With the advent of numerous video sharing and hosting sites, video content is continuously generated in the form of several TV shows, documentaries, movies, and short films. There exists certain websites such as IMDB and Rotten Tomatoes for movies and TV shows. Nielsen also provides ratings for TV shows. Although there exists several avenues such as these, there is also a grave necessity for the content creators to understand the overall unbiased sentiment for the content.
The goal of our project, Social Rating, is to build a comprehensive search engine to understand the sentiment for the content on social media. We gather tweets from Twitter and comments from YouTube to calculate the overall sentiment of a given video title. In the report attached, we discuss in detail the architecture of our system. We also discuss the implementation details. We conclude the report by presenting the future work, use cases, and results gathered from Nielsen and IMDB for comparison. This project was done as a final project for the course Cloud Computing and Big Data.
I was the incharge of data pipeline development. I contributed substantially to designing and refining the architecture of the project. I also developed modules for Topic Modelling (using mallet), Sentiment Analysis (using Stanford NLP toolkit), searching Youtube and Twitter, and also deployed the modules developed by me as a web service (RESTful web service) on the Amazon Elastic Beanstalk environment. The language I used to develop these modules is Java
More details of the project can be found in the links present below.
Slides -> http://www.slideshare.net/grohitbharadwaj/socialrating
Demo Video -> https://www.youtube.com/watch?v=ohV2FX2ZHiQ
Documentation -> https://drive.google.com/open?id=0B3SyjrvpH8YDTGhzVk13WFMxRzA