Abstract

Khaled Mahar
Automatic Correction of Programming Faults Using Semantic Networks
Software testing and debugging is a vital stage in the software engineering life cycle. It shows how reliable the software is. This paper is concerned about the automatic correction of some logical programming faults. There are researches that solve this problem by using model-based approach, knowledge-based approach, pattern matching, proof-directed debugging, genetic programming, and patches. Improvements in the performance and types of the corrected bugs vary among those techniques. The proposed technique corrects other bug types with improved performance than the knowledge-based and pattern matching techniques. Semantic networks and rule-based systems together are supportive elements for the proposed technique. A detailed description of the proposed technique is illustrated along with a java prototype called Automatic Correction. A comparison is also conducted among the results of the proposed technique and that of two other well known techniques the Precompiled Fault Detection (PFD) and the After Compilation Fault Correction (ALFC) which the proposed technique was built on. This comparison lightens the contributed points and the improvements existed in the proposed technique. It proves the success and effectiveness of the proposed technique.