From Narrative to Action: A Hierarchical Large Language Model Agent Framework for Interpretable and Adaptive Human Mobility Behavior
Accepted by Engineering Applications of Artificial Intelligence, the study presents a hierarchical cognitive framework in which LLM agents generate interpretable and adaptive human mobility behavior across multiple decision levels.
We are pleased to announce that our paper, “From Narrative to Action: A Hierarchical Large Language Model Agent Framework for Interpretable and Adaptive Human Mobility Behavior,” has been accepted by Engineering Applications of Artificial Intelligence (EAAI).
The study proposes a hierarchical cognitive framework for interpretable and adaptive human mobility generation. An LLM agent transforms survey attributes into narrative and structured diaries at the macro level, reflects on executed and pending activities at the meso level, and makes location and travel-mode choices at the micro level. By linking memory, reflection, and context-aware decisions, the framework offers a transparent way to simulate how daily mobility plans are formed and adapted.