{"id":"W4416368800","doi":"10.48550/arxiv.2510.02534","title":"ZeroFalse: Improving Precision in Static Analysis with LLMs","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Software Testing and Debugging Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Static analysis; False positive paradox; Benchmark (surveying); Reliability (semiconductor); Precision and recall; Static program analysis; Software; Java","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0007750505,0.0003358324,0.0005796845,0.0009556821,0.00008516798,0.0002076332,0.001616831,0.0002163157,0.000005480053],"category_scores_gemma":[0.0005732821,0.0002860032,0.0001482198,0.001852277,0.00004454396,0.0001839609,0.00195413,0.0007014101,0.0000106153],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001570681,"about_ca_system_score_gemma":0.0003208286,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0017394,"about_ca_topic_score_gemma":0.000171825,"domain_scores_codex":[0.9975365,0.0001507448,0.0004759768,0.001085184,0.0003767989,0.0003747421],"domain_scores_gemma":[0.9969388,0.0006559743,0.0003040186,0.001861553,0.0001602778,0.00007942232],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000009979241,0.00007931952,0.9582047,0.000152364,0.0001519762,0.00007861506,0.000606326,0.003694111,0.00001888033,0.00009441638,0.000659405,0.03624984],"study_design_scores_gemma":[0.0004593635,0.0002095651,0.6310969,0.001795252,0.0004388046,0.000009303045,0.00001474784,0.3373438,0.001116886,0.02626564,0.0001197411,0.001129967],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3727663,0.0001494907,0.6223801,0.0002209854,0.0001597884,0.0002559049,0.000005567534,0.00384077,0.0002210177],"genre_scores_gemma":[0.7715569,0.00001718918,0.2278202,0.0001914767,0.00002343096,0.0000882857,0.00001494848,0.00001231385,0.0002753711],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.3987905,"threshold_uncertainty_score":0.9999592,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03196078289550708,"score_gpt":0.2928788205595702,"score_spread":0.2609180376640631,"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."}}