Synthesis","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Query optimization; Query language; Sargable; Query plan; Plan (archaeology); Query expansion; Web query classification; Information retrieval; RDF query language; Online aggregation; Web search query; Theoretical computer science; Database; Search engine","score_opus":0.06647717028824002,"score_gpt":0.2419019282067119,"score_spread":0.1754247579184719,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301385930","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032157796,0.0016608735,0.9343953,0.00073441357,0.00029021964,0.0001937617,0.0005046545,0.0043466776,0.054658215],"genre_scores_gemma":[0.11593293,0.0028787907,0.8063751,0.0005486239,0.00039298786,0.00045841892,0.0024723643,0.0034458265,0.06749498],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9983707,0.00028879882,0.00009841322,0.0003763708,0.000728278,0.00013750432],"domain_scores_gemma":[0.9990914,0.00044358257,0.000025259826,0.00023928104,0.00017583612,0.000024754581],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012768424,0.0011723316,0.0009824857,0.0010435291,0.00077150017,0.0031494936,0.0013598294,0.0006304874,0.024521079],"category_scores_gemma":[0.003978521,0.001026237,0.001414626,0.0013452165,0.0015223657,0.0035304888,0.0020746586,0.0023534333,0.0051062214],"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":[0.00011777886,0.000059116843,0.00016792148,0.0004712844,0.00005711933,0.00008924819,0.0003890474,0.023250842,0.0062469705,0.51337814,0.03526771,0.42050487],"study_design_scores_gemma":[0.000050321512,0.00006294924,0.00021853855,0.00018651636,0.00008574906,0.00015788445,0.0002117831,0.10765926,0.023374861,0.66966015,0.19828297,0.000048977676],"about_ca_topic_score_codex":0.0026393146,"about_ca_topic_score_gemma":0.0025209433,"teacher_disagreement_score":0.024521079,"about_ca_system_score_codex":0.0016352782,"about_ca_system_score_gemma":0.0015122995,"threshold_uncertainty_score":0.08203119},"labels":[],"label_agreement":null},{"id":"W4301769198","doi":"10.1007/978-3-031-01881-7_2","title":"Logical Design and User Queries","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science","score_opus":0.10484717758373471,"score_gpt":0.2562253973501651,"score_spread":0.15137821976643037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301769198","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004057529,0.009100452,0.77504444,0.008971114,0.00048378363,0.00009889172,0.00039825402,0.0014416055,0.20040403],"genre_scores_gemma":[0.3160473,0.0128370095,0.4318743,0.0048003513,0.0015844158,0.0010018024,0.0026356387,0.0023499236,0.22686933],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99590683,0.0018210772,0.00029817296,0.0006497218,0.0011087781,0.00021533483],"domain_scores_gemma":[0.997008,0.0018594478,0.00008415062,0.0005584166,0.00041331918,0.00007655807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043659024,0.0010563155,0.001111532,0.001662606,0.0022197415,0.011577226,0.0021955548,0.0022361756,0.019155623],"category_scores_gemma":[0.009599804,0.0013109192,0.0011335437,0.0025874195,0.008221178,0.019998446,0.0028791407,0.003969124,0.005132258],"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":[0.00000717407,0.000008295443,0.000027185939,0.00005115309,0.0000036472268,0.000014707578,0.00039798746,0.0002645122,0.00013699345,0.9800787,0.0058284053,0.013181332],"study_design_scores_gemma":[0.0000064750748,0.0000043065234,0.000022878088,0.000038433547,0.000007391398,0.000045276887,0.00011412456,0.0017666464,0.0003682258,0.9251289,0.07249006,0.0000072602857],"about_ca_topic_score_codex":0.0024849572,"about_ca_topic_score_gemma":0.0017015769,"teacher_disagreement_score":0.019155623,"about_ca_system_score_codex":0.0036286763,"about_ca_system_score_gemma":0.0016491119,"threshold_uncertainty_score":0.06408197},"labels":[],"label_agreement":null},{"id":"W4301778581","doi":"10.1007/978-3-031-01862-6_2","title":"Background","year":2018,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Question