Social event decomposition for constructing knowledge graph

Abstract

Given the large amount of data collected from social media, it is very difficult for users to identify social events and understand their societies. In this paper, we propose a novel method for i)decomposing and discovering social events and ii)representing social events and their relationships as a knowledge graph. In particular, the proposed method is based on Independent Component Analysis (ICA)and the SocioScope Knowledge Graph (SKG)model. To demonstrate the actual performance, the proposed method has been evaluated with the support of the SocioScope framework (Nguyen and Jung, 2018). Then, it was verified that the system can efficiently provide people with a high understandability and traceability of social events. © 2019 Elsevier B.V.