An approach to Verify Compliance of Requirement Template Using Natural Language

Abstract

Requirement Engineering is the process to elicit stakeholder requirements and developing them in to an agreed requirement document. Requirements should serve as the basis and guideline for all the software development lifecycle. Top Level system requirements are typically written in Natural Language (NL) by individuals who are not requirement experts. During Software Development, problems in these requirements can create unnecessary risks that can impact schedule and cost. Requirement Templates address these issues by providing structural rules that increase the precision of natural language requirements. When applying these templates, it is necessary to verify that these requirements are indeed documented according to the template. Manual inspections for this purpose is time consuming and usually take multiple cycles to complete. In this paper we develop an automated approach to conform requirements to two well-known templates in requirement engineering community. Further we attempt to categorize individual requirements by using guidelines of the two mentioned requirement templates.