Discovering and learning sensational episodes of news events
Xiang Ao, Ping Luo, Chengkai Li, Fuzhen Zhuang, Qing He, Zhongzhi Shi · 2014
This paper studies the problem of discovering and learning sensational 2-episodes, i.e., pairs of co-occurring news events. To find all frequent episodes, we propose an efficient algorithm, MEELO, which significantly outperforms conventional methods. Given many frequent episodes, we rank them by their sensational effect. Instead of limiting ourselves to any individual subjective measure of sensational effect, we propose a learning-to-rank approach that exploits multiple features to capture the sensational effect of an episode from various aspects. An experimental study on real data verified our approach's efficiency and effectiveness.