Fully Dynamic Materialisation Maintenance (FDMM)
FDMM is a joint project between the University of Ulm and the DLR Institute for AI Safety and Security.
This cooperation-project focuses on a key question in computer science: When data is constantly changing — for example, due to delays or disruptions in the transport network — it is not necessary to recalculate the derived logical consequences from scratch. Instead, only the actual changes can be efficiently updated. The University of Ulm is developing algorithms to address this challenge.
Contribution from the Institute for AI Safety and Security
The Institute is providing the technical foundation for this project in the form of transport ontologies. These are structured knowledge models that define the logical interrelationships between transport events. For example, they show which routes and connections are affected when a piece of infrastructure fails. These ontologies are linked to open data, such as OpenStreetMap (OSM) and GTFS timetable data, thereby enriching them with realistic transport information.
These enriched datasets form the basis for the evaluation conducted at the University of Ulm. The data is used to verify that the developed algorithms correctly and efficiently calculate the logical consequences, i.e. implicit facts derived from explicit facts.
Ontologies are a form of data description that express relationships between data using propositional logic. Using a ‘reasoner’, implicit facts can be derived from explicit facts. Therefore, ontologies can be considered a form of symbolic artificial intelligence.
PhD projects within the FDMM programme are directly related to the work of the Institute, as ontologies enhance the security of AI. The results can be reused in both internal and external projects.
Institutes and facilities involved (DLR & external)
- DLR Institute for AI Safety and Security
- University of Ulm