微博信息流行度预测研究综述

Survey on Predicting Popularity of Information in Microblogs

微博的出现改变了传统的信息传播方式。文章针对微博流行度预测的一系列问题展开综述。在对微博流行度预测问题进行定义的基础上,从微博内容、用户信息和网络结构3方面对影响微博流行度的主要因素进行介绍,对基于时间序列、基于流行病模型、基于分类模型和基于回归模型的4种微博信息预测主流方法进行比较,指出流行度预测未来的研究方向主要集中在时间粒度的划分、最小观测时间的确定、微博流行度的动态预测等方面。该研究对信息检索、舆情研判、企业营销等具有重要意义。

Microblogs have changed the traditional way of information dissemination. This paper presents a survey of predicting in-formation popularity in microblogs. Based on the definition of the popularity prediction problem of microblogs, the main factors which affect the popularity of microblogs are introduced from the aspects of content, user information and network structure. Four methods based on time series, epidemic model, classification model and regression model are compared. The future direction of the popularity prediction is pointed, which is focused on the time granularity, minimum observation time and dynamic prediction of popularity. This study is important to information retrieval, public opinion monitoring, business marketing and so on.