System For Interpretation Of Reviewer Feedback On Social Web

Abstract

The increasing popularity of social media web sites such as Amazon, Yelp and others has influenced our online decision making. Before making selection decisions on movies and restaurants, we investigate its reviewer feedback. Social media web sites provide reviewer feedback in the form of ratings, tags, and user reviews. However, overwhelming feedback details will leave the user in a quandary as to decide whether the item is desirable or not. Potential buyers either make a snap judgment based on the aggregate ratings/tags or spend a lot of time in reading reviews.In this thesis, we build a system that can analyse the reviewer feedback in the form of ratings or tags and generate meaningful interpretations. One of major component is rating interpretation that generates meaningful interpretation of the reviewer ratings associated with the item of interest. For example, given the movie "Titanic", our system returns results such as, "Young female from California like this movie" instead of average rating 7.6 from all reviewers. Furthermore, end users will be allowed to systematically explore, visualize, and observe rating patterns. Additionally, our system can also explore the social tagging behavior on the input items. For example, our system can identify movies where similar users have assigned similar tags on diverse items. The tagging behavior of different sub population is compared using tag clouds. We use IMDb movie data set to demonstrate our experiments.