Abstract: Complex networks, such as social networks and biological networks, are ubiquitous in real world. Identifying modules and their relationships in complex networks is very important to discover entities and modules behind the networks, and to understand the mechanisms of the networks. In this paper, we analyze modules and their interplays in complex network using Canonical Correlation Analysis (CCA). To improve the interpretability of the results, we further extend the solution of CCA as a LASSO optimization problem, and then propose a new mining algorithm of modules and their relationships. This algorithm enables us to disclose modules and explicitly modeling their relationships simultaneously. Extensive experiments on both synthetic and DBLP datasets demonstrate the effectiveness of the proposed method.

Key words: computer software, complex network, functional module, social network, canonical correlation analysis (CCA)