Collaborative filtering

Collaborative filtering, generally speaking, is the process of filtering for information or patterns using techniques involving collaboration between multiple data points. Collaborative filtering methods have a wide breadth of applications, ranging from monitoring data such as logs, application on financial data, e-commerce recommendations, and different web applications such as news sites.

In collaborative filtering for recommendation, the underlying assumption is that the best recommendations can be deduced from the interests of similar users. If the interests of two users match for a certain Item then it is highly likely that there interests will match for other items too as compared to another users with no similar interests. The intuition behind collaborative filtering is that people often get the best recommendations from someone with similar tastes like themselves like classmates, colleagues etc.

The primary input for any recommendation system based on collaborative...