7d ago · arXiv (AI/CL/LG, keyword-filtered) →
This paper describes an approach and use case to estimate the correctness of LLM-generated COBOL summaries without access to human-authored…