{"id":"W4411120102","doi":"10.18653/v1/2025.naacl-long.187","title":"CluSanT: Differentially Private and Semantically Coherent Text Sanitization","year":2025,"lang":"en","type":"article","venue":"","topic":"Digital and Cyber Forensics","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Victoria","funders":"","keywords":"Computer science; Natural language processing","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":[0.00004088954,0.00008792693,0.00009489628,0.00004777712,0.00006542565,0.0003414804,0.0002584739,0.00003062662,0.00001656993],"category_scores_gemma":[0.00001334394,0.00006762068,0.0000236371,0.000193046,0.00005012256,0.0002695672,0.0003767514,0.00004149981,0.00002542869],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000008628598,"about_ca_system_score_gemma":0.00002387151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000005584424,"about_ca_topic_score_gemma":0.00001940224,"domain_scores_codex":[0.9993613,0.00001242331,0.0001387587,0.0002211522,0.0001223149,0.0001440279],"domain_scores_gemma":[0.9996466,0.00003007743,0.00002138248,0.0002084552,0.0000451328,0.00004833482],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000001666406,0.00003358741,0.0008481382,0.00001158858,0.00001012318,0.000001843826,0.00003469848,0.000001337287,0.0002092981,0.8803213,0.0007720367,0.1177544],"study_design_scores_gemma":[0.001476582,0.0002159547,0.1573956,0.000215409,0.00004290307,0.00002094118,0.00003665471,0.09025814,0.01557507,0.6911063,0.0429199,0.0007366064],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06480743,0.00003492438,0.8154531,0.001199926,0.0002201619,0.000107226,8.762164e-7,0.000159915,0.1180165],"genre_scores_gemma":[0.9890535,0.00001752558,0.006021622,0.0008343552,0.00001288111,0.000003565136,0.000003282402,0.000003179019,0.004050061],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9242461,"threshold_uncertainty_score":0.3292903,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.004422774677562558,"score_gpt":0.2056662444107658,"score_spread":0.2012434697332033,"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."}}