{"id":"W4416037651","doi":"10.18653/v1/2025.arabicnlp-sharedtasks.46","title":"Noor at BAREC Shared Task 2025: A Hybrid Transformer-Feature Architecture for Sentence-level Readability Assessment","year":2025,"lang":"","type":"article","venue":"","topic":"Text Readability and Simplification","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Readability; Task (project management); Architecture; Task analysis","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.001714654,0.0009448859,0.001055704,0.0003523131,0.001274341,0.0007732447,0.002432888,0.0005132725,0.0005115017],"category_scores_gemma":[0.0003884051,0.0008275468,0.0009412271,0.001337193,0.0004511648,0.0008528593,0.0004691078,0.0009886419,0.00004567861],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001389723,"about_ca_system_score_gemma":0.001977794,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003549424,"about_ca_topic_score_gemma":0.001923825,"domain_scores_codex":[0.9928886,0.0004504428,0.001368015,0.002926994,0.0009565427,0.001409423],"domain_scores_gemma":[0.9942647,0.00119595,0.0002988216,0.002880573,0.000895366,0.0004645387],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0008590673,0.001791441,0.003375486,0.002397524,0.0004521103,0.00001199601,0.00282043,0.0005166769,0.01462669,0.02864259,0.07391404,0.8705919],"study_design_scores_gemma":[0.004384535,0.001431133,0.1361394,0.0009470945,0.000614173,0.0001133164,0.0007520413,0.2954879,0.04728739,0.05373647,0.4560829,0.00302362],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01152427,0.0004046177,0.900693,0.07309473,0.001546563,0.004600426,0.002028492,0.0002838341,0.005824066],"genre_scores_gemma":[0.8121324,0.0001134833,0.1498881,0.002849519,0.000197042,0.0009150686,0.0005320942,0.00004179999,0.0333305],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8675683,"threshold_uncertainty_score":0.9994175,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03207726001327568,"score_gpt":0.3037281443476246,"score_spread":0.271650884334349,"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."}}