Tag
This paper presents a framework for integrating semantic knowledge from general-purpose knowledge graphs into spatio-temporal traffic forecasting models to improve prediction accuracy and interpretability by providing contextual information beyond physical connectivity.