Abstract: A large number of opensource software systems were taken as samples. The evolution trend of average closeness， the stable value of average closeness， the closeness distribution， the closeness distribution of the new added nodes， and the evolution of the top 10 nodes with the highest closeness were analyzed by using complex network as research tool. The analysis results showed that the average closeness of software networks will trend to a stable value. The closeness distribution presented a single peak characteristic， meaning that there are small numbers of modules with large closeness in the software. These nodes are in the core of the software network， and they are the focus of study. It can be concluded that the closeness of software network is a useful parameter to measure software and is significant to the guidance for software development.

Key words: software networks, software evolution, closeness, centralization, complex network