A recommendation engine is a machine learning system that filters and ranks content or products to serve each user the most relevant suggestions at a given moment. These systems learn from data, including past behavior, preferences, and interactions, to predict what a user is likely to want next. They are the technology behind product suggestions on Amazon, content feeds on Netflix and TikTok, and playlist generation on Spotify.
There are three main approaches to building recommendation engines.
- Content-based filtering analyzes the attributes of items and recommends others with similar characteristics, such as suggesting crime films to someone who has watched several crime films.
- Collaborative filtering looks at patterns across many users and identifies people with similar tastes, recommending items that others with overlapping preferences have engaged with.
- Hybrid filtering combines both approaches to compensate for the weaknesses of each and is the model used by most modern platforms.
Recommendation engines rely on two types of training data. First is explicit data, which comes directly from users in the form of ratings, likes, or reviews. Second is implicit data, which is collected passively through behavior such as watch times or purchase history.