{"id":"W4416391996","doi":"10.1186/s43067-025-00288-9","title":"Size matters less: how fine-tuned small LLMs excel in BPMN generation","year":2025,"lang":"en","type":"article","venue":"Journal of Electrical Systems and Information Technology","topic":"Business Process Modeling and Analysis","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Business Process Model and Notation; Python (programming language); Pipeline (software); XML; Process (computing); Process modeling; Benchmark (surveying); Business process modeling","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003300425,0.001439939,0.0006431089,0.001005375,0.0004768758,0.00234596,0.00243033,0.001574363,0.005792116],"category_scores_gemma":[0.01655297,0.0007312619,0.001216784,0.0006719102,0.0007187212,0.00427071,0.001838301,0.002426485,0.003521397],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001404511,"about_ca_system_score_gemma":0.002087274,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01127311,"about_ca_topic_score_gemma":0.01502405,"domain_scores_codex":[0.9981892,0.0006613346,0.000118826,0.0005537395,0.0003434341,0.0001334906],"domain_scores_gemma":[0.9956933,0.002481365,0.0002022197,0.0009030928,0.0004887074,0.0002313829],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001544751,0.0006724303,0.0175014,0.001108709,0.0002252147,0.0007885685,0.0007738835,0.4558876,0.02054302,0.0131162,0.04530669,0.4425315],"study_design_scores_gemma":[0.00007524587,0.00009989281,0.000531665,0.0000462293,0.00002545328,0.00006819787,0.00006279688,0.978835,0.007718735,0.005165579,0.007349188,0.00002197495],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2716341,0.002971428,0.5828631,0.00374156,0.001072073,0.0006578926,0.005167067,0.1177255,0.01416717],"genre_scores_gemma":[0.6207977,0.0006500647,0.3535147,0.001247279,0.00009734296,0.0003107928,0.01297365,0.004737749,0.005670787],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01127311,"threshold_uncertainty_score":0.02241498,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01111295222210184,"score_gpt":0.1993258710873685,"score_spread":0.1882129188652666,"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."}}