Project Rank: An Internet Topic Evaluation Model Based on Latent Dirichlet Allocation

With the rapid development of Internet-related industries, the Internet news is containing more and more information as well as being paid more attention by human beings. In order to explore the intrinsic links among the news and extracted effective topics, we put forward ProjectRank model based on latent Dirichlet allocation. Large-scale Chinese text sets serves as initial input of this model. By combining LDA topic model with distance between probability distributions, we figure out scores of topics so as to quantify the result of topic evaluation. Moreover, this model proves to be useful to track trends of hot news in different time periods. This paper applies ProjectRank model to real Internet news data obtained from several websites and achieves good experiment results.