Method
Sequential recommendation
Next-item prediction from ordered user histories with RNN or transformers.
42 cases from 19 companies apply Sequential recommendation. 19 were checked in an online A/B test; 12 report results in production. Includes variants: SASRec.
Typical tasks: Recommend from what the user just did, Retrieve and rank from millions of items, Recommend to new users and new items, Reduce ML cost and 6 more. Companies: Yandex, Pinterest, Netflix, Meta, Avito, Airbnb and 13 more. Often combined with: Multi-task learning, Two-tower model, SASRec, In-batch negatives, CTR / feedback prediction and 74 more.






