In order to accurately excavate the micro-blog (Weibo) topic information and emotional information, we put forward Weibo Sentiment Online-LDA model on the basis of LDA. The model prejudges the emotional tendencies of the words in the text as a priori information of emotions and expands LDA model according to the emotional layer to get the topic information and the different emotional information of the topic. It also considers the influence of text information on the current time, dynamically adjusts the genetic coefficient of the topic, and ensures that the hot topic features are inherited to the next moment. The experiments show that WSO-LDA model mining matches the topic information and emotion information, and the model confusion degree is superior to other topic models.
Ma, Jing; Yao, Zhaoxu; and Sun, Mingzhu, "WSO-LDA: An Online "Sentiment + Topic" Weibo Topic Mining Algorithm" (2017). PACIS 2017 Proceedings. 223.