摘要
网络舆情参与者对舆情事件情感指向的清晰定位,可以为舆情事件主体或监管部门准确掌握民众关注焦点、制定有效的网络舆情引导措施提供科学的参考。使用PLSA模型提取网络舆情话题及其特征词,结合TF-IDF修正话题特征词;基于已经提取到的话题文本特征词,构建对应的情感词表,并应用How Net相似度算法计算每个情感词对应的正负情感倾向值;综合考虑特征词对应的所有情感词,计算得出特征词的情感值,对舆情参与者的情感指向及变化进行准确定位。
The research on the evolution of netizens sentiments provided theoretical basis for relevant organizations to understand the neti- zens sentiments of the topics they were focusing on and to make effective opinion guidance of the network public topics. This paper used the PLSA model to extract the topics and the text-words matrixes after segmenting the texts. Combining with the TF-IDF model and the PLSA model, the paper accurately identified the sub topics and the feature words. Based on the text-words matrixes which had been ex- tracted before, this paper constructed the sentiment vocabularies. The application of similarity algorithm HowNet helped us to calculate the sentiment value of each word from the semantic angle, being positive or negative. Taking all of the feature words into consideration, this paper calculated the netizens sentiments of each feature word in the network public topics.
出处
《情报杂志》
CSSCI
北大核心
2015年第11期117-122,144,共7页
Journal of Intelligence
基金
国家自然科学基金"基于情景演化的数字化应急预案动态生成机制研究"(编号:7117111)
江苏省社会科学基金"大数据时代基于话题演化视角的网络舆情监控与应对路径研究"(编号:15TQB004)
关键词
网络舆情
话题
特征词
情感分析
network public opinion topic feature words feeling orientation
