Content and Collaborative filtering for User-Movie recommendation system.
Content filtering:
- Content filtering takes user-item similarity into account. Implemented K means clustering algorithm to cluster the movies based on contents (genres of the movies). Predicted rating for a particular movie based on the ratings that user give to other movies grouped in the same cluster.
Collaborative filtering:
- Collaborative filtering takes user-user similarity into account instead of user-item. Implemented neural network in sense of matrix factorization. No information about content of the movies has been used.
More clarification and updates will be made later.