{"id":"W3157230690","doi":"10.48550/arxiv.2105.00811","title":"CBench: Towards Better Evaluation of Question Answering Over Knowledge Graphs","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Carleton University","funders":"","keywords":"Computer science; Question answering; Benchmark (surveying); Benchmarking; Suite; Information retrieval; Vocabulary; Set (abstract data type); Graph; Syntax; Artificial intelligence; Natural language processing; Task (project management); Theoretical computer science; Programming language; Linguistics","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":[0.01842633,0.003016223,0.001645374,0.00980896,0.001206526,0.00415685,0.004539714,0.002454928,0.003024876],"category_scores_gemma":[0.08873054,0.0007156657,0.001613417,0.007663028,0.001572268,0.005879862,0.004082507,0.002895891,0.001483916],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00375324,"about_ca_system_score_gemma":0.003055275,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01839183,"about_ca_topic_score_gemma":0.01387538,"domain_scores_codex":[0.9636856,0.01629769,0.003614086,0.003756682,0.01134302,0.001302792],"domain_scores_gemma":[0.9160568,0.04996275,0.004217783,0.01027636,0.01721161,0.002274595],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003600815,0.003798045,0.0419833,0.01150809,0.001629694,0.0008431704,0.004439589,0.1973083,0.05494913,0.03723457,0.1267466,0.5159586],"study_design_scores_gemma":[0.0003908582,0.001934581,0.02490463,0.0005272417,0.0002925681,0.0004608628,0.001996297,0.8301511,0.05233347,0.03466297,0.05211467,0.0002308503],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.3857688,0.007632867,0.449221,0.002085207,0.001140545,0.002778387,0.03527348,0.0928095,0.02329019],"genre_scores_gemma":[0.5661734,0.001185044,0.3368502,0.0008058572,0.0001597171,0.001651413,0.08536782,0.004291839,0.003514866],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01842633,"threshold_uncertainty_score":0.09744889,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.127953138872214,"score_gpt":0.2417656969451876,"score_spread":0.1138125580729736,"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."}}