EventPanorama: A Framework for Event Detection and Visualization from Online News
Chen Zhang, Hao Wang, Wei Wang, Cuixia Ma, Jingjing Li, Yibo Wang, Fanjiang Xu · 2016
Online News has become one of the most popular channels for consuming and understanding the real-world events. However, it is increasingly difficult for users to hold the full picture of massive events in a comprehensive perspective. In addition, how to motivate human to perceive the important rare events, which may trigger the subsequent emergency events, remains to be a challenge. To address these issues, we present EventPanorama, a framework to detect and visualize events. Firstly, we introduce a hybrid event detection method which combines topic modeling and Chance Discovery, and detects events more effectively by coupling multiple term-relations. Secondly, we propose a heterogeneous event-graph layout algorithm which takes the significance of events into consideration by leveraging latent co-occurrence relations to represent important rare events and thus enhance human cognition. An experiment demonstrates the superiority of EventPanorama by comparing with several benchmarks.