DOI: 10.5176/2251-2217_SEA36
Authors: J. Isabella and R. M. Suresh
Abstract:
The success of an organization not only depends on the quality of product and service they provide but also extensively depend on the digital medium used in the organization to interact with its customers. Websites have been successfully deployed to get new customers and retain existing customers. Websites cannot cater to the needs of all its users, however the website can be continuously upgraded with changing requirements of the customer based on feedback. This can be automated by collecting feedback automatically from different sites and find the opinion of the user which in turn can be used to update the website. Internet has facilitated access to a variety of sources of written text and it also made it easier for people to express their opinions. Opinions on almost any subject by means of specialized product review websites, discussion forums and blogs are expressed online. Opinion Mining deals with automated methods for detecting and extracting opinions from textual data. In this paper it is proposed to extract the feature set from movie opinions and compute the inverse document frequency and reduce the feature set using the proposed correlation based feature reduction. The proposed preprocessing method efficacy is tested using Naive Bayes and K Nearest Neighbour classifier.
Keywords: Opinion mining, IMDb, Inverse document frequency (IDF), Naïve Bayes, K Nearest Neighbor.
Updating...