A new modelling framework over temporal graphs for collaborative mobility recommendation systems

Over the years, collaborative mobility proved to be an important but challenging component of the smart cities paradigm. One of the biggest challenges in the smart mobility domain is the use of data science as an enabler for the implementation of large scale transportation sharing solutions. In particular, the next generation of Intelligent Transportation Systems (ITS) requires the combination of artificial intelligence and discrete simulations when exploring the effects of what-if decisions in complex scenarios with millions of users. In this paper, we address this challenge by presenting an innovative data modelling framework that can be used for ITS related problems. We demonstrate that the use of graphs and time series in multi-dimensional data models can satisfy the requirements of descriptive and predictive analytics in real-world case studies with massive amounts of continuously changing data. The features of the framework are explained in a case study of a complex collaborative mobility system that combines carpooling, carsharing and shared parking. The performance of the framework is tested with a large-scale dataset, performing machine learning tasks and interactive realtime data visualization. The outcome is a fast, efficient and complete architecture that can be easily deployed, tested and used for research as well in an industrial environment.