Product Review Information Extraction Based on Adjective Opinion Words

Product review mining can provide effective help for manufacturers, retailers and potential customers. We propose a framework of product review information extraction based on adjective opinion words. Firstly, we extract the web page information with web page template features. Secondly, we manually label seed adjective opinion word set, and expanded it to a set with words coverage reaching 90%. We also identify comparative sentence, word match and negation patterns. Finally, we summarize the product reviews from the various perspectives. In order to demonstrate our methods of the framework, we conduct the experiment with the product reviews of digital camera from www. Epinions.com. The experiment shows that the methods can provide the number of emotion polarity for the products and features.