Movie Review System using Sentiment Analysis and Social Networking Platforms (SNPs)

Abstract

Today online social networking platforms SNPs have become an integral part of or life where we share a lot of information of all the things, we do in life from shopping to movie watching. With ever growing use of SNPs recommendation systems have emerged as a hot trend for applications in e-commerce and digital media. These recommendation systems are useful as well as misguiding. Today digital media use has increased tremendously with increase in internet speeds. But users do not get proper review of a movie and a user is lured to watch a substandard movie which he does not intended to do, thus costing a user time and money. So, there is a need of developing a movie review which will give correct reviews of a digital content like movies so he can only movies which he intends to do. So, we are studying various techniques authored by various authors and create a good movie review system of our own. The first technique we studied and intends to use is movie recommendation system using tweets. The second study is movie recommendation using similarity measures. The third study does find a public shamming using SNP. These techniques are useful and we propose to use some part of each in our new movie review framework by improving the techniques drawbacks. The new framework will be a combination of data from more than one SNP and using natural language processing and machine learning on the data. We are going to use two machine learning algorithms SVM and Naïve Bayes for this purpose. For natural language processing of SNP data, we are going to use OPEN-NLP. We intend to use SNPs such as Twitter and any other movie database like IMDB etc. for data on the movie. The movie will be classified in three classes bad, good and excellent. The results from each algorithm SVM and Naïve bayes will be analyzed for each SNP and try to give user a more accurate movie review by combining all the reviews together and classes accuracy and show overall prediction results with a rating. To get more accurate results for each movie we are going to create a dataset for each movie for demonstration and will not depend on a single combined dataset as keywords for each movie may be different. We are going to combine datasets for each movie from multiple SNPs.