A Microblog Unbalanced Data Evolution Analysis Method Based on LDA Model

Abstract

As a social network platform, Microblog has a large amount of information that can be mined. Microblog differs from static texts, which is time-sensitive, and the popularity of themes is changing with time. In addition, the popularity of Microblog varies from theme to theme, and the number of Microblogs varies greatly. Therefore, it has obvious unbalance characteristics. The traditional LDA model can not mine the theme intensity evolution process, and the theme mixing phenomenon will appear in the theme of unbalanced data mining. In view of the above problems, this paper proposes a method to analyze the theme evolution law of unbalanced Microblog data by introducing time variable. The experiment shows that the problem of theme words mixing is effectively solved after the balance treatment. The introduction of time variables can mine Microblog popularity and theme intensity evolution process.