Mining textual requirements to assist architectural software design: a state of the art review

Abstract

Modern Software Engineering (SE) is characterized by the use of several models that establish and show the different states a software product goes through, from its initial conception to its end, passing across its development, setup and maintenance amongst others. Each phase produces a set of deliverables following different documentation standards, but in many cases, natural language text is a key aspect in the elaboration of such documents. This work surveys the state of the art in the application of text mining techniques to architectural software design, discussing first the role of text documents in SE. Specifically the kind of text that can be found and subsequently exploited for assisting in architects in the complex task of designing software. Intelligent text analysis techniques utilized in software engineering tasks across the software life-cycle are detailed in order to analyze works focus on automatically bridging the gap between requirements and software architectures.