Movie Review Classification Based on a Multiple Classifier

Movie Review Classification Based on a Multiple Classifier* Kimitaka Tsutsumi a, Kazutaka Shimada a and Tsutomu Endo a aDepartment of Artificial Intelligence, Kyushu Institute of Technology, Iizuka, Fukuoka 820-8502 Japan {k_tsutsumi, shimada, endo}@pluto.ai.kyutech.ac.jp Abstract. In this paper, we propose a method to classify movie review documents into positive or negative opinions. There are several approaches to classify documents. The previous studies, however, used only a single classifier for the classification task. We describe a multiple classifier for the review document classification task. The method consists of three classifiers based on SVMs, ME and score calculation. We apply two voting methods and SVMs to the integration process of single classifiers. The integrated methods improved the accuracy as compared with the three single classifiers. The experimental results show the effectiveness of our method. Keywords: Sentiment analysis, p/n classification, Integration, Movie reviews, WWW. 1. Introduction The World Wide Web contains a huge number of on-line documents that are easily accessible. Finding information relevant to user needs has become increasingly important. The most important information on the Web is usually contained in the text. We obtain a huge number of review documents that include user’s opinions for products. For example, buying products, users usually survey the product reviews. Movie reviews are likewise one of the most important information for users who go to a movie. More precise and effective methods for evaluating the products are useful for users. Many researchers have recently studied extraction of evaluative expressions and classification of opinions (Kobayashi et al., 2005; Osajima et al., 2005; Pang et al., 2002; Turney, 2002). Studies of opinion classification are generally classified into three groups: (1) classifying documents into the positive (p) or negative (n) opinions, (2) classifying sentences into the positive or negative opinions, and (3) classifying words into the positive or negative expressions. In this paper, we focus on the classification of movie review documents. Pang et al. (2002) have reported the effectiveness of applying machine learning techniques to the p/n classification. They compared three machine learning methods: Naive Bayes, Maximum Entropy and Support Vector Machines. In their experiment, SVMs produced the best performance. Osajima et al. (2005) have proposed a method for polarity classification of sentences in review documents. The method is based on a score calculation process of word polarity and outperformed SVMs in the sentence classification task. The previous studies, however, used only a single classifier for the classification task. We (Tsutsumi 2006 et al.) have proposed a method consisting of two classifiers: SVMs and the scoring method by Osajima et al. (2005). The method identified the class (p/n) of a document on the basis of the distances that were measured from the hyperplane of each classifier. It obtained * Copyright 2007 by Kimitaka Tsutsumi, Kazutaka Shimada and Tsutomu Endo. 481 the better accuracy as compared with the single classifiers. This method, however, contained a problem for the determination of the final output, namely positive or negative. We needed to normalize the classifier’s outputs manually because the scale of …

Content truncated to 500 words