Empirical Study on Specification Metrics to Predict Volatility and Software Defects

Successful software implementation needs high-quality software requirement specifications (SRSs). However, SRSs are not only difficult to evaluate quantitatively, but there is not an effective indicator to predict which SRSs are prone to modifications. Moreover, few studies have investigated the impact of SRS quality on software quality. Herein we use two specification metrics for SRSs to evaluate their effectiveness to predict future modifications in two actual developments. The results show that our metrics are useful to predict future specification modifications and our specifications are closely related with software quality.