Hot News Mining Method Based on Intelligent Computing

摘要

Internet news has been an important source of public access to information, it is of great importance to find useful information behind hot news. Therefore, in this paper, we aim to mine hot news using intelligent computing. Firstly, we proposed a framework of the hot news mining system, in which the comment evolution, area evolution, strength evolution are obtained based on topic extraction matrix and content evolution matrix. Secondly, we propose a novel hot news mining method based on Latent Dirichlet Allocation (LDA), which is a three-level hierarchical Bayesian model. In the LDA model, each element of a collection is represented as a finite mixture for underlying topics. Finally, we construct two datasets using Sina Weibo and Twitter, and experimental results show the effectiveness of the proposed algorithm.