Evolution Analysis Of Software System Based On Complex Networks

Software evolution analysis can help us understand and discover the changing process of software structures, and guide us to refactor and maintain the software systems. Previous rese arches show that the software structures represent characteristics of small-world and scale-fre e. They are typical complex networks. Studies of software static structure evolution by using the methods of complex networks can quantify variety of aspects in software evolving proces s, better describe underlying laws, and help software designers grasp changes of software stru ctures in macroscopic view. It is regarded as a novel method to study software evolution.In this thesis,29kinds of software, a total of494versions of the large-scale open source software source code are selected as software samples, each of the software contains a number of versions, and measure the degree of node, betweenness, closeness and standard structure entropy, and other parameters of sample software. We analyze and summarize large amounts of data, reveal and describe the objective laws of software evolution from different angles and at different levels.The centralization of the complex networks can help us find important nodes in the network. This paper calculates the degree centralization index, the betweenness centralization index and the closeness centralization index, and analyzes the evolution rule of the above indexes in the evolution process of software. We find, in the course of software network evolution, that the complexity of the software gradually increased and eventually stabilized at a range and modules of the average share responsibilities of the software functions in gradually decreases in the software.Software static structure has Characters of dissipative structures. In this paper, we use standard structure entropy measure of network software ordered degree to conclude that software architecture evolution toward the ordered and stabilized.The objective laws of software static topology in the evolution process, has important practical significance for developers from on macroscopic to know and master of the whole structure of the software changes, and has an important guiding meaning to the software design and the later maintenance.