基于演化聚类的社交媒体舆情分析方法综述

当前,社交媒体已经成为人们发布和获取信息的最主要渠道之一。由于大量用户的参与,社交媒体平台上已累计了海量的舆情数据,并且还在以指数级增加。如何发现和跟踪社交媒体中舆情变化是当前数据挖掘领域研究的热点问题之一。基于演化聚类的方法是分析社交媒体舆情的一类主要方法。主要从主题模型和矩阵/张量分解模型两方面来总结当前基于演化聚类的舆情分析方法,并且分析这两类方法的优缺点;最后对下一步研究进行了展望。

A survey of analyzing public opinion methods based on evolution clustering

Currently, the social media platforms have become one of the main channels for publishing and obtaining information for most people. Since a great quantity of users, the platforms have accumulated a large number of social media data and the data are increasing exponentially. Detecting and tracking the trend of public opinions in social media is a researching hotspot in data mining field. Evolutionary clustering is a type of analyzing public opinion methods. In this paper, we summarize the existing evolutionary clustering methods for analyzing public opinion from two aspects: topic model and matrix/tensor factorization based methods. And, the advantages and disadvantages of two types of methods are discussed. At last, we conclude this paper and give the prospect of next researches.