User Stories to Concept Map: An approach to Visualise Dependencies

Abstract

"Writing the user stories which captures the user’s perspective in the agile framework is the starting step of gathering requirements. As user stories are written informally with fewer restrictions but may get affected by the inherent NL issues such as ambiguity, incompleteness and inter-dependencies. In this thesis work, we have proposed an approach to automatically generate the conceptual model (i.e., concept maps) from the user stories. The approach also identifies the inter-dependencies between the user stories, and subsequently analyses the incompleteness among them. The approach makes use of natural language processing (NLP) techniques for the identification of linguistics patterns. Next, the linguistics patterns are mapped into the concepts and attributes, resulting in the generation of concept maps by applying the proposed heuristic rules. After generating concept maps from the user stories of a software system, we have recognized the dependency between the concepts for a single user story, and are able to identify inter-dependencies between the set of user stories. We have evaluated the applicability of the proposed approach by experimenting on 22 different projects available publicly. On average, we found that the generated concept maps are able to capture inter-dependency with 94.7% accuracy. We have developed tool support for realising the proposed approach."