基于观点树的网络舆情危机预警方法

摘要

对初始情感种子集预处理,增加网络情感词汇、过滤部分特征词,采用改进逐点互信息一信息检索方法对特征词进行情感倾向性计算,实现对网络论坛网帖中网民观点极性和观点强度的挖掘,进一步实施网民观点属性挖掘；构建基于三粒度挖掘结果的观点树,建立基于观点树的网络舆情危机预警架构.通过实验验证,该方法能够对网络舆情危机作出预警判断.%This paper pretreated initial emotional seeds. Added in Internet emotional words. Filtered,some feature words. Used improved point-wise mutual information-information( PMI-1R) retrieval method to calculate the sentiment orientation of feature words. Mined the opinion polarity and attitude force of Internet opinions. Furthermore, mined the opinion attributes, built an opinion tree based on the three different granularities opinion mining results. Created the framework of Internet public opinions crisis early-warning based on opinion-tree. The experiment shows that this method can early-warning Internet public opinions crisis.