Recommendation Engines, also known as recommendation systems, are awesome. Scroll down to read about the overview of the different types. Use the navigation bar to see working examples.
There are recommendation engines with and without user profiles. User profiles remember what a user has chosen before. Some recommendation engines rely extensively on user profiles.
The cold start problem refers to how to start user profiles.
Explainable AI (XAI) is the concept of trying to explain to a human how the computer came up with a particular answer. In the context of recommendation engines, XAI attempts to explain how the engine came up with the recommendation. Darpa's XAI page.
Hover over any of the boxes to learn more. The links in the navigation menu take you to working examples.
Knowledge-based recommendation engines are some of the most basic forms of recommendation engines. They simply remove (i.e. filter out) unwanted content or query a database to get the desired content.
Content-based recommendation engines recommend items that are similiar to the ones you have selected based solely on the content of the items you have selected.
Collaborative filtering takes into account what other people have done and recommends items that are similar to what other people have selected. This is the most common non-hybrid recommendation engine type.
The most commonly used type of recommendation engines actually used in real life. This type takes different parts (e.g. content based and collaborative filtering) and merges them.