{"id":"W4387163691","doi":"10.1109/icalt58122.2023.00103","title":"Innovating Assessment with Conversational Agents: A Technology-Enhanced Approach to Formative Assessments","year":2023,"lang":"en","type":"article","venue":"","topic":"Intelligent Tutoring Systems and Adaptive Learning","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta; Concordia University of Edmonton","funders":"","keywords":"Conversation; Formative assessment; Casual; Computer science; Quality (philosophy); Multimedia; Psychology; Mathematics education","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040596,0.0001460965,0.0001591023,0.0004723888,0.000224042,0.0001677352,0.000511234,0.00005012357,0.0000109057],"category_scores_gemma":[0.00002641078,0.000113747,0.00002313244,0.002509905,0.00001986079,0.000531974,0.0003063054,0.000204385,0.0001736165],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001470381,"about_ca_system_score_gemma":0.00008849789,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001253069,"about_ca_topic_score_gemma":6.490339e-7,"domain_scores_codex":[0.9985166,0.00003821294,0.0002538188,0.0003858823,0.0004639461,0.0003414964],"domain_scores_gemma":[0.9992031,0.00004709773,0.0001259767,0.0002838571,0.0002858022,0.00005415596],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000003103192,0.00005296209,0.00222224,0.00002263383,0.00006986042,0.00000486037,0.001728859,0.004897885,0.001248008,0.9715558,0.0007158681,0.01747792],"study_design_scores_gemma":[0.001673256,0.001153227,0.03507163,0.0004375838,0.00001270005,0.00003063156,0.01333959,0.8977321,0.01317774,0.00382863,0.0322298,0.001313084],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03750564,7.464054e-7,0.9039147,0.000441884,0.0001684231,0.0003096914,9.086812e-7,0.000655292,0.05700266],"genre_scores_gemma":[0.7260731,2.752496e-7,0.2678073,0.0002348219,0.00002756641,0.0001253871,0.000007393671,0.000008817973,0.00571532],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9677272,"threshold_uncertainty_score":0.4638469,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03761938246569699,"score_gpt":0.3214596459764846,"score_spread":0.2838402635107876,"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."}}