{"id":"W7082280679","doi":"10.48448/6wgv-hx27","title":"CausalLink: An Interactive Evaluation Framework for Causal Reasoning","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Dalhousie University","funders":"","keywords":"Causal reasoning; Causal model; Key (lock); Causal inference; Causality (physics); Outcome (game theory); Causal structure; Psychological intervention","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":[0.001835835,0.0002913963,0.0002826937,0.0004080401,0.000324112,0.000361672,0.00206927,0.0003276163,0.000300912],"category_scores_gemma":[0.002446319,0.0002750472,0.00005810383,0.0008924031,0.0003008015,0.0004518785,0.0004557779,0.0003982454,0.00002659415],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001887471,"about_ca_system_score_gemma":0.00165912,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001470154,"about_ca_topic_score_gemma":0.0001029511,"domain_scores_codex":[0.9972456,0.00008110447,0.0002684,0.001169539,0.0007130605,0.0005223285],"domain_scores_gemma":[0.9974855,0.0002691271,0.0003358113,0.001046534,0.0007141269,0.0001489552],"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":[0.00001336696,0.0001956844,0.00008635759,0.0001391498,0.00007328206,0.000009563367,0.0008380597,0.001249578,0.0004425297,0.8240947,0.05731487,0.1155429],"study_design_scores_gemma":[0.0003693576,0.0001669983,0.00006219659,0.0008173188,0.00005668239,0.00001090694,0.0002201426,0.6259673,0.001026087,0.2043928,0.1662683,0.0006419313],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.000008247463,0.0001057621,0.7595897,0.001216373,0.001151039,0.0005134854,0.00001478825,0.0002930725,0.2371075],"genre_scores_gemma":[0.01420947,0.00001907917,0.801122,0.0008211868,0.0007462586,0.0001573992,0.0001078668,0.00004071474,0.1827761],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.6247177,"threshold_uncertainty_score":0.9999702,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03333665887984914,"score_gpt":0.3511308942757784,"score_spread":0.3177942353959292,"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."}}