Topic Detection of Chinese News Based on Word Entropy
Bo Zhu, Min Hou, Yuyin He · 2016
We propose a method of automatic news topic detection in large-scale data.First, topic words are detected based on their word entropy.Then, the topic word co-occurrence net is constructed via the semantic relationships of topic words represented by their orders in which they appear within the original text.Finally, implied communities are detected in the topic word co-occurrence net through modularity measures.Each implied community is regarded as a news topic.Experimental results show that this method can be used to effectively identify the key topic of each news report, with the presence of topic content in human-readable form.