Multi-layer network community detection model based on attributes and social interaction intensity

Abstract

With the rapid development of mobile communications and electronic technology, relevant network systems have become larger in scale and exhibited different hierarchical relationships. The complex relationships and links among different users make it more difficult to ultimately obtain data on a whole system. Even with the availability of powerful aggregation tools, most companies cannot afford the associated human and financial costs. However, existing multi-layer network community detection methods are well suited for attribute-based community detection. Therefore, this paper proposes a multi-layer local community detection model that is based on attribute and structure information. This model can effectively utilize node attribute information and the similarity strength information revealed by social exchanges to improve the accuracy of community detection in multi-layer networks. Unlike classical multi-layer and global community detection algorithms, this algorithm is robust on most datasets because of its modularity and computational efficiency.