SPARC: Sequence-aware Progressive Attribute Routing and Compression Framework for Generative Recommendation

2026-07-28

AI Summary

This work addresses the limitation of existing static semantic ID methods, which neglect interaction context and risk losing critical information when directly compressing multi-attribute user behaviors. To overcome this, the authors propose a context-aware progressive attribute routing and compression mechanism. First, sequential modeling captures dependencies among attribute fields to generate context-aware representations. These, along with original and identity embeddings, are then routed into multiple slots to preserve complementary information. Finally, a lightweight cross-item interaction module compresses the multi-slot representations into a single token, enhancing user history representation without increasing input length. Evaluated on Taobao and Amazon datasets, the proposed method significantly outperforms both traditional and generative baselines, demonstrating the effectiveness of preserving context-conditioned information.

Abstract

Generative recommendation tokenizes items as discrete Semantic IDs (SIDs) and autoregressively generates target items from users' historical SID sequences. Although existing SIDs incorporate multimodal and structured information, they are typically statically assigned and independent of the current interaction context. In industrial scenarios, each behavior also contains heterogeneous attributes, such as category, brand, price, behavior type, and timestamp. Fully expanding these features greatly increases the input length, while directly compressing them into a single representation may prematurely discard context-relevant information. We propose \textbf{SPARC}, \uline{\textbf{S}}equence-aware \uline{\textbf{P}}rogressive \uline{\textbf{A}}ttribute \uline{\textbf{R}}outing and \uline{\textbf{C}}ompression Framework for Generative recommendation. SPARC first models the sequential dependencies of each field type to obtain context-aware field representations. It then routes the original, contextual, and identity representations of different fields into multiple slots to preserve complementary information under a fixed capacity. Finally, lightweight cross-item interaction integrates the intermediate tokens and compresses each historical item into a single token. Following the principle of contextualizing before compression, SPARC enriches user-history representations without increasing the input length of the generative backbone. Experiments on industrial Taobao and public Amazon datasets demonstrate that SPARC outperforms strong conventional and generative baselines. Further comparisons with static compression variants show that the improvement of SPARC comes from context-conditioned information retention rather than merely increasing the expressiveness of the compression module.

Problem

• generative recommendation
• heterogeneous attributes
• context-aware representation
• sequence modeling
• information compression

Innovation

• generative recommendation
• context-aware compression
• attribute routing
• sequence modeling
• Semantic IDs