15 An Unsupervised Approach to Mine Customer Opinion on Products or Services and to Rank the Customer Reviews Based on Its Features

Internet is a rich source of information, ranging from latest news to latest research, from product information to product review. We are using information like customer reviews of a product, present on internet, to plot an Opinion versus Feature graph. This will help the customer to take decision, whether to purchase a product or not and also for the manufacturer to known about the different Features the consumers are expecting. This is much needed, because manufacturer may have list of Features, but they may not have noticed some of the Features, that may have expressed by the customers. It is domain independent i.e. it can be used for any product ranging from car to cell phone, from availing a healthcare service of a hospital to enquiring about the profile of a company, before going to give interview in that company. During this process we are trying to summarize all the customer reviews of a product or any services on Internet, which is a non-redundant data extract from the original reviews. It is clearly different from traditional text Summarization. This summary is structured (but shorter) rather than another free text document as produced by most text summarization systems. We are only interested in Features of the product that customer has opinion on and in the sentiment of the customer opinion i.e. is it positive or negative. Normally we use to download reviews from different web site on a product manually. This process involves going to multiple search results before we land up with what we want, hence it is time consuming and also we will not get satisfied because of not willing to read all the reviews. We are here trying to create an application for automating this process by producing an abstract output which saves a lot of time, clears users confusion and thus being useful.