Method

Pairwise ranking

Learn from item pair order (RankNet, PairLogit).

21 cases from 15 companies apply Pairwise ranking. 12 were checked in an online A/B test; 2 report results in production.

Typical tasks: Rank search results, Retrieve and rank from millions of items, Recommend from what the user just did, Recommend to new users and new items and 4 more. Companies: Expedia Group, Pinterest, VK, eBay, Avito, OLX and 9 more. Often combined with: Gradient boosting, A/B test, Matrix factorization, Two-tower model, ALS and 51 more.

Safe and Efficient Multi-Epoch Embedding Trainingsocial2026-11
Winner's Solution Walkthrough: 1st Place (AvitoTech ML Cup 2026)classifieds2026-05
Improving Quality of Recommended Content through Pinner Surveyssocial2025-12
Ranking VK Storiessocial2025-05
Does High Click-through Rate Lead to High Conversion Rate?travel2025-05
From RankNet to LambdaMART: Leveraging XGBoost for Enhanced Ranking Modelsclassifieds2025-02
Channel-Smart Property Search: How Expedia Tailors Rankings for Youtravel2024-07
Transforming Recommendationssocial2024-05
How we improved the 'Unfamiliar' setting recommendation algorithm in Yandex Musicmedia-streaming2024-05
Decoding the Regional Fashion Signatures using AIe-commerce2024-04
Learning Embeddings for Lodging Travel Conceptstravel2024-01
Evolving Recommendations: A Personalized User-Based Ranking Modele-commerce2023-10
Deploying Transformers in Yandex Musicmedia-streaming2023-05
A Job Posting Summary Generation Service Based on an LM Fine-Tuned with RLHFclassifieds2023-05
Text Deduplication: A Review—banking2023-05
Learning to rank diverselytravel2023-01
Multi-Relevance Ranking Model for Similar Item Recommendationmarketplace2022-06
Poster generation for films and cartoons in an online cinemamedia-streaming2022-06
Machine Learning and Personalization in Auto.ru Searchclassifieds2022-06
Going from 0 to 1 modeling User Preferences for Personalized Recommendationsmarketplace2022-03
Rethinking Zomato Search (Part One)food-delivery2022