Harnessing consumer reviews for marketing intelligence: a domain-adapted sentiment classification approach

With the success and proliferation of Web 2.0 applications, consumers can use the Internet for shopping, comparing products, and publishing product reviews on various social media sites. Such consumer reviews are valuable assets in applications supporting marketing intelligence. However, the rapidly increasing number of consumer reviews makes it difficult for businesses or consumers to obtain a comprehensive view of consumer opinions pertaining to a product of interest when manual analysis techniques are used. Thus, developing data analysis tools that can automatically analyze consumer reviews to summarize consumer sentiments is both desirable and essential. Accordingly, this study was focused on the sentiment classification of consumer reviews. To address the domain-dependency problem typically encountered in sentiment classification and other sentiment analysis applications, we propose a domain-adapted sentiment-classification (DASC) technique for inducing a domain-independent base classifier and using a co-training mechanism to adapt the base classifier to a specific application domain of interest. Our empirical evaluation results show that the performance of the proposed DA-SC technique is superior or comparable to similar techniques for classifying consumer reviews into appropriate sentiment categories.