{"id":"W2238711864","doi":"10.1561/9781680830231","title":"Trends in Cleaning Relational Data: Consistency and Deduplication","year":2015,"lang":"en","type":"book","venue":"now publishers, Inc. eBooks","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":84,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Data quality; Computer science; Data deduplication; Data integrity; Relational database; Consistency (knowledge bases); Anomaly detection; Data consistency; Data mining; Data science; Automation; Process (computing); Data warehouse; Database; Quality (philosophy); Artificial intelligence; Engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01098459,0.0006821703,0.0009068547,0.005752338,0.001483241,0.009137965,0.003817372,0.002053112,0.008178614],"category_scores_gemma":[0.03156237,0.0008918704,0.000937893,0.01483554,0.002689613,0.01826668,0.004818475,0.00340649,0.004612912],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002779635,"about_ca_system_score_gemma":0.002466084,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003295495,"about_ca_topic_score_gemma":0.002718965,"domain_scores_codex":[0.9854219,0.002179791,0.001132245,0.001568705,0.009298247,0.0003991807],"domain_scores_gemma":[0.9565598,0.0169724,0.003476115,0.007733978,0.01421522,0.001042458],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001221323,0.00008134067,0.0054431,0.002153936,0.00006082347,0.0002263973,0.00119317,0.001954921,0.004160232,0.07652466,0.1081716,0.7999077],"study_design_scores_gemma":[0.00002497834,0.0001086706,0.005311004,0.001621764,0.00005749675,0.001804945,0.001825817,0.007627277,0.008124229,0.05804368,0.9153707,0.00007936966],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02758486,0.373941,0.3885497,0.1010334,0.006620106,0.0005447807,0.003810104,0.00509287,0.0928233],"genre_scores_gemma":[0.1338096,0.3315489,0.4535151,0.02265475,0.00760151,0.0004296119,0.009878591,0.002350946,0.03821102],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01098459,"threshold_uncertainty_score":0.05809271,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3277735594755778,"score_gpt":0.4006223382773515,"score_spread":0.07284877880177365,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}