Knowledge-based AI Application in Transportation

Project Overview

Human travel trajectory mining is crucial for transportation systems, enhancing route optimization, traffic management, and the study of human travel patterns. While previous studies have primarily focused on spatial-temporal information, the integration of semantic data has been limited. We introduce a novel pipeline for human travel trajectory mining, annotating GPS trajectories with POIs and visit purpose.

System Flow Diagram — First application of LLM to POI classification with semantic information inference

Key Contributions

A novel data mining framework integrating LLM-based POI classification with probabilistic activity inference algorithm, bridging spatial-temporal and natural language analyses in trajectory mining.
Framework adaptability across various regional POI datasets without additional training, effectively handling incomplete data.
First application of LLM to POI classification, achieving precise point-level inference through semantic information.

Results

93.4% accuracy and 96.1% F-1 score in POI classification
91.7% accuracy with 92.3% F-1 score in activity inference

POI Analysis — Framework adaptability demonstration across different regional POI datasets. Example of POIs and classification results from (a) LA county and (b) Egypt.

Trajectory Analysis — Examples of semantic trajectories post-data mining: (a) Activity-annotated stay points in LA. (b) Activity timelines.