Change Impact Analysis by Concept Propagation

Abstract

Software maintenance is an important phase of the software life cycle. An important task in this phase is to locate code fragments affected by user change requests. However, performing this task manually is costly and requires prior knowledge of the software structure. In previous studies, Latent Semantic Indexing (LSI) has been applied to map the user change queries to the relevant code segments automatically. However, due to the lack of domain knowledge embedded in the source code, LSI might be unable to perform this mapping accurately. In this paper, we have proposed a domain knowledge propagation method to obtain more relevant impact set for each change request. This method spreads the user interface originated domain knowledge to the program classes according to the program dependency graph.<br /> The proposed method has been applied to ArgoUML case-study which is an open-source project associated with its change requests. It was observed that applying the concept propagation resulted in 5% increase in the accuracy of the plain LSI method.