{"id":"W4403950719","doi":"10.23889/ijpds.v9i5.2932","title":"Leveraging Full Count Census Data through Record Linkage","year":2024,"lang":"en","type":"article","venue":"International Journal for Population Data Science","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadiana.org","funders":"","keywords":"Census; Record linkage; Computer science; Linkage (software); Count data; Data science; Statistics; Medicine; Population; Biology; Environmental health; Mathematics; Genetics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"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":[0.09699666,0.001108795,0.001649227,0.01633536,0.004061727,0.01552897,0.005240385,0.002760464,0.009683423],"category_scores_gemma":[0.2374557,0.001943354,0.002120149,0.02653952,0.001985287,0.01867303,0.01663855,0.004496115,0.006163872],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005069987,"about_ca_system_score_gemma":0.01800083,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05431979,"about_ca_topic_score_gemma":0.05091891,"domain_scores_codex":[0.9068782,0.05398287,0.004967794,0.007089298,0.02510753,0.001974369],"domain_scores_gemma":[0.8555277,0.07667145,0.00527556,0.0257027,0.03474145,0.00208117],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00008703401,0.00009958618,0.01256384,0.001346789,0.0003737876,0.0001854053,0.006238254,0.005081367,0.0005690641,0.06547791,0.1809989,0.7269781],"study_design_scores_gemma":[0.0001077835,0.00009052157,0.007650192,0.002577783,0.0002460786,0.0002625691,0.005081887,0.01670833,0.001517733,0.1383813,0.8270872,0.000288676],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0139275,0.01233701,0.8292179,0.0571722,0.003382422,0.002954364,0.01836703,0.005389821,0.05725194],"genre_scores_gemma":[0.06565481,0.0212508,0.8630447,0.008592328,0.002738264,0.003710344,0.02016156,0.002186562,0.01266072],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.09699666,"threshold_uncertainty_score":0.5129735,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.5370824940918127,"score_gpt":0.5509539962253384,"score_spread":0.01387150213352562,"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."}}