The Best of the Two Worlds: Harmonizing Semantic and Hash IDs for Sequential Recommendation

2025-12-11

AI Summary

To address poor long-tail item performance and head–tail recommendation imbalance in sequential recommendation, this paper proposes a dual-branch framework integrating Semantic ID (SID) and Hashed ID (HID). Conventional HID lacks semantic generalization capability, while SID suffers from collaborative dominance and quantization distortion, leading to an inherent trade-off between head and tail performance. Our work introduces, for the first time, dual-branch modeling with hierarchical alignment: the HID branch preserves unique collaborative identifiability, whereas the SID branch captures multi-granularity semantic sharing. A cross-representation alignment loss jointly optimizes both branches. Extensive experiments on three real-world datasets demonstrate significant improvements in Recall@20, with simultaneous gains in both head and tail item recommendation quality, consistently outperforming state-of-the-art methods.

Technology Category

• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Recommender Systems
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Applications of semantic technologies for improving search, browsing, recommendation, personalization

Abstract

Conventional Sequential Recommender Systems (SRS) typically assign unique Hash IDs (HID) to construct item embeddings. These HID embeddings effectively learn collaborative information from historical user-item interactions, making them vulnerable to situations where most items are rarely consumed (the long-tail problem). Recent methods that incorporate auxiliary information often suffer from noisy collaborative sharing caused by co-occurrence signals or semantic homogeneity caused by flat dense embeddings. Semantic IDs (SIDs), with their capability of code sharing and multi-granular semantic modeling, provide a promising alternative. However, the collaborative overwhelming phenomenon hinders the further development of SID-based methods. The quantization mechanisms commonly compromise the uniqueness of identifiers required for modeling head items, creating a performance seesaw between head and tail items. To address this dilemma, we propose extbf{ ame}, a novel framework that harmonizes the SID and HID. Specifically, we devise a dual-branch modeling architecture that enables the model to capture both the multi-granular semantics within SID while preserving the unique collaborative identity of HID. Furthermore, we introduce a dual-level alignment strategy that bridges the two representations, facilitating knowledge transfer and supporting robust preference modeling. Extensive experiments on three real-world datasets show that ame~ effectively balances recommendation quality for both head and tail items while surpassing the existing baselines. The implementation code can be found onlinefootnote{https://github.com/ziwliu8/H2Rec}.

Problem

• Balances head and tail item recommendation performance
• Harmonizes semantic and hash IDs for sequential recommendation
• Addresses long-tail and collaborative overwhelming issues in SRS

Innovation

• Dual-branch architecture harmonizing semantic and hash IDs
• Dual-level alignment strategy bridging two representations
• Balancing head-tail performance via collaborative-semantic integration