Product Review Mining Based On Opinion Words

With the fast development of E-commerce and the population of Web 2.0, more and more consumers go shopping online. Product reviews may have more affects on potential consumers' behavior. There are a lot of reviews mixed with much information in different product communication platforms; review opinion mining can provide effective help for manufacturers, retailers and consumers.We propose a framework of product review information extraction based on adjective opinion words. Firstly we present the object of review opinion mining, got six aspects of review information, establish common product feature set, and take Epinions as an example to study review information and digital camera feature set. Secondly, we crawl web review pages for review opinion mining based on review information form and web page template, propose extraction algorithm and take Epinions as an example to implement it.Thirdly,we extract information based on review opinion mining, web page information extraction algorithm is presented with web page template features, later we manually labele seed adjective opinion words set, and expand it to a set with words coverage reaching 90%,we also identify comparative sentence, word match and negation patterns, and then demonstrate opinion word driven feature extraction algorithm and several ways to summarize the reviews. Finally in order to demonstrate our methods of the framework, we conduct the experiment with the product reviews of digital camera from www.Epinions.com.After summarizing a lot of digital camera reviews and classifying them to be positive or negative; our methods can provide the number of emotion polarity for the products and features.? The innovative results of this research are some new algorithms to study opinion words in product review mining.They're expected to help evterprises and consumers more conveniently access costomers' feedback on the product or service in the field of E-commerce with wider applications.