Study on Topic Tracking System Based on KNN

摘要：

Text classification is the key technology for topic tracking, and vector space model (VSM) is one of the most simple and effective topics representation model. Feature selection algorithm in VSM is an important means of data preprocessing, and it can reduce vector space dimension and improve the generalization ability of the algorithm. Therefore, it is necessary for feature selection algorithms to be in-depth and extensive research. So we develop a topic tracking system to study how feature dimension and the value of K-neighbors affect topic tracking. Tben we get the variation law that they affect topic tracking, and add rip their optimal values in topic tracking. Finally, TDT evaluation methods prove tbat optimal topic tracking performance based on adjusting the value of Kneighbors for text increases by 7.246％ more than feature dimension.

关键词： knn information gain tdt evaluation topic tracking

作者: Shengdong Li Xueqiang Lv Dong Liu Shuicai Shi

作者单位: Chinese Information Processing Research Center,Beijing Information Science and Technology University Chinese Information Processing Research Center,Beijing Information Science and Technojogy University Web Data Management Lab. Wuhan University Wuhan China

会议类型: 国际会议

会议名称: 2010 2nd International Conference on Signal Processing System(2010年信号处理系统国际会议 ICSPS 2010)

会议地点: 大连

会议语种:英文

页码: 971-975

在线出版日期: 2010-07-05（万方平台首次上网日期，不代表论文的发表时间）