SeMob: Semantic Synthesis for Dynamic Urban Mobility Prediction

2025-09-24

AI Summary

Human mobility prediction is highly susceptible to disruptions caused by exogenous events, yet existing models struggle to effectively capture the semantic content of event-related textual descriptions and their spatiotemporal impacts. To address this, we propose a multi-agent large language model (LLM)-driven dynamic mobility forecasting framework. First, collaborative agents automatically extract, reason over, and ground event text semantics in spatiotemporal contexts. Second, we design a progressive fusion architecture that injects event semantics as priors into a spatiotemporal graph neural network, enabling joint semantic–spatiotemporal modeling. This approach significantly enhances prediction consistency and robustness in regions proximal to events. Evaluated on a newly constructed city-scale event-augmented dataset, our method reduces MAE by 13.92% and RMSE by 11.12% compared to state-of-the-art baselines, demonstrating the efficacy of semantics-guided spatiotemporal dynamic modeling.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Human mobility prediction is vital for urban services, but often fails to account for abrupt changes from external events. Existing spatiotemporal models struggle to leverage textual descriptions detailing these events. We propose SeMob, an LLM-powered semantic synthesis pipeline for dynamic mobility prediction. Specifically, SeMob employs a multi-agent framework where LLM-based agents automatically extract and reason about spatiotemporally related text from complex online texts. Fine-grained relevant contexts are then incorporated with spatiotemporal data through our proposed innovative progressive fusion architecture. The rich pre-trained event prior contributes enriched insights about event-driven prediction, and hence results in a more aligned forecasting model. Evaluated on a dataset constructed through our pipeline, SeMob achieves maximal reductions of 13.92% in MAE and 11.12% in RMSE compared to the spatiotemporal model. Notably, the framework exhibits pronounced superiority especially within spatiotemporal regions close to an event's location and time of occurrence.

Problem

• Predicting urban mobility changes from external events
• Integrating textual event descriptions with spatiotemporal data
• Improving forecasting accuracy near event locations and times

Innovation

• LLM-powered semantic synthesis pipeline for mobility prediction
• Multi-agent framework extracts and reasons about spatiotemporal text
• Progressive fusion architecture integrates textual contexts with spatiotemporal data