JOURNAL OF CHINA UNIVERSITIES OF POSTS AND TELECOM ›› 2018, Vol. 25 ›› Issue (4): 38-47.doi: 10.19682/j.cnki.1005-8885.2018.1015

• Artificial intelligence • Previous Articles     Next Articles

Density PSO-based software module clustering algorithm

Sun Jiaze, Ling Beilei   

  1. 1. School of Computer Science and Technology, Xi’an University of Posts and Telecommunications, Xi’an 710121, China
    2. Shaanxi Key Laboratory of Network Data Intelligent Processing, Xi’an University of Posts and Telecommunications, Xi’an 710121, China
  • Received:2017-11-27 Revised:2018-08-20 Online:2018-08-30 Published:2018-11-02
  • Contact: Sun Jiaze,
  • About author:Sun Jiaze,
  • Supported by:
    This work was supported by the special fund for key discipline construction of general institutions of higher learning from Shaanxi Province, and the Industrial Research Project of Shaanxi Province (2018GY -014).

Abstract: Software module clustering is to divide the complex software system into many subsystems to enhance the intelligibility and maintainability of software systems. To increase convergence speed and optimize clustering solution, density PSO-based (DPSO) software module clustering algorithm is proposed. Firstly, the software system is converted into complex network diagram, and then the particle swarm optimization (PSO) algorithm is improved. The shortest path method is used to initialize the swarm and the probability selection approach is used to update the particle positions. Furthermore, density-based modularization quality (DMQ) function is designed to evaluate the clustering quality. Five typical open source projects are selected as benchmark programs to verify the efficiency of
the DPSO algorithm. Hill climbing (HC) algorithm, genetic algorithm (GA), PSO and DPSO algorithm are compared in the modularization quality (MQ) and DMQ. The experimental results show that the DPSO is more stable and more convergent than other traditional three algorithms. The DMQ standard is more reasonable than MQ standard in guiding software module clustering.

Key words: software module clustering, complex network, PSO, MQ, modularity density

CLC Number: