Абстрактный

Improvisation of Fault Classification and Detection

Thilagavathi R, Anitha Raj B, Gajalakshmi N

Software faults are a major threat for the dependability of software systems. When we intend to study the impact of software faults on software behavior, the issue of distinguishing faults categories and their frequency distribution arises immediately. For this clear detection, clear classification is needed. Very little is actually known about the types of faults that programmers insert into their software. It is becoming more important that these faults are classified into different categories. In this project, a programming technique is implemented where programmers are required to categorize their faults at each iterative build of the software build cycle. Experiments were carried out that measured the number of faults at each build both using this technique and not using this technique. The result suggests that requiring programmers to categorize their faults during the software build cycle decreases the total number of faults in a program. Then Faults are detected based on their classified types. To provide enhanced detection, an efficient graph mining technology is implemented and loc counts are individually features for localization of faults. The project suggests programmers to categorize their faults during the software build cycle and then detecting them will decrease the total number of faults in a program.