Leveraging Graph Embedding for Opinion Leader Detection in Dynamic Social Networks

Yunming Hui, Mel Chekol, Shihan Wang

Abstract: Detecting opinion leaders from dynamic social networks is an important and complex problem. The few methods in this field are poor in generalisation and cannot fully consider various dynamic features. In this paper, we propose a novel and generic method based on dynamic graph embedding and clustering. Inspired by the existing knowledge about dynamic opinion leader detection, the proposed method can exploit both the topological and temporal information of dynamic social networks comprehensively. It is also generalisable, as shown experimentally on three different dynamic social network datasets. The experimental results show that the proposed method runs faster than competitors.