Chunking Strategy for Retrieval Augmented Generation in Regulation Documents

Large language model (LLM) have been proven to be capable of performing various text processing tasks. LLMs have the potential to be used as a tool for analyzing regulation documents. To enhance the ability of LLMs to analyze regulation documents, we propose a regulation document chunking method that improves the accuracy of the analysis results. Our proposed method produces better accuracy than the sequential chunking method.