Analytics for complex event recognition

Specification, use cases and prototype implementation of machine learning tools for learning complex events from data by using semantic unsupervised labelling for spatiotemporal data. Representational learning will be investigated to achieve a latent representation of spatiotemporal traces and users’/drivers’ behavioural characteristics. The developed machine learning algorithms will be capable of online learning complex event patterns, in order to effectively handle the volume and velocity of the data. The core capabilities of the tools and the application of the algorithms developed, will be demonstrated through an open research data pilot with the use of open data sources. The specification report will include several use cases to allow the use and extension of the tools/algoritmes by development teams across Europe.