answering; Natural language processing; Dialog box; Parsing; Natural language; Section (typography); Context (archaeology); Artificial intelligence; Natural language understanding; Semantic interpretation; Syntax; Linguistics; World Wide Web","score_opus":0.04900264844609401,"score_gpt":0.29041669427853817,"score_spread":0.24141404583244416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301778581","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011031062,0.005476496,0.011212724,0.0044696177,0.003746625,0.00030451495,0.00850158,0.0011771147,0.9640083],"genre_scores_gemma":[0.0061240434,0.005245336,0.0058216676,0.0025508308,0.0013191293,0.00038220486,0.009005492,0.00056504534,0.9689863],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992524,0.000084833424,0.000029753042,0.00021609737,0.00030620332,0.00011068437],"domain_scores_gemma":[0.9990207,0.00014914406,0.000042874457,0.00013515381,0.00040365945,0.0002484797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00084562204,0.0010675046,0.0006510498,0.0016669346,0.0015796963,0.0033299983,0.0018868966,0.0013999316,0.54480577],"category_scores_gemma":[0.002806334,0.00028153168,0.00049136695,0.0015143459,0.0005999031,0.0024730712,0.0022489855,0.0016388694,0.40653628],"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":[0.00010051907,0.00010819353,0.00025443154,0.00046881146,0.0000054433976,0.000108021464,0.00020085332,0.00021236848,0.00075403985,0.081518225,0.7007273,0.21554182],"study_design_scores_gemma":[0.000003850557,0.000011171395,0.000080677506,0.00008361505,0.0000018147557,0.000042392883,0.000032747623,0.000028890196,0.00014127578,0.006571313,0.99299896,0.0000031723757],"about_ca_topic_score_codex":0.002485386,"about_ca_topic_score_gemma":0.0033718178,"teacher_disagreement_score":0.54480577,"about_ca_system_score_codex":0.0013573497,"about_ca_system_score_gemma":0.0026525902,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4301881750","doi":"10.1007/978-3-031-01880-0_1","title":"Introduction","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"The Internet; Politics; Rest (music); Political science; Sociology; History; Media studies; Literature; World Wide Web; Law; Computer science; Art","score_opus":0.10199095845631526,"score_gpt":0.2267811334885907,"score_spread":0.12479017503227544,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301881750","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004927348,0.002078609,0.005180641,0.004918207,0.0074532293,0.00019696378,0.0043585803,0.0013434793,0.9739775],"genre_scores_gemma":[0.0013901005,0.0010605667,0.0015035804,0.001359667,0.00089215033,0.0001066879,0.0026366985,0.00033458127,0.990716],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991277,0.00009137643,0.000029516836,0.00018530738,0.00046954383,0.00009661635],"domain_scores_gemma":[0.9986156,0.00016969435,0.000043125703,0.00018530534,0.0006703783,0.0003159159],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0008682383,0.000875915,0.0005284051,0.001680312,0.0014828426,0.004778203,0.001643764,0.001694075,0.68903244],"category_scores_gemma":[0.003266912,0.000287707,0.0005357027,0.001551731,0.00056476163,0.0029063206,0.0028944127,0.0017368351,0.5916702],"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":[0.00002407613,0.000031198335,0.00010339526,0.00012929316,0.0000018759548,0.00003425578,0.00009801635,0.000076863886,0.00028413953,0.018148363,0.8385496,0.1425189],"study_design_scores_gemma":[0.0000021222725,0.000005704645,0.0000760306,0.000054311702,8.490543e-7,0.000019216692,0.00004276204,0.000017830407,0.00006647495,0.0024413944,0.997271,0.0000022117465],"about_ca_topic_score_codex":0.002926636,"about_ca_topic_score_gemma":0.0046699606,"teacher_disagreement_score":0.68903244,"about_ca_system_score_codex":0.0015776332,"about_ca_system_score_gemma":0.0027478703,"threshold_uncertainty_score":0.44355744},"labels":[],"label_agreement":null},{"id":"W4302041524","doi":"10.1007/978-3-031-01846-6_3","title":"Query