Evidence-based Software Verification and Validation Technique

Abstract

Current software verification and validation (V&V) practice lacks a systematic and integrated methodology; moreover, the quality of V&V process is difficult to judge. This paper proposes an Evidence-Based software Verification and Validation (EB V&V) technique to solve these problems. In the paper, V&V process is viewed as an evidence searching process; we first propose an evidence taxonomy, which divides evidence into syntactic, semantic, complete, and exeception impact (also called safety) types of evidence. Methods to obtain these types of evidence along with evidence formats have been discussed. A quality-chart approach is then developed to evaluate the quality of V&V process as well as that of the evaluated software. An evidence database can be built to store and organize the obtained evidence for further utilization. A case study is also presented.