Translating Requirements in Property Specification Patterns using LLMs

This paper introduces ReqH, an innovative tool designed to streamline the translation of natural language requirements into Property Specification Patterns. The tool leverages the capabilities of Large Language Models, which are renowned for their ability to comprehend and generate human-like text. ReqH aims to address the challenges of translating informal requirements into formal specifications, a process that is crucial in industrial contexts, particularly within safety and security-critical domains which demand rigorous formalisation to ensure the reliability and security of systems. We present some preliminary results from evaluating our methodology on a dataset of semi-automatically generated automotive requirements. The findings indicate that Large Language Models, when applied to this translation process, show significant potential for improving the accuracy and efficiency of requirement specification.