Multiplex network community detection algorithm based on motif awareness

Multiplex networks have shown high utility in modeling complex systems, e.g., academic social networks. Composite community structures contribute to revealing meaningful grouping patterns in multiplex networks. Multiplex network community detection, especially using higher-order interaction information, has become a popular research topic in complex network analysis. We propose a multiplex network community detection algorithm based on motif awareness (CDMA) to reduce information loss during the aggregation of multiplex networks, which can solve the community detection problem for cold start nodes and improve the quality of multiplex network community detection. Interlayer differences are fully considered in the topology of the multiplex network throughout the algorithm. First, the motif is used to mine the higher-order interaction information in each layer of the network topology, and a multiplex network aggregation model is constructed to aggregate the multiplex network into a single-layer composite network. In addition, the information propagation method is used to transform the node information in the single-layer composite network into a vector. Furthermore, we propose a node importance evaluation index and select the nodes with the greatest significance to solve the initial node selection sensitivity problem of the k-means algorithm. The comparative experimental results on real multiplex network datasets, e.g., SCHOLAT, show that the proposed CDMA algorithm can play an important role in addressing the issue of effective detection of the community structure of multiplex networks with higher accuracy.