Chinese Article Classification Oriented to Social Network Based on Convolutional Neural Networks

Xiang Zhu, Jiuming Huang, Zhongcheng Zhou, Yi Han

Abstract: With the development of social network media, text classification oriented to the social network has attracted increasing attention. The enormous amounts of user generated content in the social network could offer tremendous information in various topics. Article automatic classification in social network is a challenging problem. However, traditional methods for Chinese article classification are insufficient in the social network due to the high volume, topic variety and data sparseness. In this paper, we propose a novel framework for Chinese article automatic classification oriented to the social network. Sentence extraction techniques are utilized to get the summarization of an article. In addition, word vector model is leveraged to represent the extracted sentence. Finally, a convolutional neural network is built to predict the category of an article in the social network. A thorough evaluation is conducted on real data in the social network, verifying the effectiveness and efficiency of the algorithm.