Software re-engineering using imperialist competitive algorithm

Abstract

Maintenance is one of the important phases of the software development life cycle which contributes to the effective and long term use of any software system. It can become cumbersome and costly for software maintainers when subsystem boundaries are not clearly defined. Further, the problem gets worse due to the system evolution, lack of current documentation and lack of original design documentation. The application of clustering techniques and tools helps software maintenance programmers to recover high-level views of system designs and hence leads to better understanding and maintenance of software systems. In this paper, we have used a sociopolitical Imperialist Competitive Algorithm for software module clustering and compared it with existing evolutionary approaches. We conclude that the novel socio-political approach produces better quality clusters as compared to the earlier evolutionary genetic approaches.