Collaborative filtering is based on the opinion that people who decided to make a purchase in the past will decide in the future, and that they will likely prefer similar kinds of items as they did in the past. The system generates recommendations using data about rating profiles for different users or goods.
The collaborative filtering approach has lots of advantages. One of them is that it is capable of accurately recommending complex items such as movies without requiring an "understanding" of the item itself. Many algorithms have been used in measuring user similarity or item similarity in recommender systems.
Further, there are several types of collaborative filtering algorithms:
- User-User Collaborative Filtering
- Item-Based Collaborative Filtering
- Other simpler algorithms
Although the collaborative filtering method is clear enough, there may be some problems while implementing it. For example, a cold start
can be an issue. For a new user or item, there isn't enough historical data to make accurate recommendations. There may be an issue of a product cold start or user cold start. The user cold start problem occurs when new users enter a website or app for the first time and the system has no information about them or their preferences. In this case, the system fails to recommend anything. Similarly for new products, as they have no reviews, likes, clicks, or other interactions among users, so no recommendations can be made.
One of the methods to deal with the issue is to recommend trending products to the new customer in the early stages. Here the selection can be narrowed down based on contextual information – their location, which site the visitor came from, a device used, etc. Behavioral information will be collected after a few clicks during that first visit, and start to build up from there.
Another way to deal with the problem of the cold start is by using metadata about the new product when creating recommendations.