Semantics","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Semantics (computer science); Ranking (information retrieval); Tuple; Rank (graph theory); Computer science; Set (abstract data type); Possible world; Position (finance); Information retrieval; Theoretical computer science; Mathematics; Programming language; Discrete mathematics; Combinatorics","score_opus":0.06341861255133178,"score_gpt":0.2510255668324861,"score_spread":0.18760695428115431,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4302041524","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025510127,0.0045253416,0.73980075,0.004199472,0.001176148,0.00029286565,0.0020640416,0.0051752017,0.24021523],"genre_scores_gemma":[0.13568068,0.011130953,0.4791474,0.004722642,0.0027611272,0.00088014745,0.013475994,0.005515828,0.34668526],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99615896,0.0006504574,0.00037827756,0.0008466796,0.001750451,0.00021522344],"domain_scores_gemma":[0.99837935,0.00052814384,0.00005071312,0.00053150667,0.00044309333,0.00006715738],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002347947,0.001551256,0.0012421653,0.0024761693,0.0015184162,0.008373818,0.002146711,0.0015076761,0.03680983],"category_scores_gemma":[0.005685093,0.0010944429,0.0015327944,0.0034735564,0.0028173926,0.015073974,0.0030218717,0.0040257596,0.021649584],"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":[0.000029948847,0.00003084988,0.0000639625,0.00017725627,0.000014089798,0.000030096298,0.00021880527,0.0006383572,0.0010594762,0.87109804,0.039214026,0.08742511],"study_design_scores_gemma":[0.000014138592,0.000016223537,0.00006330854,0.00009523207,0.000022429278,0.00016858849,0.0001022894,0.0049965633,0.0018581365,0.64891756,0.3437265,0.000018990335],"about_ca_topic_score_codex":0.001870823,"about_ca_topic_score_gemma":0.001183715,"teacher_disagreement_score":0.03680983,"about_ca_system_score_codex":0.0022659742,"about_ca_system_score_gemma":0.0018805849,"threshold_uncertainty_score":0.12314117},"labels":[],"label_agreement":null},{"id":"W4302411531","doi":"10.1007/978-3-031-01885-5_5","title":"Conclusions and Open Problems","year":2012,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Substring; Search engine indexing; Computer science; Scalability; Field (mathematics); Sequence (biology); Theoretical computer science; Information retrieval; Data mining; Data structure; Database; Programming language; Mathematics; Biology","score_opus":0.07604548345264996,"score_gpt":0.2840277626762075,"score_spread":0.20798227922355755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4302411531","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004482712,0.043884333,0.059729684,0.2378815,0.044800885,0.000083984545,0.0014617256,0.00093077443,0.60674447],"genre_scores_gemma":[0.10610855,0.051018726,0.05119633,0.05911071,0.04578345,0.00045408317,0.0039246366,0.0015068759,0.6808966],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976877,0.00070045586,0.00009019409,0.00047689676,0.00072686974,0.00031788976],"domain_scores_gemma":[0.99507457,0.0016120244,0.0001455833,0.0007629402,0.001813806,0.0005911044],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003066277,0.0011124908,0.001021186,0.00172322,0.002839061,0.006515961,0.0027431,0.002836074,0.14814353],"category_scores_gemma":[0.012023834,0.0003282483,0.0011550472,0.0021848308,0.0032666512,0.013107763,0.003910219,0.0047973706,0.0430702],"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":[0.00008949279,0.00009118793,0.00015850175,0.00037564494,0.000015935788,0.00006287076,0.0002669772,0.0008656362,0.00021852955,0.55808246,0.3573683,0.082404435],"study_design_scores_gemma":[0.00001357803,0.000010541464,0.00011191401,0.00022705086,0.0000079576,0.000041600433,0.000425049,0.0006110647,0.00016118077,0.71427023,0.2841096,0.00001022437],"about_ca_topic_score_codex":0.0023954166,"about_ca_topic_score_gemma":0.0017893253,"teacher_disagreement_score":0.14814353,"about_ca_system_score_codex":0.0026231396,"about_ca_system_score_gemma":0.0025142843,"threshold_uncertainty_score":0.4955895},"labels":[],"label_agreement":null},{"id":"W4302543190","doi":"10.1007/978-3-031-01881-7_4","title":"On