{"id":"W3119848647","doi":"10.2196/23101","title":"Incorporating Domain Knowledge Into Language Models by Using Graph Convolutional Networks for Assessing Semantic Textual Similarity: Model Development and Performance Comparison","year":2021,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"U.S. National Library of Medicine; National Center for Advancing Translational Sciences; National Institute of Mental Health; National Institutes of Health; U.S. Department of Veterans Affairs","keywords":"Computer science; Natural language processing; Artificial intelligence; Semantic similarity; Information retrieval; Encoder; Automatic summarization; Sentence; Documentation; Unified Medical Language System; Graph; Knowledge base; Theoretical computer science","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"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":[0.001114536,0.0002186405,0.0003594179,0.0001019803,0.0005847752,0.0002663175,0.000436161,0.000244184,0.000003034488],"category_scores_gemma":[0.0001121061,0.0002110987,0.00004369373,0.0003616898,0.0001277946,0.001010726,0.0005835395,0.0006540961,0.000001423024],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001449655,"about_ca_system_score_gemma":0.001113946,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001067323,"about_ca_topic_score_gemma":0.00002873649,"domain_scores_codex":[0.9976669,0.00009814279,0.0009274087,0.0002291152,0.0006476924,0.0004307718],"domain_scores_gemma":[0.9985741,0.0002944371,0.0003220344,0.0002692241,0.0002180244,0.000322216],"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":[0.00002601158,0.0003907791,0.01147918,0.004587418,0.0001033382,0.0000200326,0.1139093,0.3697587,0.00005621657,0.0387423,0.002097896,0.4588289],"study_design_scores_gemma":[0.0004912727,0.00002658373,0.00007755015,0.0003622497,0.000005618182,0.00003970421,0.001999388,0.995481,0.00003842713,0.001043062,0.0001863812,0.0002487977],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3459929,0.0003519215,0.6529494,0.0002003053,0.00008266697,0.0001848974,0.000001874933,0.00007933664,0.0001566468],"genre_scores_gemma":[0.5988964,0.000006073341,0.4004834,0.0004590831,0.00004508468,0.00002858878,0.00006074665,0.00000948352,0.00001116897],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6257222,"threshold_uncertainty_score":0.8608354,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03909626719284735,"score_gpt":0.3488209884979172,"score_spread":0.3097247213050698,"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."}}