M2P2: Movie’s Trailer Reviews Based Movie Popularity Prediction System

Abstract

In today’s world, knowing which movie will be popular among the masses, motivates movie’s crew, before movie comes on the PVR curtain. Trailers and teasers usually come out, before releasing any movie. Moreover, the people are giving comments or feedback about the movie based on the trailer or teaser. These reviews or comments may be used to predict the popularity of the movie in the near future. Therefore, this work focuses for predicting the popularity or success of the movie based on analysis of the trailer comments. This work exposes the importance of the TextBlob and Enchant library. In addition to this, dictionary-based recommendation system has been developed in order to enhance the accuracy of the proposed popularity prediction system. For experimental process, the reviews of the people on trailers before release of the movie have been used to extract the features, and then the same features are mapped on corresponding BookMyShow popularity index after the movie has been released. A percentage of accuracy of the model shows that the proposed model can be used as an accurate movie recommendation system.