A methodology for semi-automatic assistance in elicitation and analysis of textual user requirements.

摘要

Requirements Engineering (RE) is a sub-discipline within Software Engineering increasingly recognized as a critical component in the success of a software development project. With the escalating complexity of software requirements, problems of traditional requirements engineering techniques, including the use of natural language text, are becoming increasingly apparent.;The approach is based on a syntactic analysis and formalization of the text written in natural language and enriched with domain-related information extracted from reusable domain-specific data models. The applicability of this research is illustrated with a case study. A prototype implementing our methodology is developed as a proof-of-concept. The results of controlled experiments designed to evaluate our approach prove the validity of the methodology. The thesis discusses future work, issues, problems, and priorities. Furthermore, it proposes recommendations for textual requirements comprehension research.;This research aims to assist software analysts in dealing with the challenges that exist in correctly understanding user requirements during the interactive process of requirement elicitation and analysis. It proposes a methodology related to visualization of textual requirements and was of making them shared, reviewed and debated by the stakeholders. The proposed methodology serves as a basis for a semi-automated process aimed at capturing the conceptual model of the software system under development and its high-level services from user requirements text. The extracted information can be used by analysts in their in-depth study of the requirements text and in avoiding the risks associated with specifying poor or invalid requirements.