Semantic Trajectory Data Mining with LLM-Informed POI Classification

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, leading to constraints in efficiency and accuracy. Semantic information, such as activity types inferred from Points of Interest (POI) data, can significantly enhance the quality of trajectory mining. However, integrating these insights is challenging, as many POIs have incomplete feature information, and learning-based POI algorithms require the integrity of datasets to do the classification. In this paper, we introduce a novel pipeline for human travel trajectory mining, annotating GPS trajectories with POIs and visit purpose. Our approach first leverages the strong inferential and comprehension capabilities of large language models (LLMs) to link POI with activity types and then uses a Bayesian-based algorithm to infer activity for each stay point in a trajectory. In our evaluation using the OpenStreetMap POI dataset, our approach achieves a 93.4% accuracy and a 96.1% F-1 score in POI classification, and a 91.7% accuracy with a 92.3% F-1 score in activity inference.