horse
Handy Open Recommendation Service
- Frontend app url:
- http://test-horse.s3-website-eu-west-1.amazonaws.com/
- Backend app url:
- http://horse-env.pqmnkwtbum.eu-west-1.elasticbeanstalk.com/
- Frontend app repository:
- https://github.com/pragmaticcoders/horse-frontend
Recommendation assumptions
- User movie recommendations are influenced by general movie popularity.
- Movies liked by followed users are more likely to be recommended. That relation is recursive, but with exponentialy decreasing power.
- Similarity score between any pair of users can be calculated. It is based on a count of commonly liked movies and is adjusted according to a total number of movies graded by both users.
- Movie recommendations are influenced by all points above.
Installation
Python 3.4 is required.
$ cd horse $ pip install . $ python application.py
Testing
$ tox
Populate with example dataset
$ curl -X POST http://127.0.0.1:5000/populate -d @./examples/populate.json --header "Content-Type: application/json"