NLP--film-genres-from-synopsis

Detailed explanation of the code is available as blog post or in this video

Details about the API are discussed in this video.

1. Background

Public movies’databases such as IMDB provide genre information to assist searching. The tagging of movies’genres is still a manual process which involves the collection of users'suggestions sent to known email addresses provided by the database. Movies are often registered with inaccurate genres. Automatic genres classification of a movie based on its synopsis not only speeds up the classification process by providing a list of suggestion but the result may potentially be more accurate than that of an un-trained human.

2. The data

Data have been downloaded from Radix challenge Kaggle dataset

https://gitlab.com/radix-ai/challenge

3. The model

This Python notebook contains the code to obtain automatically the film genres from the synopsis.

Best results have been obtained with deep learning, coding a very simple Keras LSTM. Higher accuracy than classical ML models such as Logistic Regression (50% vs. 10%)

This model has been deployed to a REST API using the Python library Flask.