As a new direction of research on natural language processing,Topic Tracking aims at monitoring the stream of news stories to find additional stories on a topic that was identified using several sample stories.The research on topic tracking in China is just starting.In the dissertation,studies on topic tracking in Chinese news stories are carried out.We propose several algorithms for topic tracking,and adopt unigram model for Chinese topic tracking .We achieve two results as follows:(1) we try to apply several strategies to combine SVM and KNN to deal with topic tracking.Experiments show that methods with cascaded classifiers,which are trained on selected samples by KNN strategy,are more effective than others.(2) we propose unigram model for Chinese topic tracking and analyze some factors that effect topic tracking performance. Experiments show it has better tracker achieves.Next step ,we should do several works to improve the tracking system.For example ,combine information filter and topic tracking to apply in the challenge application. |