We used the dataset movie.zip published at Cornell University website that contains more than 27,000 reviews of movies published in IMDb to perform a sentiment analysis.
For the sentiment analysis, we downloaded a dataset from Stanford University that contains a list of 5o,000 highly polar movie reviews. This dataset is used to build two dictionaries of words: one that assignes a value to a word according to how positive it is, and another one that tells you how negative it is. Taking this two dictionaries into account, the sentiment of a review is calculated.
The following list shows the top-10 movies by positive reviews grade:
Leviathan and Paradise have the most positive review grades. We can corroborate this reading the reviews, where we can find sentences such as “It is one of the most beautiful films I have ever seen” in Leviathan, or “This is one of the best films I’ve seen up to now, a movie that you mustn’t miss” in Paradise.
On the other hand, we can also find the top-10 movies by negative reviews grade:
In this case, we can also check that this rating are correct. For example, we can read “This has got to be the worst movie I’ve blown five bucks on in a very long time” in Shocker, or “a lot of the movie is so bad that it’s funny” in The Guardian.
moviesanalysis.wordpress.com 