Abstract
Product reviews on the web sites help not only consumers to purchase products but also developers to analyze consumers’ needs. Because huge amount of the reviews are presented on the various sites, however, it is a hard task for them to read and to find only the reviews which match their viewpoint that they focus on. Though, to overcome this issue, many researchers in the field of the natural language processing tried to find review sites, classification of reviews according to their viewpoints does not have been succeeded because the corpus for classification must be needed and building it takes a lot of cost. In this paper, we propose a method to build the corpus for each type of products automatically and also propose a method for automatic classification of method of the review. In our method of classification, we focused on the property of review by extending the Tf-Idf. As the classification results contained many errors of classifications in the similar viewpoints, we built the improved method. In this method, we divided the classification process into two-stage. As the result, we could classify reviews by over 80 point.
Chapter PDF
Similar content being viewed by others
References
Kaji, N., Kitsuregawa, Y.: The automatic collection of evaluation sentences from an HTML document galaxy. Natural Language Processing Association (2004)
Suzuki, M.: The automatic classification of the text with difference of the word frequency between categories. Journal of Japan Industrial Management Association (2008)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Tachizawa, Y., Oka, M., Mori, H. (2013). An Automatic Classification of Product Review into Given Viewpoints. In: Yamamoto, S. (eds) Human Interface and the Management of Information. Information and Interaction for Learning, Culture, Collaboration and Business,. HIMI 2013. Lecture Notes in Computer Science, vol 8018. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-39226-9_65
Download citation
DOI: https://doi.org/10.1007/978-3-642-39226-9_65
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-39225-2
Online ISBN: 978-3-642-39226-9
eBook Packages: Computer ScienceComputer Science (R0)