# Towards Next-Generation Urban Decision Support Systems through AI-Powered Construction of Scientific Ontology using Large Language Models -- A Case in Optimizing Intermodal Freight Transportation

**Authors:** Jose Tupayachi, Haowen Xu, Olufemi A. Omitaomu, Mustafa Can Camur, Aliza Sharmin, Xueping Li  
**arXiv:** [2405.19255](https://arxiv.org/abs/2405.19255)  
**Date:** 2024-05-29  
**Summary page:** https://www.summarizepaper.com/en/paper/2405.19255v3/

## Key points

- Integration of AI models into optimization systems is increasingly prevalent in urban and environmental management
- Deep understanding of domain science and informatics expertise is required for tackling complex issues
- Study explores potential of using pre-trained Large Language Models (LLMs) to enhance urban decision support systems
- Innovative approach involves utilizing ChatGPT API as central reasoning core within an integrated workflow
- Methodology automates creation of scenario-based ontology by analyzing research articles and technical manuals
- Output manifests as knowledge graphs formulated in ontology languages such as OWL, RDF, and SPARQL
- AI-generated ontologies enhance development of urban decision support systems through improved data modeling, integration of datasets, simulation models, decision-making metrics, and workflows
- Feasibility validated through comparative analysis with Pizza Ontology commonly used in ontology software tutorials
- Real-world case study focuses on optimizing multi-modal freight transportation system
- Methodology showcases how it can revolutionize urban planning strategies

## Summary

The integration of Artificial Intelligence (AI) models into optimization systems is becoming increasingly prevalent in the rapidly evolving landscape of urban and environmental management. However, tackling complex issues in this domain often requires a deep understanding of domain science and informatics expertise to derive data-driven solutions for informed decision-making. In light of this, a study conducted by Jose Tupayachi, Haowen Xu, Olufemi A. Omitaomu, Mustafa Can Camur, Aliza Sharmin, and Xueping Li explores the potential of harnessing pre-trained Large Language Models (LLMs) to enhance urban decision support systems. The researchers delve into utilizing the ChatGPT API as a central reasoning core within an integrated workflow that incorporates natural language processing, methontology-based prompt tuning, and transformers. This innovative approach automates the creation of scenario-based ontology by analyzing existing research articles and technical manuals related to urban datasets and simulations. The output of this methodology manifests as knowledge graphs formulated in widely adopted ontology languages such as OWL, RDF, and SPARQL. By leveraging these AI-generated ontologies, the development of urban decision support systems is significantly enhanced through improved data and metadata modeling, seamless integration of complex datasets, coupling multi-domain simulation models, and formulating effective decision-making metrics and workflows. To validate the feasibility of their methodology,a comparative analysis is conducted juxtaposing their AI-generated ontology with the well-known Pizza Ontology commonly used in tutorials for popular ontology software like protégé. The study culminates in a real-world case study focusing on optimizing the intricate urban system of multi-modal freight transportation. By generating anthologies comprising various domain data and simulations tailored to support informed decision-making processes,the researchers showcase how their methodology can revolutionize urban planning strategies. This comprehensive exploration sets the stage for next-generation urban decision support systems empowered by AI-powered construction of scientific ontology using Large Language Models.