Research And Application On Microblog User Interest Recognition

With the development of mobile Internet technology and the popularity of mobile terminals, there have been many social websites and applications on the Internet. As a social application, microblog has attracted a large number of users, with its convenience of operation and rapid propagation. A user receiving hundreds of microblogs every day, which leads to the situation of information overload, increases the difficulty of the user's information and knowledge acquisition. On the other hand, more and more merchants treating microblog as a marketing platform, which makes the advertisements directed delivery become a problem with highly commercial value. Microblog user interest recognition can contribute to solve the problems discussed above. The main work of this paper is as follows:At first, we tried user interest recognition based on topic model. In this method, we see the collection of user's microblogs as a microblog document, then use the labeled LDA to inference the topic distribution over the microblog document, this topic distribution will be used as the user's interest topic distribution. But there would be a serious deviation in the labeled LDA topic assignment to the interest related word, if many noise words appear in the context.Next, we tried user interest recognition based on microblog classification. We classified the user's microblogs, to reduce the affection of the noise words, then recognized the user's interest by the classification results. We tried two classifiers, the first one is linear SVM with bigram as its feature, and the second one is a classifier based on convolutional neural network. The experimental result showed this method achieved a better performance. But neither of the classifiers could deal with the microblogs containing many noise words well.Then we proposed a topic augmented convolutional neural network approach to recognize user interest. The proposed approach first obtained the category distribution of users' microblogs. It then recognized users' interest through the maximum likelihood estimation over the category distribution of users' microblogs. The experimental result shows the proposed topic augmented convolutional neural network approach achieves a significantly improvement on the microblog classification and user interest recognition.Finally, we introduce an application of user interest model, a news recommendation module of a chat robot called Ben Ben.