Empirical Evaluation of Novel Change Recommender System for Dependency Detection in Object Oriented Environment

Abstract

Maintenance of a software and ensuring its consistency throughout the lifecycle is a challenging as well as a significant task for the maintenance personnel as infinite requests of change keeps on coming from the clients. It takes a vigilant approach from the maintenance staff to carefully analyze each such change request by considering its impact on the existing software functionality. This vigilant approach of analyzing the change request prior to its implementation is called Change Impact Analysis (CIA) and it is the most crucial activity under Software Maintenance. Performing CIA for an Object-Oriented-System becomes more demanding due to the presence of various object-oriented concepts and complex relationships and dependencies among the software artifacts. Detecting these dependencies is a major activity under CIA where several techniques and tools have been proposed in the literature where the focus of researchers has been mostly on JAVA based systems. This paper propose a novel algorithm for software artifact dependency detection which is implemented as a C# based tool which performs the task of detecting and presenting code dependencies and their inter-relationships and present the results in a most intrigued manner. The functionality of proposed approach is unique in a sense that it presents a three way output which provides clarity to the maintenance personnel to understand the code relationships and it further assists them in deciding about the inclusion or exclusion  of  the proposed change request in the software. Also, the paper provides detailed insights about the functionality, performance and effectiveness of the proposed approach by performing an empirical evaluation of the proposed approach on presenting the results of the empirical evaluation.