Topic detection based on keyword

Topic Detection is a sub-task of Topic Detection and Tracking, its main task is to find and organize topics that system didn´t know. By analyzing hundreds of website news reports, we find that usually there exist some keywords in text, and early study didn´t pay enough attention to this, we propose a topic detection algorithm according this. The algorithm is based on K-means clustering algorithm, choose keywords and enhance the weight of keywords. Experiment proves it can improve the efficiency of system.