Automated program repair using genetic programming and model checking

摘要: Automated program repair is still a highly challenging problem mainly due to the reliance of current techniques on test cases validate candidate patches. This leads increasing unreliability final patches since are partial specifications software. In present paper, an automated method proposed by integrating genetic programming (GP) and model checking (MC). Due its capabilities verify finite state systems, MC employed as appropriate criterion for evolving programs calculate fitness in GP. The application evaluation, which novel context repair, addresses important gap heuristic approaches repair. Being focused fault detection based desired aspects, it enables programmers detect faults according definition properties. Creating general method, this characteristic can be effectively customized different domains corresponding faults. Apart from various types faults, capable handling concurrency bugs not case many methods. To evaluate was implemented tool, named JBF, Java programs. meet objectives study, some experiments were conducted certain with known automatically repaired JBF tool. obtained results encouraging remarkably promising.