Method

Matrix factorization

Latent-factor decomposition of the interaction matrix (ALS, SVD, SVD++).

23 cases from 18 companies apply Matrix factorization. 14 were checked in an online A/B test; 4 report results in production. Includes variants: ALS.

Typical tasks: Retrieve and rank from millions of items, Recommend to new users and new items, Recommend from what the user just did, Balance relevance and diversity and 4 more. Companies: VK, MegaFon, Netflix, Expedia Group, Target, Avito and 12 more. Often combined with: ALS, Gradient boosting, Pairwise ranking, A/B test, Popularity baseline and 48 more.

Measuring the Impact of Personalized Recommendationsmedia-streaming2026-10
Ranking VK Storiessocial2025-05
Developing Multimodal Content Models for Recommendationssocial2025-05
A Recommender System Algorithm Accounting for Cultural Factors to Solve Cold-Start Problemsgeneral2025-05
Speeding Up Inference on Graph Models for Warm Start (AIRI)general2025-05
Does High Click-through Rate Lead to High Conversion Rate?travel2025-05
Contextual Offer Recommendations Engine at Targetretail2024-08
How We Moved from a Markov Chain to a SASRec Modelclassifieds2024-05
Experts Are Shocked: A Brilliant Strategy for Finding an Audience for Clickbait Contentsocial2024-05
From Small to Great: Finding and Growing Promising New Authors on VKsocial2024-05
Transforming Recommendationssocial2024-05
Warming up a Cold Usermedia-streaming2024-05
How we improved the 'Unfamiliar' setting recommendation algorithm in Yandex Musicmedia-streaming2024-05
Decoding the Regional Fashion Signatures using AIe-commerce2024-04
Transforming Recommendations at ASOSe-commerce2024-04
How We Relaunched Recommendations at Auto.ruclassifieds2023-05
Deploying Transformers in Yandex Musicmedia-streaming2023-05
Creating the Conditions to Launch a Recommender System with Manual Labelingsocial2023-05
MegaFon RecSys Workshop, part 2: a two-level recommender (LightFM + CatBoost) and DSSM modelsmedia-streaming2022-06
MegaFon RecSys Workshop, part 1media-streaming2022-06
Beyond Matrix Factorization: Using hybrid features for user-business recommendationsmarketplace2022-04
Going from 0 to 1 modeling User Preferences for Personalized Recommendationsmarketplace2022-03
Lightweight Representation Learning For Efficient And Scalable Recommendationadtech2022-01