Identifying Architecturally Significant Functional Requirements

Abstract

Failure to identify and analyze architecturally significant functional and non-functional requirements (NFRs) early on in the life cycle of a project can result in costly rework in later stages of software development. While NFRs indicate an explicit architectural impact, the impact that functional requirements may have on architecture is often implicit. The skills needed for capturing functional requirements are different than those needed for making architectural decisions. As a result, these two activities are often conducted by different teams in a project. Therefore it becomes necessary to integrate the knowledge gathered by people with different expertise to make informed architectural decisions. We present a study to bring out that functional requirements often have implicit architectural impact and do not always contain comprehensive information to aid architectural decisions. Further, we present our initial work on automating the identification of architecturally significant functional requirements from requirements documents and their classification into categories based on the different kinds of architectural impact they can have. We believe this to be a crucial precursor for recommending specific design decisions. We envisage ArcheR, a tool that (a) automates the identification of architecturally significant functional requirements from requirement specification documents, (b) classify them into categories based on the different kinds of architectural impact they can have, (c) recommend probing questions the business analyst should ask in order to produce a more complete requirements specification, and (d) recommend possible architectural solutions in response to the architectural impact.