Framework Information Based Java Software Architecture Recovery

Software systems tend to become more and more complex as they evolve, which makes it difficult to review, understand, and maintain the source code without complete architectural information, especially in case of large-scale systems. Software architecture recovery is considered an important method contributing to solving this problem. Hierarchical clustering is one of the techniques used to extract architectural information from lower level software representations, such as the source code. This paper is aimed at improving the accuracy of existing hierarchical clustering algorithms by allowing users to parameterize and configure framework information as framework-specific features. We have implemented our approach as an Eclipse plugin and have applied it to recovering the architecture of Java programs. Experiments are carried out on our benchmark built upon Java web applications in which the Spring Framework is used. The experimental results show that our approach can improve the accuracy of the recovered architecture to some extent.