Study on key technology of topic tracking based on VSM

摘要

Text classification is the key technology for topic tracking, and vector space model (VSM) is one of the most simple and effective models for topics representation. On the basis of information gain algorithm and chi square χ2 in VSM, we have studied how feature selection algorithm and feature dimension in VSM affect topic tracking. And then we get the variation law that they affect topic tracking, and add up their optimal values in topic tracking. Finally, TDT evaluation method proves that their optimal values can make topic tracking gain very good tracking performance. In addition, we also prove in the experiment that chi square χ2 in VSM has better performance for topic tracking than information gain algorithm.