Product retrieval based on semantic similarity of consumer reviews to natural language query. (19th July 2010)

• Record Type: Journal Article
• Authors: Sugiki, Kenji; Matsubara, Shigeki
• Is Part Of: International journal of knowledge and web intelligence. Volume 1:Number 3/4(2010)
• Page Start: 209
• Page End: 226
• Publication Date: 2010-07-19
• Subjects: natural language processing -- NLP -- information retrieval -- thesaurus -- sentiment analysis -- opinion mining -- product retrieval -- semantic similarity -- consumer reviews -- natural language queries -- product reviews -- consumer opinions; Information retrieval -- Periodicals; Web site development -- Periodicals; Internet programming -- Periodicals; 004.67805
• Languages: English
• ISSNs: 1755-8255

Abstract

In this paper, we propose a product retrieval method using product reviews. The method calculates the score of a product based on the similarity between a user's request extracted from a natural language query and consumers' opinions extracted from reviews of the product. We also propose the method to generate the thesaurus automatically from consumers' reviews. The thesaurus expresses semantic similarities between opinion expressions. We implemented an accommodation retrieval system and constructed the opinion thesaurus automatically from 652, 875 reviews. The experimental results show the effectiveness of our method.