Practical Physical Design","year":2011,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Architecture and Computational Design","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science","score_opus":0.07922744575156246,"score_gpt":0.26473700617450385,"score_spread":0.1855095604229414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4302543190","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011440986,0.007338275,0.090330616,0.0035357934,0.0016734703,0.00003385142,0.00010552256,0.00031577394,0.89552265],"genre_scores_gemma":[0.04306521,0.0075285784,0.028639743,0.0009796636,0.00075827754,0.00011194977,0.00019482333,0.00037711125,0.9183446],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9996871,0.000071011564,0.000009348098,0.00004810976,0.00015568874,0.000028932016],"domain_scores_gemma":[0.99984574,0.00004810781,0.000004555079,0.000050054077,0.00003961084,0.000011956745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00038431352,0.0009852584,0.00052346924,0.0007207749,0.0010695298,0.0020782908,0.0008675436,0.0010202981,0.05701974],"category_scores_gemma":[0.0008655141,0.00050880114,0.0004227641,0.00052921387,0.0022594333,0.0022924684,0.001682779,0.002096144,0.016775679],"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":[0.0000064596184,0.000015429623,0.000034549987,0.000094968214,0.0000049589667,0.000022229611,0.0001560246,0.0026298899,0.00053981703,0.85893583,0.07402825,0.06353161],"study_design_scores_gemma":[0.0000040294667,0.000010689591,0.00006780144,0.000094236726,0.000003581818,0.00004964681,0.000057992762,0.0019491605,0.00045663523,0.41922873,0.57807076,0.000006697016],"about_ca_topic_score_codex":0.0010649861,"about_ca_topic_score_gemma":0.001901339,"teacher_disagreement_score":0.05701974,"about_ca_system_score_codex":0.0016235829,"about_ca_system_score_gemma":0.0008291937,"threshold_uncertainty_score":0.19075006},"labels":[],"label_agreement":null},{"id":"W4388975825","doi":"10.1007/978-3-031-45043-3_4","title":"Text to Data","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Natural language processing; Parsing; Artificial intelligence; Principle of compositionality; Semantic computing; Graph; Embedding; Information retrieval; Semantic Web; Theoretical computer science","score_opus":0.13425949397111772,"score_gpt":0.2972451479885176,"score_spread":0.16298565401739987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388975825","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00031079553,0.0011177729,0.003183657,0.0053929216,0.009771827,0.00042142448,0.09057466,0.0055461703,0.8836808],"genre_scores_gemma":[0.0011145632,0.0006828764,0.0012235893,0.0019414611,0.00094723864,0.00024848556,0.02689243,0.002709998,0.9642394],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9982332,0.00017636477,0.00012023317,0.0003509547,0.00097593846,0.00014338408],"domain_scores_gemma":[0.9931726,0.001421758,0.00025575358,0.0013377537,0.003210416,0.00060172175],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0012307069,0.0014506322,0.0015625883,0.005012567,0.0016503667,0.0094522815,0.002353815,0.00162704,0.8537714],"category_scores_gemma":[0.014573166,0.00076113973,0.000988118,0.006280774,0.0007878461,0.005373264,0.0038391193,0.0027676905,0.8540971],"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":[0.000020442267,0.0000093681965,0.000034720975,0.00013762187,0.0000019179724,0.000009954303,0.000016138594,0.000026893333,0.000094625546,0.0016656416,0.9624136,0.035568964],"study_design_scores_gemma":[0.000007327412,0.0000038105184,0.00010217602,0.00007328122,0.0000017463817,0.000008897094,0.000020479432,0.000028303179,0.000098235294,0.001013128,0.9986388,0.0000038577705],"about_ca_topic_score_codex":0.006282926,"about_ca_topic_score_gemma":0.006240529,"teacher_disagreement_score":0.8537714,"about_ca_system_score_codex":0.0023371384,"about_ca_system_score_gemma":0.003179286,"threshold_uncertainty_score":0.2085774},"labels":[],"label_agreement":null},{"id":"W4388976293","doi":"10.1007/978-3-031-45043-3_5","title":"Evaluation","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; SQL; Subject (documents); Focus (optics); Interface (matter); Information retrieval; Database; World Wide Web","score_opus":0.1483799302113748,"score_gpt":0.31213625806319206,"score_spread":0.16375632785181726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976293","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028142646,0.005290796,0.07818233,0.0067373584,0.0050591114,0.008173441,0.048320282,0.0057414677,0.81435263],"genre_scores_gemma":[0.33171132,0.0035229898,0.08241567,0.006759971,0.0015260301,0.00994793,0.10218526,0.005486017,0.4564448],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98318374,0.005959168,0.0009311369,0.002280505,0.0067588557,0.0008865385],"domain_scores_gemma":[0.9696833,0.009949893,0.0009222093,0.0046193767,0.013869064,0.00095606444],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013838006,0.0015688138,0.0011330545,0.003890877,0.0015074321,0.004991166,0.0020297968,0.001590199,0.17116737],"category_scores_gemma":[0.071816675,0.0003626682,0.001185426,0.0027777809,0.0012249344,0.0031184293,0.0022248568,0.0014295358,0.062971145],"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":[0.002287092,0.00065691525,0.00641181,0.0013328884,0.00019616852,0.00010827815,0.0003595996,0.0028537032,0.0012885865,0.02873404,0.37014684,0.58562404],"study_design_scores_gemma":[0.0008579018,0.0008857172,0.01278136,0.0016876366,0.0005150802,0.00040581267,0.0011913128,0.013640967,0.007121219,0.042203598,0.9185859,0.00012346442],"about_ca_topic_score_codex":0.008399563,"about_ca_topic_score_gemma":0.008085255,"teacher_disagreement_score":0.17116737,"about_ca_system_score_codex":0.0036292197,"about_ca_system_score_gemma":0.0058697667,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4388976676","doi":"10.1007/978-3-031-45043-3_1","title":"Overview","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Casual; Dialog box; Computer science; Rendezvous; Turing; Context (archaeology); World Wide Web; Programming language; Engineering; Political science; History","score_opus":0.14406673079216062,"score_gpt":0.3003912531240619,"score_spread":0.15632452233190128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976676","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006414319,0.01188921,0.010243526,0.008086586,0.0073036263,0.0002045589,0.004493238,0.0022982834,0.9548395],"genre_scores_gemma":[0.0026867301,0.0070017492,0.0038463871,0.0024667685,0.0014319527,0.00013257947,0.00443878,0.00060716417,0.97738796],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99922884,0.00007523327,0.000032069205,0.00014627307,0.00042319103,0.00009433389],"domain_scores_gemma":[0.9990036,0.00010233087,0.000035876354,0.00016195951,0.00046017,0.00023609646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00089262024,0.0008053485,0.00053375616,0.0020959359,0.0013094023,0.0044168057,0.001814519,0.0015941474,0.45746785],"category_scores_gemma":[0.0022714313,0.00039862425,0.00054970605,0.0019980273,0.00050224864,0.003632894,0.0026440371,0.0015379948,0.3536277],"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":[0.000029303434,0.00002165091,0.00012270652,0.0001867213,0.0000040268237,0.000039639566,0.000087874054,0.000096749725,0.0003105843,0.027658097,0.79143214,0.18001044],"study_design_scores_gemma":[0.0000014044093,0.000003128395,0.00004794979,0.000034661465,0.0000010832125,0.00002359558,0.000016329675,0.000012113932,0.000045558205,0.0017549822,0.99805766,0.0000015566771],"about_ca_topic_score_codex":0.0036718321,"about_ca_topic_score_gemma":0.006318627,"teacher_disagreement_score":0.45746785,"about_ca_system_score_codex":0.0015474919,"about_ca_system_score_gemma":0.00290305,"threshold_uncertainty_score":0},"labels":[],"label_agreement":null},{"id":"W4388976681","doi":"10.1007/978-3-031-45043-3_6","title":"Data-to-Text","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; View; Graph