Online News Clustering System for Event Detection
In the information age, there are huge amount of news stories on the Internet. People wish to learn what is going on and how it is going every day. In this paper, an online news clustering system is proposed. We use two-stage clustering algorithm in this system. The first stage is micro-clustering for event detection, in which online news stories are clustered into micro-clusters. Then we perform an event tracking process, where those new microclusters are compared with previous generated microclusters, either merged into old ones or be regarded as a new event. At the same time, these microclusters will be classified into candidates and outliers according to the number of stories it contains. The second stage is a macro-clustering algorithm, which runs on the result of microclustering to combine all candidates to its related events. Based on this two-stage clustering approach, the system can provide an overview of one specific event and its related events. The system has been realized and online for extracting every days events from news stories.
Clustering Event Detection Evet Tracking
Shun Wang Fang Li
Deptof Computer Science & Engineering Shanghai Jiaotong University Shanghai, China
国际会议
重庆
英文
136-140
2011-01-21（万方平台首次上网日期，不代表论文的发表时间）