{"id":"W4416036783","doi":"10.18653/v1/2025.emnlp-main.442","title":"Beyond Seen Data: Improving KBQA Generalization Through Schema-Guided Logical Form Generation","year":2025,"lang":"","type":"article","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Atomic Energy of Canada Limited; Institute for Catastrophic Loss Reduction","keywords":"Generalization; Set (abstract data type); Feature (linguistics); Relation (database)","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004967055,0.001248287,0.001133204,0.002237186,0.0006774531,0.00229096,0.003884934,0.001816175,0.004192255],"category_scores_gemma":[0.02685319,0.0005427159,0.002183673,0.001936329,0.001260043,0.008316038,0.004273807,0.003200505,0.002350194],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001015599,"about_ca_system_score_gemma":0.002196833,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01163787,"about_ca_topic_score_gemma":0.01618958,"domain_scores_codex":[0.9968135,0.001161908,0.0002774302,0.0009791692,0.0006297038,0.0001382833],"domain_scores_gemma":[0.9895858,0.005077242,0.0003111843,0.003331705,0.00151422,0.0001798474],"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":[0.0004284068,0.0008990199,0.01182639,0.0009762418,0.000362185,0.0004815941,0.001510883,0.1066513,0.0149531,0.01855993,0.06135694,0.7819941],"study_design_scores_gemma":[0.0002513767,0.0002307048,0.002084638,0.0001577859,0.0002247411,0.0004147687,0.0005129034,0.889599,0.01291775,0.06225181,0.03128737,0.00006721573],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1151632,0.005931349,0.8217143,0.004526578,0.0003817077,0.001155469,0.007860047,0.0348282,0.008439157],"genre_scores_gemma":[0.4390686,0.001541587,0.5223401,0.003393131,0.0002017956,0.0005184932,0.02781746,0.001040534,0.004078348],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01163787,"threshold_uncertainty_score":0.0262686,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0634349050262608,"score_gpt":0.3447075123731154,"score_spread":0.2812726073468546,"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."}}