Automating Opinion Analysis in Film Reviews: The Case of Statistic Versus Linguistic Approach

Websites dedicated to collecting and disseminating opinions about goods, services, and ideas,attract a diversity of opinions comprising attitudes and emotions. www.flixster.com is an example of a participative web site, where enthusiastic reviewers share their feelings/views on movies– usually expressing polar opinions. The participative web-sites usually contain substantial amount of data which is continually been updated.The contents of such websites is regarded as a key source of information by academic and commercial researchers keen to gauge this sample of public opinion. The key challenge is to automatically extract the reviewers opinion. Our goal is to use the reviews for building a model which can then be used to predict the user’s verdict on a movie. We explore two different methods for extracting opinion. The first, machine learning method that uses a naive Bayesian classifier. The second method builds upon existing NLP techniques to process opinions and build dictionaries: those dictionaries are then used to determine the polarity of a comment comprising a review. We compare and contrast the relative merits of the two methods with special reference to movie review data bases.