Advancing Transportation Modeling with Large Language Models : A Case Study in the Swedish Context

Sammanfattning

The complexity of comprehensive transport models often creates a significant barrier between sophisticated analytical tools and their practical application by users across various domains. This study addresses this research gap by exploring the potential of Large Language Model (LLM) technology to enhance the accessibility and usability of complex transportation models. Using the Swedish SAMPERS 4 as a case study, we introduce LLM4SAMPERS4, a Retrieval Augmented Generation (RAG) pipeline designed to bridge the divide between intricate model documentation and user comprehension. The research methodology combines qualitative user interviews with a technical implementation, leveraging SAMPERS 4 user manuals as a knowledge base. Evaluation through representative queries demonstrates LLM4SAMPERS4's capacity to provide domain-specific, contextually relevant assistance, outperforming generalpurpose LLM models in this specialized context. While showing promise in integrating multimodal information and offering targeted guidance, the study also identifies areas for future refinement. This research contributes to the growing body of literature on AI applications in transportation modeling and planning, suggesting new approaches to interacting with and utilizing complex analytical tools.