Mining the impact of social news on the emotions of users based on deep model

In this paper, Deep Belief Nets(DBN) model and Support Vector Machine(SVM) are used to mine the features hidden in social news, which can influence the emotions of men. Three feature selection methods for text modeling are used to build the input vectors of DBN, with the purpose of keeping the text information to the greatest extent. We take advantage of the deep features abstracted by DBN to build social news text classifier. At the same time, three optimal models are used as inputs of SVM to train and classify the social news. We get a conclusion that DBN not only reduces the dimension of original features, but also makes the abstracted features with more text information and shows better performance in determining the influence on people's emotions by social news.