Abstract: ［Purpose/Significance］The prediction of microblog entries' influence in the context of terrorist events and the revelation of their evolution patterns can help the counterterrorism departments foresee potential problems and hazards in time and take effective measures to respond quickly. ［Design/Methodology］In this paper, the features of users, time and contents of microblog entries in terrorist incidents were extracted and a microblog influence prediction model was developed, trained and evaluated based on the logistic regression model. ［Findings/Conclusion］The accuracy rate of the proposed model reaches 85.8%, which means it can effectively predict the influence of microblog entries. This paper analyzes the characteristics of highinfluence microblogs about terrorist incidents. A quantitative method of microblog topics’ influence according to h-index is proposed and the topics’ evolutionary patterns are further explored. ［Originality/Value］The findings can help identify microblog entries of high influence about terrorist events, assess the scale of microblog information dissemination, understand the contents, intensity and evolution of public attention, and assist counter terrorism departments in public opinion management.

Key words: Terrorist events, Microblog, Influence, Prediction, Evolution, Topic identification, Sentiment analysis, Hindex