Empirical Evaluation and Review of a Metrics-Based Approach for Use Case Verification

In this article, an empirical evaluation and review of some metrics–based verification heuristics for use cases are presented. This evaluation is based on empirical data collected from requirements documents developed by Software Engineering students at the University of Seville using REM, a free XML–based requirements management tool developed by one of the authors. The analysis of the empirical data has not only confirmed the validity of the intuitions that gave rise to the verification heuristics but has also made possible to review and adjust some of their parameters, consequently enhancing their accuracy in predicting defects in use cases. One of the most interesting results derived from the analysis of empirical data is a number of possible enhancements that could be applied to the underlying metamodel of use cases implemented in REM, in which the heuristics are based on, thus providing an important feedback to our current research in Requirements Engineering. ACM Classification: D.2.1 (Software – Software Engineering – Requirements/Specifications), D.2.8 (Software – Software Engineering – Metrics), D.2.9 (Software – Software Engineering – Management – Software Quality Assurance)