Metrics to Predict Future Modifications and Defects Based on Software Requirements Specifications (SRSs), accepted for IEIE Transactions on Smart Processing and Computing (Scopus indexed)

Taketo Tsunoda, Hironori Washizaki, Yosiaki Fukazawa, Sakae Inoue, Yoshiiku Hanai, Masanobu Kanazawa, “Metrics to Predict Future Modifications and Defects Based on Software Requirements Specifications (SRSs),” IEIE Transactions on Smart Processing and Computing, Vol. 8, pp. 1-9, 2019. (invited from TENCON 2018)(Scopus indexed)

In software development, the quality of the upstream process greatly affects the quality of the downstream process. However, few have applied metrics to estimate quality, controlled the quality quantitatively, or verified the relationship between specifications and software quality. One reason is that specifications are described in a natural language, making it difficult to quantitatively evaluate software metrics such as complexity. Although high-quality software requirement specifications (SRSs) lead to successful implementation, neither a simple quantitative evaluation nor an effective indicator to predict modification-prone SRSs exist. Herein the effectiveness of two specification metrics (number of pages and number of previous modifications) to predict future modifications of SRSs and software defects are evaluated. We confirmed that specification quality measured by the specification metrics and software quality measured by the number of defects are related. We also revealed that future modifications are correlated with the size of SRSs.