空间推理：为基于大语言模型的生成下POI推荐启用地理推理
Reasoning Over Space: Enabling Geographic Reasoning for LLM-Based Generative Next POI Recommendation
AI总结：
本文提出ROS框架，通过地理信息作为决策变量提升生成式推荐的地理推理能力，采用分层空间语义ID和三阶段移动链式推理，结合空间引导强化学习，实现跨城市迁移和命中率提升。
摘要：
基于大语言模型（LLMs）的生成式推荐将预测重新定义为序列生成，但现有方法在利用地理信号方面存在局限。本文提出Reasoning Over Space（ROS）框架，利用地理信息作为推理过程中的关键决策变量。ROS引入分层空间语义ID（SID），将粗到细的地域和POI语义离散化为组合性标记，并赋予LLM三阶段移动链式推理（CoT）范式，以建模用户个性、构建意图对齐的候选空间并执行基于地域的剪枝。进一步通过空间引导强化学习（RL）将模型与真实地理对齐。在三个广泛使用的基于位置的社交网络（LBSN）数据集上的实验表明，ROS在命中率上比最强的LLM基线模型高出超过10%，并且在使用较小的骨干模型的情况下提高了跨城市迁移能力。
英文摘要：
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.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
