Coarse-to-fine Dynamic Uplift Modeling for Real-time Video Recommendation

TL;DR

This paper introduces Coarse-to-fine Dynamic Uplift Modeling (CDUM), a novel approach for real-time video recommendation that combines offline user preferences with online real-time interests, improving personalization and effectiveness.

Contribution

The paper proposes a new uplift modeling framework for video recommendation, addressing treatment design and real-time interest capture, and demonstrates its effectiveness through extensive experiments and deployment.

Findings

CDUM outperforms existing models in offline and online tests.

The model is deployed on Kuaishou, serving hundreds of millions daily.

Experimental results show significant improvements in recommendation accuracy.

Abstract

With the rise of short video platforms, video recommendation technology faces more complex challenges. Currently, there are multiple non-personalized modules in the video recommendation pipeline that urgently need personalized modeling techniques for improvement. Inspired by the success of uplift modeling in online marketing, we attempt to implement uplift modeling in the video recommendation scenario. However, we face two main challenges: 1) Design and utilization of treatments, and 2) Capture of user real-time interest. To address them, we design adjusting the distribution of videos with varying durations as the treatment and propose Coarse-to-fine Dynamic Uplift Modeling (CDUM) for real-time video recommendation. CDUM consists of two modules, CPM and FIC. The former module fully utilizes the offline features of users to model their long-term preferences, while the latter module leverages online real-time contextual features and request-level candidates to model users' real-time interests. These two modules work together to dynamically identify and targeting specific user groups and applying treatments effectively. Further, we conduct comprehensive experiments on the offline public and industrial datasets and online A/B test, demonstrating the superiority and effectiveness of our proposed CDUM. Our proposed CDUM is eventually fully deployed on the Kuaishou platform, serving hundreds of millions of users every day. The source code will be provided after the paper is accepted.