{"id":"W7017822690","doi":"","title":"CAMEO: Explaining Consensus and Expertise Across MOdels","year":2024,"lang":"en","type":"dissertation","venue":"UWSpace (University of Waterloo)","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Blackberry (Canada)","funders":"","keywords":"Variety (cybernetics); Feature (linguistics); Test (biology); Space (punctuation); Key (lock); Multiple Models","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":[0.0002289349,0.0002933973,0.0004071011,0.0002356074,0.0003503523,0.0001803885,0.0008166565,0.0002939744,0.00002299433],"category_scores_gemma":[0.00001879041,0.000348894,0.0001316329,0.0003280368,0.0001583269,0.0005844427,0.0003413954,0.0003185816,0.0001033118],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009641684,"about_ca_system_score_gemma":0.0001160894,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.02870034,"about_ca_topic_score_gemma":0.04321497,"domain_scores_codex":[0.9982209,0.00006051164,0.0001852402,0.0007330827,0.0003360286,0.0004642616],"domain_scores_gemma":[0.9988458,0.00007111682,0.000177627,0.000538149,0.0002076568,0.0001596993],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0000749202,0.00003874028,0.00001934271,0.0003422034,0.00009419279,0.0005483277,0.9512607,0.000580247,0.003895267,0.01666057,0.001420744,0.02506475],"study_design_scores_gemma":[0.0002827812,0.0001890437,0.00009890903,0.0007251286,0.00007619325,0.00003236794,0.7620048,0.2075924,0.01688842,0.01074879,0.000509185,0.0008519806],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9939094,0.001055346,0.002185043,0.0008739912,0.0006483484,0.0002265062,0.0000200684,0.0002151454,0.000866195],"genre_scores_gemma":[0.7050986,0.0003539352,0.009429015,0.00004202105,0.00004577405,0.00000162732,0.00005632145,0.00004352437,0.2849291],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2888107,"threshold_uncertainty_score":0.9998963,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02551582153685272,"score_gpt":0.2494166335978579,"score_spread":0.2239008120610052,"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."}}