Latent user preference model improves short video recommendations
Mend the Measurement Gap: Latent User Preference Modeling for Short-Form Video Recommendation
Information RetrievalMachine Learning
Summary
Watching a short video doesn't always mean someone genuinely likes it—people might watch because they are bored, distracted, or the video is just quick to watch. The authors found that using simple watch time can be misleading for guessing what users really enjoy, especially because short and long videos behave differently. They created a new model that figures out the hidden true preference behind watching videos by treating observed behaviors as noisy clues influenced by video length and other factors. Testing on YouTube Shorts showed this method better predicts what people actually like and improves viewer satisfaction.
What this means in practice
- •For video platform engineers: Integrate the latent value score as a ranking feature to improve recommendation relevance without biasing towards short videos.
- •For online advertiser teams: Use refined user preference signals to better target ads by distinguishing genuine engagement from incidental watch time.
Authors
Shuo Chang, Yueqi Wang, Zihuan Diao, Ali Montazer, Jiangguo Zhang, Joyneel Misra, Dapeng Hong, Tomer Margolin, Sourabh Bansod, Ningren Han
Abstract
Recommender systems rely heavily on heterogeneous behavioral feedback to infer user preference. Although abundant, these signals are imperfect measurements: the same observed behavior can arise from different underlying states, such as genuine enjoyment, passive consumption, or inattention. The challenge is especially acute in short-form video, where watch-based signals are strongly affected by measurement confounders such as video duration - the same watch time can imply different levels of preference for videos of different lengths, while ratio-based metrics can systematically favor short videos. As a result, optimizing raw engagement can amplify measurement artifacts rather than improving user value. We propose a Factorized Latent Value Model (FLVM) for measuring user preference from heterogeneous behavioral feedback. The model treats observed behaviors as noisy measurements of a low-dimensional, factorized latent value state and uses structured output heads to model heterogeneous feedback signals. A restricted baseline path captures predictable variation from measurement-confounding features such as video duration, user propensity, and session context, while a routed latent path estimates preference-relevant value advantage. The resulting latent value score can be integrated into an existing recommender system as a ranking feature or ranking score. On YouTube Shorts, a major short-form video platform, this model improves offline metrics and lifts a primary viewer enjoyment metric by 2.67% in online A/B tests.