Farewell to Item IDs: Unlocking the Scaling Potential of Large Ranking Models via Semantic Tokens

TL;DR

This paper introduces TRM, a framework that replaces item IDs with semantic tokens in large ranking models, significantly improving scalability, reducing storage, and enhancing performance in recommendation and search systems.

Contribution

The paper proposes a novel semantic token-based approach (TRM) that overcomes the limitations of item ID embeddings, enabling better scalability and stability in large ranking models.

Findings

33% reduction in sparse storage

0.85% increase in AUC performance

improved user engagement metrics in deployment

Abstract

Recent studies on scaling up ranking models have achieved substantial improvement for recommendation systems and search engines. However, most large-scale ranking systems rely on item IDs, where each item is treated as an independent categorical symbol and mapped to a learned embedding. As items rapidly appear and disappear, these embeddings become difficult to train and maintain. This instability impedes effective learning of neural network parameters and limits the scalability of ranking models. In this paper, we show that semantic tokens possess greater scaling potential compared to item IDs. Our proposed framework TRM improves the token generation and application pipeline, leading to 33% reduction in sparse storage while achieving 0.85% AUC increase. Extensive experiments further show that TRM could consistently outperform state-of-the-art models when model capacity scales. Finally, TRM has been successfully deployed on large-scale personalized search engines, yielding 0.26% and 0.75% improvement on user active days and change query ratio respectively through A/B test.