Research Article
Data-Driven Influential Nodes Identification in Dynamic Social Networks
@INPROCEEDINGS{10.1007/978-3-030-92635-9_34, author={Ye Qian and Li Pan}, title={Data-Driven Influential Nodes Identification in Dynamic Social Networks}, proceedings={Collaborative Computing: Networking, Applications and Worksharing. 17th EAI International Conference, CollaborateCom 2021, Virtual Event, October 16-18, 2021, Proceedings, Part I}, proceedings_a={COLLABORATECOM}, year={2022}, month={1}, keywords={Social networks Influential nodes Data-driven}, doi={10.1007/978-3-030-92635-9_34} }- Ye Qian
Li Pan
Year: 2022
Data-Driven Influential Nodes Identification in Dynamic Social Networks
COLLABORATECOM
Springer
DOI: 10.1007/978-3-030-92635-9_34
Abstract
The identification of influential nodes in social networks has significant commercial and academic value in advertising, information management, and user behavior analysis. Previous work only studies the simple topology of the network without considering the dynamic propagation characteristics of the network, which does not fit the actual scene and hinders wide application. To solve the problem, We develop a data-driven model for the identification of influential nodes in dynamic social networks. Firstly, we introduce an influence evaluation metric BTRank based on user interaction behavior and topic relevance of the information. Combining BTRank, LH-index, and betweenness centrality, we construct a multi-scale comprehensive metric system. Secondly, in order to optimize the metric weights calculated by entropy weight method, we use simulation data to train a regression model and obtain the metric weights by Gradient Descent Algorithm. Thirdly, the weights obtained from training are used in weighted TOPSIS to sort the influence of nodes and identify influential nodes among them. Finally, We compare our model with existing models on four real-world networks. The experimental results have demonstrated significant improvement in both accuracy and effectiveness achieved by our proposed model.
