Overview:
- For the scope of this project we will be trying to predict the freshness rating for a movie, based on the user reviews. In order to enhance our predictions, we will be analyzing the sentiment of Rotten Tomatoe Reviews.
Data Sources:
- A Subset of the Movie Lens Data: https://raw.github.com/cs109/cs109_data/master/movies.dat
- Contains approximately 9,400 movies (titles & IMDB ID)
- Adapted from: http://grouplens.org/datasets/movielens/
- Rotten Tomatoes API: http://developer.rottentomatoes.com/
- Reviews API: http://developer.rottentomatoes.com/docs/read/json/v10/Movie_Reviews
- Movie Alias API: http://developer.rottentomatoes.com/docs/read/json/v10/Movie_Alias
Additional Options:
- Get the abridged cast list
- http://developer.rottentomatoes.com/docs/read/json/v10/Movie_Info
- Preprocess the comments a bit more
- removing stopwords
- stemming (Porter or KSTEM)
- convert everything into the same case
- remove "junk"
- sentiment analysis: http://www.nltk.org/api/nltk.html
The Tomatometer: http://www.rottentomatoes.com/about/
Related Kaggle: https://www.kaggle.com/c/sentiment-analysis-on-movie-reviews
Movie Lens Dataset: http://grouplens.org/datasets/movielens/
Related Work:
- http://datapsych.weebly.com/blog/bayesian-tomatoes-using-machine-learning-and-text-analysis-to-predict-reviewer-sentiment
- http://nbviewer.ipython.org/github/xbsd/content/blob/master/HW3_solutions.ipynb
- http://www.farsiteforecast.com/rotten-tomatoes/
- http://www.slate.com/articles/arts/culturebox/2011/06/slates_hollywood_careeromatic.html
- http://www.vulture.com/2013/07/summer-movie-quiz-guess-the-rotten-tomato-score.html