2019

Mohammadnazar, Hojat; Pulkkinen, Mirja; Ghanbari, Hadi

Measures taken to prevent faults from being introduced or going undetected can secure development of highly reliable software systems. One such measure is analyzing root causes of recurring faults and preventing them from appearing again. Previous methods developed for this purpose have been reactive in nature and relied heavily on fault reporting mechanisms of ogranizations. Additionally, previous efforts lack a defined mechanism for innovating corrective actions. In this study, we strive to complement the existing methods by introducing a proactive and qualitative method that does not rely on fault data. During the course of the research, in addition to an extensive literature search, an empirical field study is conducted with representatives of companies active in safety-critical and business-critical domains. Our proposed method relies on identifying mismatches between development practices and development context in order to predict erratic behaviors. Corrective actions in this method are innovated by resolving these mismatches. The use of the method is demonstrated in two safety-critical projects. Evaluation of the proposed method is done by two experts with respect to proactivity, resource-intensity, and effectiveness.