Document Level Sentiment Analysis Based On Naive Bayes Algorithm

With the rapid development of web2.0, people can obtain a lot amount of opinion-bearing content from internet, such as personal blogs and review site. With these information people can actively mining and understand what other people think via information technology.Sentiment analysis, which deals with the computational treatment of opinion, sentiment, and subjectivity in text, has earned a great deal of attention. There are also a lot of potential application of sentiment analysis, like question-answering systems that address opinions as opposed to facts and business intelligence systems that analyze user feedback. The research issues raised by such applications are often quite challenging compared to fact based analysis. This paper presents several sentiment analysis tasks to illustrate the new challenges and opportunities.An important task of sentiment analysis is classify sentiment-bearing review (movie review or book review) as positive or negative, also known as text sentiment polarity classification. This paper will present some algorithms and a comprehensive survey in sentiment polarity classification.we first summarize sentiment of a review, and then use machine learning algorithm to do the classification. In order to extract subjective sentence from a given review, previous research tent to use huge amount of labeled data to train a subjectivity classifier. But as known to all, manually annotate sentences as subjective or objective is a time-consuming and tough job. So we use an algorithm to automatically label sentences as subjective or objective from un-annotated corpus as training set. We also consider proximity between each sentence in a given review when we do the sentence-level subjectivity classification, which can easily be modeled by graph-cut-based model.Finally, we train and classify each review using their subjective part only.