Sentiment Analysis and Mining of Opinions

Abstract

Now days the way of expressing opinions on certain products that people purchase and the services that they receive in the various industries has been transformed considerably because of World Wide Web. Social Networking sites fascinate people to post feedbacks and reviews online on blogs, Internet forums, review portals and much more. These opinions play a very important role for customers and product manufacturers as they tend to give better knowledge of buying and selling by setting positive and negative comments on products and other information which can improve their decision making policies. Mining of such opinions have focused the researchers to pay a keen intention in developing such a system which can not only collect useful and relevant reviews online in a ranked manner and also produce an effective summary of such reviews collected on different products according to their respective domains. However, there is little evidence that researchers have approached this issue in opinion mining with the intent of developing such a system. Our work will focus on what opinion mining is the existing works on opinion mining, the challenges in the existing techniques and the workflow of mining opinions. Consequently, the aim of this chapter is to discuss the overall novel architecture of developing an opinion system that will address the remaining challenges and provide an overview of how to mine opinions. Existing research in sentiment analysis tend to focus on finding out how to classify the opinions and produce a collaborative summary in their respective domains, despite an increase in the field of opinion mining and its research, many challenges remain in designing a more comprehensive way of building a system to mine opinions. This chapter addresses the problem of how to classify sentiments and develop the opinion system by combining theories of supervised learning.