AI Insight
Researchers developed the Controllable Retrieval Model (CRM), a recommendation system that incorporates desired watch time as an input condition to retrieve content more effectively. The model learns to predict the next item based on observed watch time during training, then uses desired watch time targets at inference to guide recommendations without requiring retraining. Implemented as both a deep neural network two-tower model and a Decision Transformer, CRM was deployed in Kuaishou's short-video platform serving over 400 million users, achieving watch time improvements of 0.372% to 0.457% in A/B tests.
Why it matters
This approach demonstrates a practical method for recommendation systems to optimize for specific engagement metrics while maintaining computational efficiency through precomputed embeddings. The real-world deployment results suggest that outcome-conditioned retrieval can meaningfully improve user engagement in large-scale video platforms, potentially applicable to other content recommendation domains.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
-cross
Abstract: Retrieval models for recommendation need to account for outcomes such as watch time while searching large item corpora efficiently. We present the Controllable Retrieval Model (CRM), an outcome-conditioned retrieval framework that adds watch time as an input to the user tower. During training, CRM learns to predict the next item given its observed watch time; at inference, we supply a desired watch time to steer retrieval without retraining. The item tower is independent of the condition, so item embeddings can still be precomputed and indexed for approximate nearest neighbor (ANN) search. We implement CRM as both a DNN-based two-tower model and a Decision Transformer. The Transformer interleaves item tokens with watch-time-to-go condition tokens, using the target watch time as a numerical prompt for next-item prediction. For each online request, we choose between the average and maximum watch times in the user’s recent history as the condition. CRM is deployed in Kuaishou’s short-video recommendation system, which serves over 400 million users. Online A/B tests show video watch time improvements of 0.372% in Kuaishou and 0.457% in Kuaishou Lite.
Source: CRM: Retrieval Model with Controllable Condition