Abstract: An approach of sentiment classification for online comments based on intuitionistic fuzzy reasoning is presented on the basis of the analysis of characteristics of sentiment classification. The approach employs membership function, non-membership function and hesitant function to depict uncertainties of features, quantitatively, by sample training, as well as sentiment expressions influenced by adverbs of degree, conjunctions and negative words are considered. Then the semantic orientation of a text is synthesized on the level of phrases, sentences and texts in sequence by means of aggregations of intuitionistic fuzzy information of features. The validity and high performance of the presented approach is testified when comparable experiments are done using public corpus.

Key words: Chinese information process, text categorization, sentiment classification, intuitionistic fuzzy set

摘要： 网页在线评论的情感分类关系到个人决策、企业管理甚至社会安全。提出了一种基于直觉模糊推理的情感分类方法，通过样本库的学习将特征在分类时的不确定性分别用隶属度、非隶属度、犹豫度定量地描述，同时定量地考虑程度副词、转折词、否定词对表达情感的作用，然后通过对特征的直觉模糊信息的集结，按词组、句子、文本三个级别依次合其情感倾向，得到文本的情感倾向。在对公开语料库的比较实验中证明了该方法的正确性和分类性能。

关键词: 中文信息处理, 文本分类, 情感分类, 直觉模糊集