The rapid development of mobile and sensor technologies is leading to substantial changes in the transportation industry, with the research paradigm moving from physical mechanism-based models to data-driven models. Traffic sensors, such as loop detectors, probes, cameras on road networks, smart cards, and quick response codes in public transport, and global positioning system (GPS), Bluetooth, and Wi-Fi devices, are generating massive datasets, and providing unprecedented opportunities for the transportation industry to understand diverse travel behaviours and improve the traffic modelling process.

However, due to the nonlinearity and heterogeneity of big data, new challenges are arising in the data-driven analysis, mining, and application in the transportation industry. To overcome these challenges, graph-based machine learning is gaining increasing attention as a powerful tool. For instance, urban road networks can be directly modelled as graphs, based on which a series of neural networks and their variants can be employed for urban traffic analytics. Recently, pilot studies on traffic analysis using heterogeneous graphs were carried out. A heterogeneous graph is a graph with multiple types of nodes and links. Compared with homogeneous graphs, heterogeneous graphs have two advantages. Firstly, heterogeneous graphs are an effective tool for fusing information, that is, not only from different types of objects and their interactions but also from heterogeneous data sources. Secondly, the coexistence of multiple types of objects and relationships in heterogeneous graphs, which contain rich structural and semantic information, provides a new way to discover hidden traffic patterns in a more precise and interpretable way. Although some efforts have been made to explore graph-based methods in mining transportation data, it is still necessary to fully explore the benefits of these methods for ITS applications (e.g. traffic state analysis, traffic demand prediction, anomaly detection, traffic signal control).

The aim of this Special Issue is to collate original research and surveys on the use of graph-based models in big data analysis and mining within transportation systems. Submissions should not focus on freight or supply chain logistics.

Potential topics include but are not limited to the following:

• Graph-based deep learning in traffic management and operation solutions
• Graph-based analysis of the resilience of transportation systems
• Graph-theoretic matching, classifying and assessing of transportation network
• Data mining theory, technologies and applications in ITS based on heterogeneous graphs
• Novel network representations and graph extraction of transportation systems
• Security, integrity and privacy solutions for ITS using graph-based deep learning
• Collaborative decision-making for improving the efficiency of road networks using multiple agent graphs
• Multi-scale mobility feature analysis and prediction based on graphical modelling
• Knowledge graph theory, technologies and applications for ITS