HGenPush: A Heterogeneous Generative Recommendation Architecture for Industrial Push Notification Systems

2026-07-03

AI Summary

This work addresses the limitations of existing generative recommender systems, which are often confined to a single content modality and rely on inefficient autoregressive generation, thereby struggling to jointly recommend high-quality content and credible creators. To overcome these challenges, we propose HGenPush, an end-to-end heterogeneous generative recommendation framework that, for the first time, unifies video and creator recommendation within a single architecture. HGenPush introduces a non-autoregressive multi-token prediction mechanism to enhance generation efficiency and incorporates modules for multi-scenario user behavior modeling and preference alignment, leveraging user feedback as reward signals to optimize output quality. Deployed on Kuaishou’s short-video recommendation platform, the system achieves a statistically significant 0.181% increase in daily active users.

Abstract

With the explosive growth of content platforms, recommendation systems need to better satisfy user demands to enhance user satisfaction and retention. Taking short-video platforms as an example, users not only seek high-quality content but also trusted authors. Although generative recommendation systems have achieved breakthroughs in recent years, existing methods primarily generate single-type recommendation content and typically employ the inefficient autoregressive paradigm to generate semantic IDs. In this paper, we propose an end-to-end heterogeneous generative recommendation architecture called HGenPush. First, we design a hybrid user behavior understanding module that integrates multi-scenario and multi-perspective behaviors to capture precise user interest. Then, we design a dual-branch heterogeneous generative recommendation module that integrates video recommendation and author recommendation within a unified framework. In addition, to improve generation efficiency, we design a lightweight multi-token prediction method that discards the autoregressive paradigm. Finally, we design a user consumption preference alignment module, which leverages user feedback as reward signals to guide the model toward generating higher-quality content, thereby enhancing user experience and engagement. Through these designs, HGenPush simultaneously fulfills users' demands for high-quality content and trusted authors. We have deployed HGenPush on the push notification system of Kuaishou, a large-scale short-video platform, achieving a significant 0.181% increase in daily active users.

Problem

• generative recommendation
• heterogeneous recommendation
• push notification systems
• user satisfaction
• short-video platforms

Innovation

• heterogeneous generative recommendation
• multi-token prediction
• dual-branch architecture
• user preference alignment
• push notification system