Reasoning Over Space: Enabling Geographic Reasoning for LLM-Based Generative Next POI Recommendation

TL;DR

This paper introduces ROS, a framework that enhances large language models for POI recommendation by integrating geographic reasoning through hierarchical spatial tokens and a three-stage Chain-of-Thought process.

Contribution

ROS is the first to incorporate geographic signals into LLM-based recommendation via hierarchical spatial tokens and a mobility-aware reasoning paradigm.

Findings

ROS achieves over 10% relative gains in hit rate over baselines.

ROS improves cross-city transfer performance.

ROS outperforms existing LLM-based recommenders despite using smaller models.

Abstract

Generative recommendation with large language models (LLMs) reframes prediction as sequence generation, yet existing LLM-based recommenders remain limited in leveraging geographic signals that are crucial in mobility and local-services scenarios. Here, we present Reasoning Over Space (ROS), a framework that utilizes geography as a vital decision variable within the reasoning process. ROS introduces a Hierarchical Spatial Semantic ID (SID) that discretizes coarse-to-fine locality and POI semantics into compositional tokens, and endows LLM with a three-stage Mobility Chain-of-Thought (CoT) paradigm that models user personality, constructs an intent-aligned candidate space, and performs locality informed pruning. We further align the model with real world geography via spatial-guided Reinforcement Learning (RL). Experiments on three widely used location-based social network (LBSN) datasets show that ROS achieves over 10% relative gains in hit rate over strongest LLM-based baselines and improves cross-city transfer, despite using a smaller backbone model.