{"id":"W4394680908","doi":"10.1021/jacs.4c00338","title":"Artificial Intelligence for Retrosynthetic Planning Needs Both Data and Expert Knowledge","year":2024,"lang":"en","type":"article","venue":"Journal of the American Chemical Society","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":32,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Defense Advanced Research Projects Agency; Institute for Basic Science; Canada Research Chairs; Deutsche Forschungsgemeinschaft","keywords":"Retrosynthetic analysis; Bridge (graph theory); Artificial intelligence; Domain (mathematical analysis); Expert system; Domain knowledge; Applications of artificial intelligence; Chemistry; Knowledge base; Process (computing); Computer science; Perspective (graphical); Data science; Management science; Engineering","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.001592404,0.0001339215,0.0002883314,0.00002698753,0.0001493902,0.000288791,0.001264365,0.00003863418,0.0000412708],"category_scores_gemma":[0.0007878084,0.00008216072,0.0001519794,0.0003432193,0.0007790085,0.0002186002,0.000542759,0.0002710251,0.000006273282],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000756287,"about_ca_system_score_gemma":0.0001167296,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001417783,"about_ca_topic_score_gemma":2.342276e-7,"domain_scores_codex":[0.9986521,0.00008344702,0.0004470168,0.0002571132,0.0002934502,0.0002668937],"domain_scores_gemma":[0.9985555,0.0005747588,0.0003303071,0.0003734577,0.00007376742,0.00009225483],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00002976126,0.00002086852,0.00006829481,0.00003811362,0.0000118467,0.000001009212,0.001872665,0.00005996561,0.9873472,0.00021732,0.005591846,0.004741143],"study_design_scores_gemma":[0.00004486553,0.0001407718,0.00007289814,0.0003265798,0.00005884131,0.0001654544,0.001692241,0.09215337,0.8923023,0.003762067,0.009044888,0.0002356995],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9618746,0.0007143867,0.03326517,0.002942978,0.001011125,0.00008482336,0.00002636473,0.00003595369,0.00004462029],"genre_scores_gemma":[0.9643416,0.00005255615,0.03435071,0.0004184011,0.0007789342,0.000002972921,8.892781e-7,0.00001898111,0.00003500294],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09504484,"threshold_uncertainty_score":0.3350417,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04960269485982668,"score_gpt":0.3581205273893054,"score_spread":0.3085178325294788,"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."}}