PopcornBook

Solo Project by Rishabh Jain

Check it out at http:https://popcornbook.net

Most of the machine learning code is redacted, however you can see how the implementation of the engine works.

HOW IT WORKS:

I scraped plots of about 17,000 books from Wikipedia and stored it in a data file which can be found here. 

Using NLTK, roots for different words were found and weights were allotted in a custom dictionary accordingly.

Then, this data was fed into a Term Frequency Inverse Document Frequency vectorizer, which basically means that if a term occurs
more frequently across books, its weight is reduced significantly. This greatly helps in uniquely identifying plot structures.
This engine is then serialized and stored to reduce stress on the server and user wait time.

When the user enters a movie name, the server fetches the plot data from IMDB and asks the vectorizer to generate a prediction 
based on this data.



If you decide to use my code, please make sure to cite it properly.