PDM: A Simple Approach for Identifying Influential Users in Social Networks

In the realm of social computing, identifying highly influential users holds significant theoretical and practical implications. Traditional methods frequently depend on centrality measures such as degree, closeness, betweenness, and K-shell within social networks. Although these approaches are straightforward, they often fall short due to their reliance on static topological structures. To address these issues, a method for rapidly constructing user features based on user historical data and social network topology was designed, and a classifier for high-influence users was developed. When aggregating the features of neighboring nodes using the social network topology, the proposed PDM framework performs rapid calculations based on the Laplacian matrix of the network structure, making it more efficient than using GCN networks. Our model demonstrates superior performance compared to existing methods across two real-world social network datasets. By effectively leveraging both historical data, this approach offers a more robust strategy for identifying high influential users.