database; Information retrieval; Graph; Relational database; Ask price; Database design; Query language; Natural language; Natural language processing; Theoretical computer science","score_opus":0.1522734746914532,"score_gpt":0.29960791980145984,"score_spread":0.14733444511000665,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976681","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00057876884,0.0011483177,0.12134283,0.004632498,0.0041187303,0.00052165875,0.06322752,0.06842936,0.7360004],"genre_scores_gemma":[0.0055644806,0.0013189533,0.030660009,0.0028454121,0.0011199128,0.00050477585,0.060176536,0.034253582,0.86355627],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981614,0.00026821482,0.00016404886,0.00042728905,0.00085827115,0.00012077659],"domain_scores_gemma":[0.9927638,0.0022104955,0.00017826035,0.00262728,0.0018148973,0.00040532672],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0019783888,0.0018101336,0.0013973542,0.0027240564,0.001094435,0.00883363,0.003302669,0.0015556368,0.6889398],"category_scores_gemma":[0.014547537,0.0012195719,0.0013125826,0.0041910307,0.00078285066,0.008739987,0.0051704305,0.0025756338,0.72433686],"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":[0.000043203545,0.00002143476,0.00006101433,0.00023750168,0.000008713879,0.00003686349,0.00005694808,0.00021677472,0.00072877354,0.009178372,0.85124826,0.13816226],"study_design_scores_gemma":[0.000011544414,0.000005962895,0.00007199463,0.000070191534,0.0000043133837,0.000045221142,0.000027389611,0.0004929458,0.00091180374,0.006527395,0.99182117,0.000009990676],"about_ca_topic_score_codex":0.002112794,"about_ca_topic_score_gemma":0.0020594462,"teacher_disagreement_score":0.6889398,"about_ca_system_score_codex":0.001409206,"about_ca_system_score_gemma":0.0020951768,"threshold_uncertainty_score":0.44368958},"labels":[],"label_agreement":null},{"id":"W4388976713","doi":"10.1007/978-3-031-45043-3_2","title":"Building an NLIDB: The Basics","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Leverage (statistics); Sect; Construct (python library); Computer science; Baseline (sea); Rest (music); Programming language; Artificial intelligence; Political science","score_opus":0.09428031179532508,"score_gpt":0.2902417105555458,"score_spread":0.19596139876022073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976713","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014874443,0.0016130733,0.89794725,0.0025314395,0.0005299159,0.00022902848,0.001947026,0.014481141,0.07923369],"genre_scores_gemma":[0.017220212,0.0020887607,0.8920112,0.00086835626,0.0003133646,0.0005446728,0.006515993,0.0070681944,0.07336929],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981888,0.00042582146,0.00015171083,0.0003672204,0.00074193336,0.00012454069],"domain_scores_gemma":[0.99819297,0.0006523421,0.00004967262,0.0005298953,0.00042035026,0.00015479875],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027355277,0.0007780897,0.00095605204,0.0017705534,0.0018627917,0.007484994,0.00229488,0.0013202142,0.058799904],"category_scores_gemma":[0.007882097,0.0011726901,0.0011256814,0.0027395475,0.0014960185,0.012433934,0.0056876736,0.0026841098,0.06308035],"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":[0.00007674133,0.00008820785,0.0006063221,0.00055956363,0.000019795305,0.000102554724,0.0012500678,0.0026741114,0.0047287066,0.20820238,0.14576964,0.63592184],"study_design_scores_gemma":[0.000014744672,0.000015878168,0.0002125669,0.00021868628,0.000015387637,0.00018727362,0.00031540985,0.016203368,0.0045144544,0.12445811,0.85381424,0.000029919145],"about_ca_topic_score_codex":0.0046553886,"about_ca_topic_score_gemma":0.004239787,"teacher_disagreement_score":0.058799904,"about_ca_system_score_codex":0.0015159624,"about_ca_system_score_gemma":0.002016144,"threshold_uncertainty_score":0.19670528},"labels":[],"label_agreement":null},{"id":"W4388976789","doi":"10.1007/978-3-031-45043-3_7","title":"Interactivity","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Interactivity; Computer