Concept 1 / 9
What is a Recommender System?
Recommender systems emerged as a serious research and industry topic in the 1990s, driven by the rapid growth of the web as a medium for commerce and information exchange. Key milestones:
| Year | Milestone |
|---|---|
| 1992 | Tapestry (e-mail filtering) — coined the term collaborative filtering |
| 1994 | GroupLens — first news recommender system |
| 1995 | Ringo — music recommendation |
| 1997 | Term "Recommender System" formally established; continuous extension to new domains |
Definition: The general idea of a recommender system is that a recommendation seeker receives a recommendation drawn from the universe of alternatives. The system uses information and data about the user's preferences and similarities to other users to generate this recommendation.
Three roles appear in every recommender scenario:
- Recommendation seeker — the person who receives the suggestion
- Preference provider — other users whose ratings and behavior inform the algorithm
- Recommender — the system that processes preferences and similarity scores to output a ranked suggestion
The underlying assumption is that past behavior (ratings, purchases, browsing history) is a reliable predictor of future preferences. The more interaction data is available — both for the target user and for similar users — the more accurate the predictions become. This enables personalization at scale: instead of one store for all customers, each customer effectively receives their own curated view of the product catalog.