{"id":"W3013190089","doi":"10.3724/sp.j.1041.2020.00528","title":"计算机动态测验中问题解决过程策略的分析：多水平混合IRT模型的拓展与应用","year":2020,"lang":"zh","type":"article","venue":"Acta Psychologica Sinica","topic":"Educational and Psychological Assessments","field":"Psychology","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Environmental science","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity","insufficient_payload"],"category_scores_codex":[0.0009339498,0.001262485,0.001461236,0.0001754775,0.0005156774,0.0002926575,0.00290383,0.001578846,0.07606864],"category_scores_gemma":[0.0007173969,0.001076124,0.0008739646,0.001754473,0.0008951733,0.0003114174,0.000437672,0.002460463,0.02970459],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009696802,"about_ca_system_score_gemma":0.0001778702,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002989201,"about_ca_topic_score_gemma":0.000002149271,"domain_scores_codex":[0.9902719,0.001461114,0.00202054,0.003336732,0.0009780347,0.001931713],"domain_scores_gemma":[0.9936965,0.001292175,0.0009967008,0.001976405,0.0002687286,0.001769502],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001857796,0.006538371,0.02242131,0.00009019433,0.0009915757,0.0003408948,0.004287149,0.000002407155,0.002404766,0.03977767,0.8935741,0.02771382],"study_design_scores_gemma":[0.003060905,0.003213821,0.2644559,0.00006876497,0.0001985103,0.00008916447,0.001204182,0.00001221936,0.00001145934,0.00870165,0.7177118,0.001271625],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1278519,0.002268517,0.0001395226,0.1974999,0.006493697,0.001084345,0.0003262844,0.0005495879,0.6637862],"genre_scores_gemma":[0.8815237,0.001045258,0.001542487,0.1037665,0.003456643,0.0001704526,0.0002062754,0.0001316111,0.008157058],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7536718,"threshold_uncertainty_score":0.9998409,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1824294846631266,"score_gpt":0.438946796307523,"score_spread":0.2565173116443963,"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."}}