Statistical Analysis of Software Coupling Measurement Based on Complex Networks

Abstract

As the scale and complexity of software systems increasing, researches have begun to place much more emphasis on the area of software measurement, especially coupling measurement in object-oriented systems. In order to understand the coupling measure, we make use of complex network theory to characterize and describe its macroscopic properties. A novel software coupling network was proposed firstly, then the well-defined weight was further added to the software coupling network and got the software coupling weight network which was used to calculate coupling measure. Some of the coupling measure methods such as coupling between objects (CBO) and message passing coupling (MPC) were calculated based on the general statistical features of software coupling weight network. The examinations show that both of CBO and MPC have the scale-free property. Finally the relationship between them was interpreted.