science; Focus (optics); Dimension (graph theory); Human–computer interaction; Key (lock); Multimedia; Mathematics","score_opus":0.09378483333534199,"score_gpt":0.28771303726163994,"score_spread":0.19392820392629795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976789","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027530675,0.0016872971,0.05460028,0.0018705697,0.00072102685,0.00006868046,0.00034338972,0.0014598934,0.9364957],"genre_scores_gemma":[0.13093789,0.0026959104,0.028978446,0.0023333863,0.00074908545,0.0005487078,0.0017695825,0.0018638627,0.83012307],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99812394,0.0005156778,0.000084420295,0.00041160063,0.00066357583,0.00020085771],"domain_scores_gemma":[0.9982999,0.0005188974,0.000067569206,0.0005974073,0.00033735312,0.00017885977],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012051883,0.0011294144,0.0005411833,0.0014817648,0.0019642725,0.007686677,0.001639756,0.0015987608,0.110038705],"category_scores_gemma":[0.0037006403,0.00057963765,0.00077846274,0.0010373105,0.0026034948,0.010119314,0.006510342,0.002278565,0.036685664],"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":[0.000080575446,0.000053027437,0.0003548848,0.00030329733,0.000022560034,0.00019153365,0.0044587324,0.0003228243,0.004683615,0.78744674,0.07673569,0.12534654],"study_design_scores_gemma":[0.0000109445255,0.000019081748,0.0002861849,0.0001486921,0.000016692053,0.00030221845,0.00091991393,0.0006111024,0.0018120116,0.121796355,0.8740601,0.000016585564],"about_ca_topic_score_codex":0.0012910315,"about_ca_topic_score_gemma":0.0011306184,"teacher_disagreement_score":0.110038705,"about_ca_system_score_codex":0.00090221665,"about_ca_system_score_gemma":0.00089579955,"threshold_uncertainty_score":0.36811614},"labels":[],"label_agreement":null},{"id":"W4388976844","doi":"10.1007/978-3-031-45043-3_3","title":"Data and Query Model","year":2023,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Syntax; Computer science; Programming language; Grammar; Abstract syntax; Natural language processing; Query language; Linguistics; Artificial intelligence; Information retrieval; Philosophy","score_opus":0.10586423601112353,"score_gpt":0.2897658059866748,"score_spread":0.18390156997555124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388976844","genre_codex":"methods","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015299396,0.007978225,0.6399617,0.010419521,0.0024093387,0.00026907286,0.0034039,0.0030758067,0.33095247],"genre_scores_gemma":[0.07942048,0.012595945,0.2986651,0.0078047127,0.0037172837,0.0011615194,0.009106786,0.0036273587,0.58390087],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99711394,0.0004715153,0.00023469445,0.0007076444,0.0013036871,0.0001685756],"domain_scores_gemma":[0.9984195,0.00045183394,0.000045870795,0.000692784,0.00031699092,0.000073054696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023304503,0.0016349613,0.0012984112,0.0020559123,0.001563336,0.010191795,0.0029424995,0.001969118,0.048257995],"category_scores_gemma":[0.005355025,0.001311295,0.0013863293,0.0035068197,0.0036196555,0.017532803,0.0032351066,0.005634194,0.030326683],"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":[0.000010717157,0.000016906793,0.000034312954,0.00007794262,0.000005260557,0.000020494832,0.00010473213,0.0005153848,0.000464311,0.9342508,0.037544437,0.026954662],"study_design_scores_gemma":[0.000009072537,0.000013716941,0.000047530168,0.000066559325,0.000015045183,0.00012040217,0.0000619467,0.0055991574,0.0007695153,0.5766057,0.4166764,0.000014898958],"about_ca_topic_score_codex":0.0038012897,"about_ca_topic_score_gemma":0.002258291,"teacher_disagreement_score":0.048257995,"about_ca_system_score_codex":0.0030997174,"about_ca_system_score_gemma":0.0024185933,"threshold_uncertainty_score":0.16143912},"labels":[],"label_agreement":null},{"id":"W4388985085","doi":"10.1007/978-3-031-45043-3","title":"Natural Language Interfaces to Databases","year":2023,"lang":"en","type":"book","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Database; Natural (archaeology); Computer science; Linguistics; History; Archaeology; Philosophy","score_opus":0.05939884682807921,"score_gpt":0.3151872066011007,"score_spread":0.2557883597730215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388985085","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014831307,0.050703574,0.23455498,0.00732449,0.0041465517,0.0001696887,0.0014190649,0.004308583,0.69588995],"genre_scores_gemma":[0.02171971,0.026704887,0.087516166,0.0035128098,0.0024800585,0.0004610775,0.004229076,0.0021692107,0.851207],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9992586,0.00018705815,0.000054792294,0.00012901715,0.0003278108,0.00004266345],"domain_scores_gemma":[0.9992434,0.00045248788,0.000023129845,0.00011011138,0.00013493959,0.000035974226],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008367917,0.001120633,0.0010634743,0.0020935435,0.0008499338,0.0065554776,0.0014893763,0.0011763497,0.07010029],"category_scores_gemma":[0.0030478537,0.00094904756,0.00065443304,0.0033557825,0.002186346,0.009551227,0.0018514055,0.0025680182,0.027826335],"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":[0.0000241557,0.00003808279,0.000048302674,0.0003982604,0.000011287063,0.000057744062,0.00046336683,0.00042284388,0.0010419228,0.49314404,0.3610824,0.14326757],"study_design_scores_gemma":[0.000009708826,0.000005354262,0.00007072862,0.0001309803,0.0000067338474,0.00008244588,0.000066371824,0.00089602143,0.00031684648,0.22247848,0.7759299,0.0000062974655],"about_ca_topic_score_codex":0.0024989326,"about_ca_topic_score_gemma":0.0030482495,"teacher_disagreement_score":0.07010029,"about_ca_system_score_codex":0.0017204547,"about_ca_system_score_gemma":0.0008798074,"threshold_uncertainty_score":0.23450881},"labels":[],"label_agreement":null},{"id":"W4390776420","doi":"10.1007/978-3-031-45043-3_8","title":"Correction to: Natural Language Interfaces to Databases","year":2024,"lang":"en","type":"book-chapter","venue":"Synthesis lectures on data management","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Natural (archaeology); Database; Natural language processing; Information retrieval; Programming language; History; Archaeology","score_opus":0.038868677709695196,"score_gpt":0.29882066628214676,"score_spread":0.25995198857245155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390776420","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00018786406,0.000839583,0.0035603044,0.12706314,0.85374004,0.000053510677,0.001288356,0.0023096746,0.010957445],"genre_scores_gemma":[0.018190479,0.0037390874,0.011200586,0.1850164,0.34298983,0.0003949265,0.0027806766,0.0041211,0.43156692],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99290633,0.0010172888,0.0009828525,0.0010167514,0.0034196903,0.0006571104],"domain_scores_gemma":[0.9522734,0.008647042,0.0015604394,0.00477604,0.031403966,0.0013390551],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039065485,0.002094246,0.002322032,0.006553338,0.005255555,0.0061909393,0.0047429437,0.010225001,0.11001343],"category_scores_gemma":[0.064003214,0.0013228303,0.0017319543,0.0049589337,0.0036287026,0.0051415768,0.003053485,0.012958982,0.07628045],"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":[0.000012615485,0.0000051743536,0.000022748773,0.000049432096,0.0000054365532,0.00005747955,0.000023857954,0.000022705013,0.000031598604,0.0018331745,0.99374044,0.0041954718],"study_design_scores_gemma":[0.00002731559,0.000012048796,0.00039795705,0.00013083367,0.000023401548,0.00023392947,0.00007724401,0.0004514771,0.00040562398,0.0058731204,0.99233717,0.000029868932],"about_ca_topic_score_codex":0.019183075,"about_ca_topic_score_gemma":0.02363202,"teacher_disagreement_score":0.11001343,"about_ca_system_score_codex":0.006006018,"about_ca_system_score_gemma":0.005163759,"threshold_uncertainty_score":0.36803156},"labels":[],"label_agreement":null}]}