{"id":"W4414622738","doi":"10.1007/s10664-025-10731-0","title":"DeepCodeProbe: Evaluating Code Representation Quality in Models Trained on Code","year":2025,"lang":"en","type":"article","venue":"Empirical Software Engineering","topic":"Software Engineering Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Huawei Technologies (Canada); Polytechnique Montréal","funders":"Fonds de Recherche du Québec - Santé; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Representation (politics); Software quality; Code (set theory); Quality (philosophy); Code smell; Source code; Software; Static program analysis","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.004536868,0.002188565,0.0008471803,0.001684425,0.0005469819,0.00141581,0.002553224,0.002790717,0.003102089],"category_scores_gemma":[0.02156004,0.0007127764,0.001222289,0.001050991,0.001101877,0.003546544,0.001891384,0.003021563,0.001403897],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001772177,"about_ca_system_score_gemma":0.002424664,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01371695,"about_ca_topic_score_gemma":0.01861103,"domain_scores_codex":[0.9977123,0.0007795061,0.0001506444,0.0006595314,0.00049807,0.0001998743],"domain_scores_gemma":[0.985094,0.009859195,0.0005890022,0.002113223,0.001953894,0.0003907566],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002272991,0.001148936,0.03116975,0.0007853502,0.0008048857,0.0002521475,0.0002833311,0.586989,0.01580817,0.002809888,0.03594882,0.3217266],"study_design_scores_gemma":[0.0001007891,0.0002704224,0.001502058,0.00004356056,0.00005841757,0.00004920417,0.00005531287,0.9889631,0.006139468,0.001811763,0.0009886229,0.00001736812],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7248684,0.003413973,0.2232999,0.001506396,0.0007809361,0.0002916768,0.005909774,0.03560763,0.004321308],"genre_scores_gemma":[0.8725873,0.0005239064,0.09947149,0.000653642,0.00008268717,0.000195076,0.020913,0.001876179,0.003696673],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9954631,"threshold_uncertainty_score":0.02727419,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1192894051877803,"score_gpt":0.4194135589370064,"score_spread":0.3001241537492261,"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."}}