Prediction of the Dissemination of Health News on Microblogging Sites Based on Ample Feature Selection and Support Vector Machine

Abstract

As a social networking service, microblogging sites provide an open platform facilitating the sharing and discussion of valuable news. This paper identifies the influencing factors of the dissemination of health news posts on Weibo, the leading microblogging site in China. The effects of these factors were tested with 863 news posts, all about public health issues. The content features, author features and a social feature of each post were evaluated, and collected into a set of ample, diverse features that characterize widely disseminated posts. In addition, the support vector machine (SVM) was adopted to differentiate between widely disseminated posts and normal posts, and compared with several classification methods through an experiment on microblog posts of health news. The results show that the SVM greatly outperformed the contrastive methods in predicting the dissemination trends of such news. The research results inform crisis managers about the public reaction towards specific news on public health issues, shedding important new light on news dissemination.

A Hotspot Discovery Method Based on Improved FIHC Clustering Algorithm

“…However, the recent studies show that the main way to discover hot topics on micro-blog is to use clustering algorithm to cluster text content [19,20]. The traditional clustering algorithms mainly focus on large and traditional text, while the texts of microblog are very short, trendy, and network language.…”