Community detection algorithm for social network based on node intimacy and graph embedding model
Di Huang,Jinbao Song,Yu He
DOI: https://doi.org/10.1016/j.engappai.2024.107947
IF: 8
2024-02-03
Engineering Applications of Artificial Intelligence
Abstract:Social network analysis has become an important research area in recent years. Community detection, as a fundamental task in social network analysis, aims to discover the hidden community structure. However, most real-world social networks are unweighted and undirected without edge attributes. Existing community detection methods cannot fully utilize the network topology information. In this paper, we propose a new community detection algorithm based on graph embedding and node intimacy. Firstly, we design an algorithm to calculate the edge weights from the unweighted and undirected social network structure. Then, we apply the Node2vec model to learn the low-dimensional representations of nodes by preserving both weight matrix and global network structures. After that, a node intimacy algorithm is adopted to calculate the intimacy between each node pair based on their embeddings. Finally, the Fuzzy Clustering Means algorithm is used to group nodes into communities by considering the learned node intimacy. Extensive experiments are conducted on both real-world and artificial benchmark networks. The results demonstrate that our proposed algorithm can capture the community structure more accurately by making better use of the network topology, compared with other state-of-the-art methods. In summary, the algorithm integrates graph embedding and node intimacy for community detection, which provides an effective approach for analyzing unweighted social networks.
automation & control systems,computer science, artificial intelligence,engineering, electrical & electronic, multidisciplinary