Restructuring of Object-Oriented Software Through Cohesion Improvement Using Frequent Usage Patterns

Abstract

Due to wide adoption of object-oriented programming in software development, there is always a requirement to produce well-designed software systems, so that the overall software maintenance cost is reduced and reusability of the component is increased. But, due to prolonged maintenance activities, the internal structure of software system deteriorates. In this situation, restructuring is a widely used solution to improve the overall internal structure of the system without changing its external behavior. As, it is known that, one technique to perform restructuring is to use refactoring on the existing source code to alter its internal structure without modifying its external functionality. However, the refactoring solely depends on our ability to identify various code smells present in the system. Refactoring aims at improving cohesion and reducing coupling in the software system. So, in this paper, a restructuring approach based on refactoring is proposed through improvement in cohesion. This paper focuses on improving the cohesion of different classes of object-oriented software using a newly proposed similarity metric based on Frequent Usage Patterns (FUP). The proposed similarity metric measure the relatedness among member functions of the classes. The metric makes use of FUPs used by member functions. The FUP consists of unordered sequences of member variables accessed by member function in performing its task. The usage pattern includes both direct and indirect usages based on sub-function calls within a member function. Based on the values of the similarity metric, we performed hierarchical agglomerative clustering using complete linkage strategy to cluster member functions. Finally, based on the clusters obtained, the source code of the software is refactored using proposed refactoring algorithm. The applicability of our proposed approach is tested using two java projects related to different domains of real life. The result obtained encourages the applicability of proposed approach in the restructuring of a software system.patterns, refactoring, hierarchical clustering, maintainability. usage patterns, refactoring, hierarchical clustering, maintainability.