{"id":"W2044979776","doi":"10.1007/s10115-012-0515-8","title":"Anonymizing bag-valued sparse data by semantic similarity-based clustering","year":2012,"lang":"en","type":"article","venue":"Knowledge and Information Systems","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Semantic similarity; Cluster analysis; Information retrieval; Vocabulary; Data publishing; WordNet; Similarity (geometry); Differential privacy; Data mining; Artificial intelligence","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":[0.004118244,0.0006759642,0.002095867,0.002761898,0.001356128,0.002887871,0.002258467,0.001563154,0.001134258],"category_scores_gemma":[0.02023804,0.0007095561,0.001350328,0.004980617,0.002018085,0.005961391,0.004650351,0.002145539,0.0007020319],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001056242,"about_ca_system_score_gemma":0.001769746,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001480144,"about_ca_topic_score_gemma":0.00163161,"domain_scores_codex":[0.99328,0.002542743,0.0004704649,0.00105623,0.00224929,0.0004013192],"domain_scores_gemma":[0.9870929,0.004455748,0.0009748631,0.006067295,0.001216833,0.0001924455],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001128631,0.0003715427,0.003900433,0.000388431,0.0004188031,0.0003735084,0.001110785,0.3969058,0.0145564,0.2106971,0.009605803,0.3605428],"study_design_scores_gemma":[0.00003274419,0.00007598528,0.0005816198,0.00002848649,0.00005649558,0.0002927381,0.0003028399,0.7985351,0.006939175,0.1898286,0.0032835,0.00004268156],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01899691,0.0001508638,0.9790524,0.0003573819,0.0000616936,0.00004013472,0.0003006481,0.0003149972,0.0007250266],"genre_scores_gemma":[0.57665,0.0006402652,0.4158823,0.0003496672,0.0003234239,0.0001580853,0.003219107,0.0001686024,0.002608586],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004118244,"threshold_uncertainty_score":0.02177966,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05574438451177503,"score_gpt":0.2877710545662532,"score_spread":0.2320266700544782,"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."}}