{"id":"W4412396369","doi":"10.1145/3726302.3730305","title":"Benchmarking LLM-based Relevance Judgment Methods","year":2025,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Benchmarking; Relevance (law); Computer science; Data science; Artificial intelligence; Political science; Business","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":[0.000557372,0.00008059473,0.000102248,0.0000884179,0.00007685206,0.00008276871,0.0006035109,0.00003456197,0.0000331276],"category_scores_gemma":[0.00006445255,0.00007226353,0.00004399832,0.00030996,0.00001163715,0.0001340136,0.0002027414,0.00008710269,0.00001052999],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005790879,"about_ca_system_score_gemma":0.00009242079,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002968064,"about_ca_topic_score_gemma":0.000006089942,"domain_scores_codex":[0.9990785,0.00008176765,0.0001778293,0.0003429687,0.0001286146,0.0001903628],"domain_scores_gemma":[0.9990227,0.0002570382,0.00003116655,0.0006178387,0.00003640439,0.00003485525],"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":[0.000001059063,0.00002564503,0.0001765019,0.00001527013,0.000006485448,0.000002243109,0.00004462316,0.004367695,0.0009186859,0.3068036,0.0009432964,0.6866949],"study_design_scores_gemma":[0.0001452743,0.00001465604,0.000201589,0.00003371782,0.000002511755,3.648908e-7,0.000004104239,0.9431467,0.01454167,0.01161422,0.03020561,0.00008956148],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0003999946,0.0001449182,0.957358,0.002959358,0.0006925337,0.00008046517,6.900711e-8,0.0001897257,0.03817494],"genre_scores_gemma":[0.06249658,0.000003702447,0.9327779,0.00284001,0.00002853132,0.00001194885,2.384791e-7,0.00000233619,0.001838797],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.938779,"threshold_uncertainty_score":0.2946821,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02706635386643755,"score_gpt":0.3433562945497067,"score_spread":0.3162899406832691,"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."}}