面向复合维信息特征的微博舆情事件感知方法

Composite Dimensional Information Characteristics Oriented towards Microblogging Public Opinion Event Detection Method

微博的短文本与半结构化特征,使得传统的基于热点词的舆情事件检测方法已不适用。对于微博的热点发现,需要充分利用微博特有的信息特征,构建适应于微博的热点感知方法。通过对微博的文本特征和社会化关系特征进行无监督聚类,提出一种基于LDA主题模型,面向复合维信息特征的微博舆情事件感知方法。实验表明,该方法在话题挖掘以及话题热点计算上有良好的效果。

The short nature of microblogging and its semi-structured characteristics, making traditional public opinion event detection methods based on the hot words is no longer applicable. Making full use of specific characteristics of microblogging and constructing a method that adapted to the characteristics is needed to discovery the hot topic of microblogging. Based on the theme model of LDA, this paper puts forward a kind of composite dimensional information characteristics oriented towards microblogging public opinion event detection method by text features and characteristics of social relations of microblogging for unsupervised clustering. Experiments show that the method has a good effect on the topic mining and hot topic of calculation.