Reproduce “Business as Rulesual: A Benchmark and Framework for Business Rule Flow Modeling with LLMs” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chen Yang, Ruping Xu, Ruizhe Li, Bin Cao, Jing Fan
The authors present ExIde to represent business instructions as linked condition–action rules, connected by sequential, conditional, or parallel relations. BREX contains 409 mostly real-world documents, with limited expert-reviewed synthetic augmentation, and 2,855 rules across 30-plus domains. Across 13 models, P5’s mean NER F1 is 0.882 versus P1’s 0.879; the results do not establish safe real-world workflow execution.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.