A Topic Detection Method Based on KeyGraph and Community Partition

Shanliang Yang, Qi Sun, Huyong Zhou, Zhengjie Gong, Yangzhi Zhou, Junhong Huang

Abstract: More and more media stream data is created on the Internet every day. It's more difficult for persons to obtain valuable information due to information overload. Topic detection is the method that extracts valuable hot topics from media stream data. It is the tool to help to solve the problem of overload information. The topic positive accuracy of cluster method is very low. In this paper, we proposed one topic detection method based on KeyGraph to improve the positive accuracy, and took experiments compared with baseline method on corpus marked by graduate students. In the result, the positive accuracy of KeyGraph method reaches 88.48% with great improvement. The result verified the effectiveness of our proposed method.