基于情感倾向书评自动分类.docVIP

基于情感倾向的书评自动分类 张朝阳 商品评论的自动分类在电子商务中有重要的应用价值。首先对图书评论语料库进行未了登录词识别。然后以《知网》情感词为种子词汇，提出了一种基于条件随机场的Bootstrapping情感词挖掘算法，并利用互信息将这些情感词分为褒义和贬义。根据句子中包含正面情感词和负面情感词的数目，同时考虑否定副词和转折连词的影响，将图书评论自动分为好评和差评。对从电商网站上随机爬取的2026条书评做自动分类，平均正确率达81.8%，平均召回率达81.1%，说明了算法的有效性。 情感挖掘　书评分类　条件随机场 AUTOMATIC CLASSIFICATION OF BOOK REVIEWS BASED ON EMOTIONAL TENDENCIES Zhang Chaoyang Automatic classification of product reviews has great application value in e-commerce. Firstly, unknown words were recognized from corpus. With HowNet emotional words as seed vocabulary, a bootstrapping emotional words mining algorithm based on Condi