Towards Large-scale Generative Ranking

TL;DR

This paper explores the potential of generative ranking in large-scale recommender systems, demonstrating its effectiveness and efficiency through theoretical analysis and real-world experiments on Xiaohongshu's platform.

Contribution

It introduces GenRank, a novel generative architecture for ranking, and provides empirical evidence of its advantages over traditional recommenders in industrial settings.

Findings

Generative ranking outperforms current recommenders in effectiveness.

GenRank achieves significant user satisfaction improvements.

Efficient deployment with comparable computational resources.

Abstract

Generative recommendation has recently emerged as a promising paradigm in information retrieval. However, generative ranking systems are still understudied, particularly with respect to their effectiveness and feasibility in large-scale industrial settings. This paper investigates this topic at the ranking stage of Xiaohongshu's Explore Feed, a recommender system that serves hundreds of millions of users. Specifically, we first examine how generative ranking outperforms current industrial recommenders. Through theoretical and empirical analyses, we find that the primary improvement in effectiveness stems from the generative architecture, rather than the training paradigm. To facilitate efficient deployment of generative ranking, we introduce GenRank, a novel generative architecture for ranking. We validate the effectiveness and efficiency of our solution through online A/B experiments. The results show that GenRank achieves significant improvements in user satisfaction with nearly equivalent computational resources compared to the existing production system.