Automated Natural Language Requirements Analysis using General Architecture for Text Engineering (GATE) Framework

Abstract

Stakeholders exchange ideas and describe requirements of the system in natural language at the early stage of software development. These software requirements tend to be unclear, incomplete and inconsistent. However, better quality and low cost of system development are grounded on clear, complete and consistent requirements statements. Requirements boilerplate is an effective way to minimise the ambiguity from the natural language requirements. But manual conformance of natural language requirements with boilerplate is time consuming and difficult task. This paper aims to automate requirements analysis phase using language processing tool. We propose a natural language requirement analysis model. We also present an open source General Architecture for Text Engineering (GATE) framework for automatically checking of natural language requirements against boilerplates for conformance. The evaluation of proposed approach shows that GATE framework is only capable of detecting ambiguity in natural language requirements. We also present the rules to minimise ambiguity, incompleteness, and inconsistency.