An Illegal Website Family Discovery Method Based on Association Graph Clustering

The explosive growth of illegal websites poses a huge threat to network security. Illegal websites often do not exist independently but carry out illegal activities in a familial manner. Existing research mainly focuses on the recognition of a single illegal website, lacking analysis of the illegal websites’ group characteristics, resulting in the inability to detect families of illegal websites. To this end, we propose a family discovery method for illegal websites based on association graph clustering. Firstly, we qualitatively analyze and quantitatively calculate the multidimensional association relationships between illegal websites. Secondly, we construct an illegal website association graph to establish the representation formalism of the complex multidimensional association relationships. Thirdly, the discovery of illegal website families is achieved by implementing feature decomposition and clustering methods on the association graph. The effectiveness of the proposed method is experimentally verified by identifying a total of 40 families of illegal websites out of a sample of 500 real illegal websites.