Revolutionizing Trajectory Mining with Large Language Models: A New Paradigm in Spatial-Temporal Data Analysis

Country
New Zealand
Year
2024

Principal Investigator: Dr. Zhiqian Chen

Led by Dr. Zhiqian Chen, this project explored the intersection of large language models (LLMs) and spatial-temporal data, aiming to revolutionize trajectory mining through advanced AI techniques. The team developed a benchmark suite to evaluate LLMs’ ability to model long-range dependencies in graph dynamics, culminating in the presentation of their work, FlowGPT, at SouthNLP 2024 and the publication of a co-authored paper at IJCAI 2024, one of the premier conferences in artificial intelligence.

Image
Headshot of Dr. Chen Zhiqian
Dr.  Chen Zhiqian

The project also enhanced the open-source library xflow.network, and laid the groundwork for future international collaboration. The project successfully advanced foundational research and proposal development, and not only deepened understanding of spatial-graph relationships but also provided valuable insights into the complexities of building international research teams. It has positioned Dr. Chen’s lab to pursue larger-scale funding and collaborative opportunities in AI-driven spatial analysis and threat detection.


Project Impact

Publications and Presentations

Yaya Zhao, Kaiqi Zhao, Zhiqian Chen, Yuanyuan Zhang, Yalei Du, and Xiaoling Lu. 2024. A graph-based representation framework for trajectory recovery via spatiotemporal interval-informed seq2seq. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI '24). Article 286, 2588–2597. https://doi.org/10.24963/ijcai.2024/286