Network analysis of software change history for understanding software evolution

When software systems grow large and complex, understanding the dependencies in the system and making changes become difficult. History-based analysis of change dependencies (i.e. co-changing relation among entities) can facilitate system understanding, maintenance and prediction of evolutionary trends. In this paper we propose techniques to understand evolution in software change dependencies. Version repository data of Android platform is used as a case study and analysis of change dependencies is presented. Experimental analysis shows that the proposed approach provides an interesting view of software evolution in the context of change dependencies that reveals previously unknown facts about changing files in the system.