Automated Identification of Type-Specific Dependencies between Requirements

Abstract

Requirements Engineering is one of the most important phases in a software project. The elicitation of requirements and the identification of dependencies between these requirements appears to be a challenging task. In this paper, we present an approach to automatically identify requirement dependencies of type requires by using supervised classification techniques. Our results indicate that the implemented approach can detect potential requires dependencies between requirements (formulated on a textual level). We evaluated our approach on a test dataset and figured out that it is possible to identify requirement dependencies with a high prediction quality. We trained and tested our system with different classifiers such as Naive Bayes, Linear SVM, k-Nearest Neighbors, and Random Forest. The results show that Random Forest classifiers correctly predict dependencies with a F1 score of ~82%.