Resolving inconsistent ratings and reviews on commercial webs based on support vector machines

Nowadays, there are more and more ratings and reviews for products and services given by customers on the commercial webs. The ratings and reviews often play critical roles as many consumers read them before making purchase decisions. Due to various reasons, the same customer sometimes gave inconsistent rating and review on commercial website. The existing studies tend to focus on the consumers' feedback on either rating or review separately, but ignore the inconsistency between them. Occasionally, consumers cannot fully express their real complex opinions simply by ratings. Consequently, they are inclined to write their true feelings in reviews. This paper harvests 852,071 ratings and corresponding reviews from the Taobao website, and resolves inconsistant ones based on text sentiment analysis and support vector machines. Our method improves the existing online customers' feedback system by intelligently resolving inconsistent, sometimes conflicting, ratings or reviews from customers, and also provides decision supports for both consumers and shopkeepers.