Formalizing and Verifying Natural Language System Requirements using Petri Nets and Context based Reasoning

Natural language descriptions are generally used to describe requirements in reactive systems. Translating the natural language requirements to a more formal specification is a challenging task. One possible approach to handle complex natural language requirements is to convert them to an intermediary formal representation. This intermediate representation is further converted into a more formal representation such as EDT (Expressive Decision Tables). In this paper, we use Petri nets in combination with domain based context reasoning as a tool to model natural language requirements. We have also built a tool, NatEDT, to generate EDT specifications. In a case study, consisting of natural language requirements across three domains, our experimental results show that Petri nets provide an efficient way of formalizing natural language requirements.