LLM-Assisted Semantic Enrichment of BPMN Models for Blockchain-Financial Processes

Benchmark BPMN Model Corpus for Blockchain, Payments, and AI/LLM Explainability Research. This repository contains a curated benchmark set of BPMN models reconstructed from academic research papers. The corpus is being developed as part of an empirical software engineering research effort focused on: BPMN-based process modeling Payment workflows Blockchain-enabled business processes AI/LLM-assisted explainability of process models Research Context This dataset supports our research that combines: Business Process Modeling (BPMN) Payment-centric workflows AI-assisted explainability (LLMs) Empirical software engineering methods Corpus Design Principles The selected BPMN models aim to be: Diverse in domain - Finance - Insurance - Logistics - E-government - Construction - Trade Diverse in financial intent type - Simple direct transfer - Smart contract payment - Escrow Diverse in model structure - Orchestration/process models - Collaboration models - Choreographymodels Source of the Models Most academic papers do not publish executable BPMN files. Therefore, models in this repository were reconstructed manually from figures using bpmn.io for orchestration and collaboration diagrams, and chor-js for choreographies Repository Content models—BPMN diagrams (.bpmn files) descriptions—Model textual description from paper (paper-excerpt) papers—List of source papers (corpus-index) outputs—LLM outputs configuration files code-unified runner manages the model representation, taxonomy injection, and prompt orchestration.