REVV: A tool to create a better understanding of software requirements through Information Visualization and NLP

Abstract

Current requirements elicitation techniques are sub-optimal as far as representing requirements inconsistencies and stakeholder disagreements. The literature in Requirements Engineering (RE) has shown that combining humans’ cognitive and analytical capabilities with automated reasoning is an effective combination to achieve such result.

In this work we introduce a novel software tool that blends Natural Language Processing (NLP) and Information Visualization (IV) techniques with the aim of identifying potential ambiguities and missing requirements. For this purpose we have constructed a conceptual framework and built a visualization that is inspired by this framework. In addition to that, this work presents an algorithm for finding ambiguities in a set of user stories. This algorithm is evaluated in a correlation study.

The algorithm and the visualization with its incorporated state-of-the-art IV techniques are all incorporated in the implemented software tool. The usefulness of this tool for identifying ambiguities and missing requirements is assessed in an evaluation study.