{"id":"W4413822057","doi":"10.1109/tbdata.2025.3604171","title":"Optimizing Deduplication Parameters via a Change-Estimation Analytical Model","year":2025,"lang":"en","type":"article","venue":"IEEE Transactions on Big Data","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Data deduplication; Database","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001354556,0.000155966,0.0002076862,0.0008832903,0.0003811737,0.0002505268,0.0009893752,0.000115467,0.00007514007],"category_scores_gemma":[0.0002870401,0.0001341432,0.00006977136,0.002167908,0.00008494111,0.0007057516,0.00001658114,0.000243976,0.0001967769],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006677178,"about_ca_system_score_gemma":0.00008963251,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003545482,"about_ca_topic_score_gemma":0.00004901316,"domain_scores_codex":[0.9976404,0.00007727741,0.0006630519,0.00068436,0.0007089798,0.0002259376],"domain_scores_gemma":[0.9974189,0.0004675943,0.0001617461,0.001612119,0.0002635248,0.00007604848],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004141173,0.0001282591,0.00001163938,0.000004327951,0.00001775363,6.609724e-7,0.0001532673,0.1165293,0.0002861068,0.0006486933,0.002225855,0.8799528],"study_design_scores_gemma":[0.0002750452,0.00001793653,0.00008795336,0.00003256674,0.00003254213,0.000002819449,0.00007173175,0.9941574,0.001138407,0.003039365,0.001018487,0.0001256964],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.007771219,0.00001051155,0.9871938,0.002398549,0.0007584686,0.0002736351,0.0001453952,0.00009387067,0.001354599],"genre_scores_gemma":[0.95604,0.00000934485,0.04090526,0.001915532,0.00003232507,0.00005161007,0.00007216373,0.00001055133,0.0009632309],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9482688,"threshold_uncertainty_score":0.54702,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.4476223141122868,"score_gpt":0.4268905043208662,"score_spread":0.02073180979142053,"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."}}