Similarity Based Hot Spot News Clustering

Perhaps BigData is one of the hottest topics nowadays due to the development of Web 2.0. As more and more data are available, relevant information becomes increasingly more difficult to locate, especially if the information are available on different servers (or websites). In this paper, similarity based information clustering, which is based on HowNet, is proposed to cluster relevant information. We develop a software which combined with the database for automatically storing and calculating similarities among enormous Chinese news headlines using two different calculating methods. A series of experiments with varying parameters have been conducted using the data obtained from five popular websites to show the effectiveness of the proposed approach. The experimental results reveal that the approaches are able to find the most related news. Furthermore, we could classify the different news into proper categories, which leads to a considerable contribution to readers.