{"id":"W4412377345","doi":"10.1145/3726302.3730159","title":"A Human-AI Comparative Analysis of Prompt Sensitivity in LLM-Based Relevance Judgment","year":2025,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Relevance (law); Sensitivity (control systems); Computer science; Artificial intelligence; Political science; Engineering; Law","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":[0.036864,0.001005092,0.0009425845,0.002866886,0.0010762,0.002948158,0.001156521,0.001695697,0.00324944],"category_scores_gemma":[0.3123539,0.0005604396,0.0006041109,0.001747735,0.001494123,0.002687427,0.00364218,0.00197881,0.002034115],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001393279,"about_ca_system_score_gemma":0.001099809,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001902759,"about_ca_topic_score_gemma":0.002007543,"domain_scores_codex":[0.9451596,0.03803252,0.003073966,0.00688635,0.006011467,0.0008361734],"domain_scores_gemma":[0.5097679,0.4282497,0.01247983,0.02178894,0.02444322,0.003270389],"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":[0.01833126,0.002257894,0.1557063,0.005745331,0.001076362,0.001230065,0.04612036,0.0274177,0.1209058,0.006756207,0.0326147,0.581838],"study_design_scores_gemma":[0.001988569,0.008620767,0.5448107,0.001500965,0.000908679,0.003856312,0.01629737,0.2103116,0.1052348,0.03982017,0.06498644,0.001663647],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8912592,0.003391689,0.07879243,0.0005878552,0.0004922155,0.001000765,0.001772526,0.004101698,0.0186016],"genre_scores_gemma":[0.9550153,0.0003113018,0.03906139,0.0004052148,0.0001214391,0.0006512047,0.001991463,0.000555903,0.001886676],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.036864,"threshold_uncertainty_score":0.1949578,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0379838156217058,"score_gpt":0.3335992028802042,"score_spread":0.2956153872584985,"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."}}