Glossary term
Glossary term
Evaluation and Benchmarks
The part of a recommendation system that provides a value or ranking for each item produced by the candidate generation phase.
Created for this library
A retail recommender team uses a heavy scoring model on the top candidates retrieved by a cheaper candidate generator.
A search team uses scoring with a deep ranker on the top few hundred candidates while a cheap recall model handles the long tail.
An e-commerce team uses scoring with business signals like inventory and margin in addition to relevance for its homepage ranker.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License