The research of hot-event's heat prediction in microblog platform

Abstract

With rapid development of microblog, more and more people are going to use microblog to transmit information and express their opinions. Therefore, predicting and monitoring the online hot-event tendency has become a significant support for our national security and development of society. Traditional prediction model tends to have a non-liner and unstable performance in time series. As a result, the accuracy of prediction cannot be guaranteed. To address that problem, this paper proposes an improved regression prediction model which combines the derivative and regression curve, and gets a consequence about heat tendency for hot-event. According to experiment result, the proposed model outperforms LS_SVM model and several other traditional models.