A Study on Feature Subsumption for Sentiment Classification in Social Networks using Natural Language Processing

In the past, whenever a customer wants to buy some product he used to consult his family members or friends. But this thing has been changed over the last few years where the users are generally finding out the reviews from the internet before purchasing the products. It is easy to process a review if the opinions are less, but for few popular products the reviews can be more that sometimes they will be in hundreds or thousands. It is a quite time taking process for the customers to go through all these reviews. So a system that could automatically summarize the opinions could be useful to the customers. This paper studies existing methods for sentiment classification and proposes new method Sentiment Classification for Dynamic Data Features (SCDDF) that not only considers many sites for sentiment classification but also aggregates the opinions using Bayesian Networks and Natural Language Processing techniques. We consider the various products and their features and classify them. Bulk amount of dynamic data is considered rather than the static one. It takes as input a collection of comments from the social networks and outputs ranks to the comments in the social networks, for each product, and also classifies the comments posted. Thus the user can evaluate the product and its features.