摘要
网络舆情是近年来研究的热点。现有的舆情监测系统大多采用基于统计和特征关键词的方法，这种方法没有考虑到文字背后的语义关联，可能在两个文档共同出现的单词很少甚至没有，但两个文档是相似的。而主题模型正是近年来大量使用的语义挖掘的方法，其中LDA主题模型使用最多。结合了基于统计和LDA主题模型两种方法，应用于网络舆情监控。在基于统计的基础上进一步进行语义的挖掘，能够对舆情信息更加准确的分析。
Network public opinion is a research hotspot in recent years. Most of the existing public opinion monitoring system used the count of word or key-word, in this way, lost the semantics of the document. Topic model is a good semantics mining model, and is very popular in recent years. LDA is most commonly used. Combining the statistics and LDA topic model, designed a pub- lic opinion monitoring system. Based on counting of key-word, mining the semantics of documents, can exactly analyses the pub- lic opinion.
作者
钟荣飞
ZHONG Rong-fei （University of South China, Hengyang 421001, China）
出处
《电脑知识与技术》
2015年第3期90-93,共4页
Computer Knowledge and Technology
关键词
网络舆情
舆情监控
主题模型
LDA
语义挖掘
network public opinion
opinion monitor
topic model
LDA
semantics mining
