Study of computational techniques to deal with ambiguity in SRS documents
Mohd Shahid Husain · 2022
According to the CHAOS studies, one of the vital reasons for the failure of most of the software projects is the bad quality of the requirement specifications document. Mostly, the knowledge engineers capture the software requirements in natural languages like English. These software requirement specification (SRS) documents can be ambiguous and inconsistent as ambiguity is an intrinsic characteristic of natural languages. Ambiguity occurs when different stakeholders may interpret a requirement differently. Traditionally, after all the coding is done, the software is tested for bugs and other functional issues. Moreover, there was no effort to ensure the quality of the SRS documents. This chapter gives the overview about the basic concepts of requirement engineering, natural language processing, and ambiguities. This chapter helps the readers to be familiarized with the idea of requirement engineering, and its importance in the software project development life cycle. It also discusses the phenomena of natural language processing and how ambiguities play a vital role in the SRS documentation. This chapter discusses some of the noteworthy works done by the researchers to provide an ambiguity-less SRS expressed in natural languages by using inspection techniques, checklists, controlled languages, and NLP tools. It also discusses different machine learning techniques proposed by the researchers to resolve the ambiguities present in a document. Finally, it gives insight about the state of the art in the field.