Naman Jogani
10/22/2019
David Bellanger
A Survey of Collaborative Filtering Techniques
Article addresses the approach of collaborative filtering. This is one of the most successful
approaches to building recommender systems. Collaborative filtering uses a database of preferences for
items by users to predict additional topics or products a new user might like.
The example of amazon and Barnes and Nobles website illuminates the memory based CF
methods that are deployed into commercial systems. The reason being that they are easy to implement
and customization of CF systems for each user decreases the search effort for users. However, there are
certain limitations to this approach aswell. The similarity values are based on common items and
therefor are unreliable when data are sparse, and the common items are therefore few. Along with that
they are dependent on human ratings, and cannot recommend for new users and items.
