{"id":"W6979329680","doi":"","title":"Support Evaluation for the TREC 2024 RAG Track: Comparing Human versus LLM Judges","year":2025,"lang":"en","type":"article","venue":"ArXiv.org","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institute for Information and Communications Technology Promotion; Natural Sciences and Engineering Research Council of Canada; National Institute of Standards and Technology; Ministry of Science and ICT, South Korea","keywords":"Scratch; Human error; Language model; Documentation","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008711097,0.0001109683,0.0001368824,0.00006934291,0.000317212,0.000108259,0.0007461583,0.0000473373,0.00005933117],"category_scores_gemma":[0.0001113917,0.00008884445,0.00008220253,0.0002053166,0.00002942274,0.000232755,0.000175321,0.0001244458,0.00004088909],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009052309,"about_ca_system_score_gemma":0.0001169107,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004373624,"about_ca_topic_score_gemma":0.0001322642,"domain_scores_codex":[0.9988712,0.0000516347,0.0002401486,0.000369041,0.000231566,0.0002363776],"domain_scores_gemma":[0.9988624,0.0002775512,0.00006346086,0.0006538016,0.0001104855,0.00003233235],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00012936,0.0003453347,0.4129135,0.0001621638,0.000460038,0.000008074396,0.005106543,0.01846834,0.003566607,0.1486481,0.0377818,0.3724101],"study_design_scores_gemma":[0.002640992,0.0001314938,0.168764,0.00003872827,0.0001374505,0.000001526099,0.0002322699,0.8039038,0.002802052,0.003388671,0.01769973,0.0002593275],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7547894,0.0002108988,0.2282882,0.003080124,0.00238505,0.000542749,9.302041e-7,0.0001280341,0.01057466],"genre_scores_gemma":[0.995547,0.000004224051,0.00180494,0.0002672558,0.0001527028,0.00009354578,0.000005214404,0.000006465167,0.002118655],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7854354,"threshold_uncertainty_score":0.3622971,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1847326692448298,"score_gpt":0.3754335193551211,"score_spread":0.1907008501102913,"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."}}