舆情事件中评论对象的情感分析及其关系网络研究

Sentiment Analysis of Objects in Public Opinion Events and Their Relation Network Research

【目的/意义】从舆论对象的情感变化和关系变化展现舆情走向,为舆情监测和分析提供新的研究方法和研究角度。【方法/过程】文章提出舆论对象分析模型,利用依存句法分析,识别和抽取舆论对象-情感词对,进行情感分析,并对舆论对象的关系网络进行研究。【结果/结论】实验结果表明,模型能有效识别主要舆论对象及其情感词,直观地展现网民对舆论对象随时间演化的情感表达和关系认知。通过舆论对象情感变化与舆情事件发展的拟合,可为舆情监测、分析等相关研究提供新的研究视角。

[Purpose/significance] In order to describe and analyze the public event from the angle of public opinion objects. [Method/process] This paper puts forward a sentiment and relation network analysis model of public opinion objects based on the dependency parsing. [Result/conclusion] The results show that the model can effectively identify the objects of the public opinion event and their sentiment words, and display the trend of public opinion from the change of the sentiments and the change of the relations between those objects, and also, the results reveal some rules about how event goes.