Why Users Go There: World Knowledge-Augmented Generative Next POI Recommendation

2026-05-12

AI Summary

This work addresses the challenge that existing generative point-of-interest (POI) recommendation methods struggle to effectively integrate dynamic real-world information—such as local events and cultural trends—that influences users’ mobility decisions. To bridge this gap, we propose AWARE, a novel framework that introduces large language model (LLM) agents to generate narrative texts embedding spatiotemporal context, thereby precisely aligning dynamic external knowledge with individual user behavior patterns while mitigating noise interference. By synergistically combining contextual narrative generation, spatiotemporal behavior modeling, and generative recommendation, AWARE significantly outperforms current state-of-the-art baselines across three real-world datasets, achieving relative performance improvements of up to 12.4%.

Abstract

Generative point-of-interest (POI) recommendation models based on large language models (LLMs) have shown promising results by formulating next POI prediction as a sequence generation task. However, the knowledge encoded in these models remains fixed after training, making them unable to perceive evolving real-world conditions that shape user mobility decisions, such as local events and cultural trends. To bridge this gap, we propose AWARE (Agent-based World knowledge Augmented REcommendation), which employs an LLM agent to generate location- and time-aware contextual narratives that capture regional cultural characteristics, seasonal trends, and ongoing events relevant to each user. Rather than introducing generic or noisy information, AWARE further anchors these narratives in each user's behavioral context, grounding external world knowledge in personalized spatial-temporal patterns. Extensive experiments on three real-world datasets demonstrate that AWARE consistently outperforms competitive baselines, achieving up to 12.4% relative improvement.

Problem

• next POI recommendation
• world knowledge
• large language models
• user mobility
• contextual dynamics

Innovation

• world knowledge augmentation
• LLM agent
• next POI recommendation
• contextual narrative generation
• personalized spatio-temporal modeling