Bursty feature based topic detction and summarization

Thousands of news is available on the Web every day, it is almost impossible for people to read all of them. Because people are usually interested in “what´s new”, or “what´s hot”, it is quite necessary to find out these hot topics. In this paper, we propose a bursty feature based topic detection and automatic summarization method, which can help people have a gist of what´s happening daily. It first identifies bursty features in the news stream; and then these features are grouped into topics; finally, a centroid based summarization method is used to generate summary. Through the proposed method, bursty topic can be detected quickly, and the generated summary can help people get the general idea of the topic effectively.