README

Bachelor

Name: Chi Sam Mac

Student-number: s2588382

How to run the classifier

First you need to have all subtitles files (directory subtitle, with subdirectories the genres: Comedy, Drama, Documentary, Thriller, etc.)

The Doc2vec model: d2v_150.model

The movie_classification.py file.

Then run movie_classification.py* as: python3 movie_classification.py Comedy Drama Documentary Horror

*POS tag feature commented to increase efficiency of program, as the feature does not add anything to performance of the classifier. To turn it on uncomment the following lines: 35, 392-412, 427, 534-537, 549.

Files neccesary to gather all data from scratch

Directory "subtitles" with in it directories of the genres ("Action", "Comedy", "Documentary", "Drama", "Horror", "Thriller")

movies.csv (movieLens data roughly 9k different movies)

movies2.csv (movieLens data roughly 27k different movies)

transform_data.py

get_subs.py

scrapper.py

doc2vec.py

grid_search.py (for parameters algorithms)

gs_feature_weights.py (for weights of features)

base_program.py (classifier with basic BOW model)

movie_classification.py

Steps to gather all data from scratch and run the classifier

Step1: python3 transform_data.py #To get the data of which movies to get subtitles from

Step2: python3 get_subs.py #This downloads all the subtitles for the movies chosen before

Step3: python3 doc2vec.py #Create a doc2vec model

Step4: python3 movie_classification.py (Don't need grid_search.py/gs_feature_weights.py, already used for parameters/weights in classifier in movie_classification.py)