NLP and LLMs for Regulatory Reporting Obligations

Large corpus of regulatory texts from European Commission are Difficult to read and interpret. Especially, Reporting Obligations, are one type of regulations that are crucial for many organization. Failure to comply the result in subsequent fines. In this work, we present a framework for transforming regulatory texts in reporting obligations into structured format which enable to find out who was a reporter, to report what, to whom and by when. We proposed 2 approaches to extract addresser, addressee, action result, reported verb.