Research on Improvement of Text Processing and Clustering Algorithms in Public Opinion Early Warning System

Abstract

In order to provide the necessary data for Public opinion monitoring and trend warning, this paper did some researches on text processing and clustering algorithms based on hot topics of the Weibo. Data that get from Weibo were classification data which contain two properties. To adapt this feature and meet the requirement of public opinion trends warning, hamming distance was used to do text similarity computing. By improving the traditional K-means algorithm, a new K-mode algorithm which is used to text clustering on hot topics was achieved. Simulation and results analysis indicated the text processing method was accurate and suitable to the microblog public opinion early warning.