网络舆情热点规模预测研究  

Research on the Prediction of the Scale of Online Hot Topics

作　　者：罗泰晔[1] Luo Taiye(Xinhua College of Sun Yat-sen University, Guangzhou 510520)

机构地区：[1]中山大学新华学院,广州510520

出　　处：《情报杂志》2016年第10期181-184,145,共5页Journal of Intelligence

基　　金：广东省高校教学质量与教学改革工程项目"独立学院电子商务专业应用型人才培养的综合改革"资助(编号:粤高教函[2013]113)

摘　　要：[目的/意义]提出一种能实时预测舆情热点规模的方法。[方法/过程]通过对历史数据的分析发现网络热点话题帖的回帖时间间隔成伽马分布,并以此为基础利用回归分析得出了实时预测舆情热点规模的公式。该公式能够利用过去一个小时的回帖时间间隔特征来预测下一个小时的回帖规模。[结果/结论]用历史数据证明了该方法具有良好的预测效果。[ Purpose/Significance ] The prediction of the scale of online public opinion is an important means for online public opinion su-pervision. The goal of this paper is to propose a method that can predict the scale of online hot topics in real time. [ Method/Process] Ananalysis on historical data revealed that the time interval of replies follows a Gamma distribution, based on this discovery and regression a-nalysis, a formula was formed to predict the development of online hot topics. The formula can use the time interval characteristics of re-plies in the previous hour to predict the reply scale in the next hour. [ Result/Conclusion] Historical data were used to prove the effective-ness of the method.

关 键 词：网络舆情 回帖分布 幂律分布 伽马分布 回归分析 

分 类 号：G206[文化科学—传播学]