Social Data Mining and User Preference Modeling Based on Machine Learning Algorithm

Abstract

In order to study the relationship between what people post on social networks and their preference for current affairs, we proposed a user preference evaluation model. We used text mining technology to pre-process the Weibo data, and made comparison of nine machine learning algorithms to obtain an optimal model of current affairs text classification. Furthermore, the classification process is extended to the users' scientific research ability and psychological state, three user preference evaluation models are obtained. The results show that the Linear Support Classifier is the optimal classification algorithm.