Final Thesis: Use of Natural Language Processing for the Evaluation of Requirements Specifications (in German)

Abstract: The main reasons for software project failure are the lack of user involvement, incomplete requirements, unrealistic expectations, and poor quality of requirements specification. A requirements template is a blueprint for the syntactic structure of individual requirements. The application of requirements templates is an effective and simple approach for reducing the ambiguity of natural language requirements and thus increases the precision of the requirements and the quality of the requirements specification. The requirements templates define the syntax for requirements description and are applied for evaluating, whether the requirements follow this syntax. But the manual evaluating of the conformance is a challenge when dealing with a huge amount of requirements. Manual evaluation is time-consuming and expensive. This thesis presents an approach for automated evaluating the compliance of requirement descriptions using natural language processing techniques and requirements templates. This increases the quality of the requirements and project success.

Keywords: Uni1, software monolith, microservices, components

PDF: Master Thesis

Reference: Thi Yen Le. Anwendung von Natural Language Processing zur Evaluierung der Anforderungsbeschreibungen. Master Thesis. Friedrich-Alexander-Universität Erlangen-Nürnberg: 2021.