HF-SID: High-Fidelity Semantic IDs for Generative Retrieval in Location-Based Services
2026-08-31 • Information Retrieval
Information Retrieval
AI summaryⓘ
The authors address problems in how location-based services represent places using short codes called Semantic IDs (SIDs). They note that existing methods lose important details like exact geographic distance, varying numerical attributes, and hierarchical categories. To fix this, the authors propose HF-SID, which improves the details in the representation by using 3D coordinates, better number encoding, and a learning method to distinguish similar places nearby. Their approach keeps the ID short without extra cost while preserving more useful information.
Generative retrievalLocation-Based Services (LBS)Semantic ID (SID)3D Cartesian coordinatesType-aware embeddingsContrastive learningPoint-of-Interest (POI)Hierarchical affiliationGeographic distance approximation
Authors
Haowen Lin, Jing Li, Zhibin Hao, Fangye Wang, Lihui Su, Song Yang, Xiaojiang Zhou, Pengjie Wang
Abstract
Generative retrieval has attracted increasing attention in Location-Based Services (LBS), where each Point-of-Interest (POI) is represented as a Semantic ID (SID). As the SID is the only channel through which POI information reaches the generative model, whatever it fails to preserve is irrecoverable at decoding time, and LBS retrieval is especially sensitive to the fine-grained differences that existing SIDs blur. Specifically, (1) LLMs embed continuous coordinates discontinuously, so their numeric differences do not reflect true geographic distance; (2) dynamic numerical attributes differ vastly in scale, so an identical gap may be decisive for one attribute yet negligible for another; and (3) short text cannot convey hierarchical affiliation, as text-similar POIs may belong to different hierarchies. We therefore propose HF-SID, which restores geographic, numerical, and structural fidelity at the representation stage, before any information is committed to a discrete code. It transforms coordinates into a continuous 3D Cartesian form and encodes each numerical value as a single unit, consolidated inside the LLM by Geo-CPT and Num-CPT with type-aware embeddings; a Structure-based Contrastive Learning objective, applied only to the last-layer residual, then separates co-located POIs that share a coarse tag but differ at the fine level. Because these mechanisms enrich the representation rather than lengthen the identifier, HF-SID uses a 3-token SID at no extra decoding cost. On a large-scale industrial