Supervised Learning Approaches for Rating Customer Reviews

Social media (blogs, reviews, and forums) have become highly popular in recent years, and people are expressing their views, thoughts on a mobile or a movie or a book through reviews. Reviews have a great influence on people in decision making, which has led researchers and market analysts to analyze the opinions or sentiments of users in reviews and statistically model their preferences. Sometimes reviews are also rated in terms of satisfaction score on any product or movie by the customer. These ratings usually vary on a scale of one to five (stars) or from very bad to excellent. In our work, we address the problem of attributing a numerical score (one to five stars) to a review. We view it as a multi-label classification (supervised learning) problem and present two approaches, using naive Bayes (NB) and logistic regression (LR). We focus more feature representations of reviews widely used; problems associated with them and present solutions.