A Model-based Multi-Agent Personalized Short-Video Recommender System

TL;DR

This paper introduces a model-based multi-agent reinforcement learning framework for personalized short-video recommendation, effectively modeling user preferences and maximizing watch-time, and has been successfully deployed at scale.

Contribution

It presents a novel multi-agent RL framework for short-video recommendation that addresses sample bias and is suitable for industrial deployment.

Findings

Outperforms alternative methods in offline evaluations

Successfully deployed in a large-scale platform

Serves hundreds of millions of users

Abstract

Recommender selects and presents top-K items to the user at each online request, and a recommendation session consists of several sequential requests. Formulating a recommendation session as a Markov decision process and solving it by reinforcement learning (RL) framework has attracted increasing attention from both academic and industry communities. In this paper, we propose a RL-based industrial short-video recommender ranking framework, which models and maximizes user watch-time in an environment of user multi-aspect preferences by a collaborative multi-agent formulization. Moreover, our proposed framework adopts a model-based learning approach to alleviate the sample selection bias which is a crucial but intractable problem in industrial recommender system. Extensive offline evaluations and live experiments confirm the effectiveness of our proposed method over alternatives. Our proposed approach has been deployed in our real large-scale short-video sharing platform, successfully serving over hundreds of millions users.