Birds of a Feather Cluster Nearby: a Proximity-Aware Geo-Codebook for Local Service Recommendation

2026-04-25

AI Summary

This work addresses a critical limitation in existing semantic encoding methods for recommendation systems, which often neglect geographical constraints, yielding results that are semantically relevant but spatially infeasible. To bridge this gap, the authors propose Pro-GEO, a novel approach that explicitly models geographic proximity as an orthogonal rotation transformation within high-dimensional embeddings. By introducing a local coordinate system centered on geographic anchors and incorporating geographically aware rotational positional encoding, Pro-GEO achieves a balanced fusion of semantic and spatial signals. Integrated with semantic ID tokenization and a generative recommendation architecture, the method demonstrates substantial improvements on large-scale industrial datasets: it reduces average geographic clustering distance by 45.60% and increases Hit@50 by 1.87%, effectively harmonizing semantic relevance with geographic feasibility.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Search and Optimization: Learning to Search
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Generative recommendation systems are increasingly adopted in local service platforms, where semantic relevance alone is insufficient without strict geographic feasibility. A key technical challenge lies in semantic ID (SID) tokenization, which directly impacts the recommendation performance. However, existing semantic codebooks neglect geographic constraints, often resulting in recommendations that are semantically relevant yet geographically unreachable. To address this limitation, we propose Pro-GEO, a Proximity-aware GEO-codebook. Pro-GEO establishes a geo-centroid local coordinate system to capture intra-cluster spatial relationships and a geo-rotary position encoding mechanism that models geographic proximity as orthogonal rotational transformations in the high-dimensional embedding. This design enables semantic and spatial signals to be jointly modeled in a balanced manner, without reducing geographic information to a weak auxiliary feature. Extensive experiments conducted on a large-scale industrial dataset reveal that Pro-GEO significantly outperforms state-of-the-art methods. In particular, Pro-GEO reduces the average geographic clustering distance by 45.60% and achieves a 1.87% improvement in Hit@50, highlighting its effectiveness for real-world local service recommendation.

Problem

• local service recommendation
• geographic feasibility
• semantic ID tokenization
• geo-codebook
• proximity awareness

Innovation

• proximity-aware
• geo-codebook
• semantic ID tokenization
• rotary position encoding
• local service recommendation