Sentiment classification on Chinese reviews based on ambiguous sentiment confined library

Sentiment bag of words plays an important role in sentiment analysis. In the paper we propose a method to construct an AMBIGUOUS SENTIMENT CONFINED LIBRARY without hard work. Followed by preprocessing and extracting features based on our AMBIGUOUS SENTIMENT CONFINED LIBRARY Maximum Entropy classifier is trained by our selected corpus and used to classify Chinese review sentiment. Comparing with extracting features based on National Taiwan University Sentiment Dictionary experiments show that our method can evidently improve the performance of Chinese review sentiment classification.