Mining Suspicious Tax Evasion Groups in a Corporate Governance Network

There is a new tendency for corporations to evade tax via Interest Affiliated Transactions (IAT) that are controlled by a potential "Guanxi" between the corporations' controllers. At the same time, the taxation data is a classic kind of big data. These issues challenge the effectiveness of traditional data mining-based tax evasion detection methods. To address this problem, we first coin a definition of controller interlock, which characterizes the interlocking relationship between corporations' controllers. Next, we present a colored and weighted network-based model for characterizing economic behaviors, controller interlock and other relationships, and IATs between corporations, and generate a heterogeneous information network-corporate governance network. Then, we further propose a novel Graph-based Suspicious Groups of Interlock based tax evasion Identification method, named GSG2I, which mainly consists of two steps: controller interlock pattern recognition and suspicious group identification. Experimental tests based on a real-world 7-year period tax data of one province in China, demonstrate that the GSG2I method can greatly improve the efficiency of tax evasion detection.