Automated Mining and Checking of Formal Properties in Natural Language Requirements

Bridging the gap between natural language requirements (NLR) and precise formal specifications is a crucial task of knowledge engineering. Software system development has become more complex in recent years, and it includes many requirements in different domains that users need to understand. Many of these requirements are expressed in natural language, which may be incomplete and ambiguous. However, the formal language with its rigorous semantics may accurately represent certain temporal logic properties and allow for automatic validation analysis. It is difficult for software engineers to understand the formal temporal logic from numerous requirements. In this paper, we propose a novel method to automatically mine the linear temporal logic (LTL) from the natural language requirements and check the consistency among different formal properties. We use natural language processing (NLP) to parse requirement sentences and map syntactic dependencies to LTL formulas by using our extraction rules. Also, we apply the automata-based model checking to assess the correctness and consistency of the extracted properties. Through implementation and case studies, we demonstrate that our approach is well suited to deal with the temporal logic requirements upon which the natural language is based.