AI for Data Management

The research group AI for Data Management is led by Diego Calvanese, a world-leading expert in artificial intelligence for data management, who received the AAAI Classic Paper Award 2021, which testifies the strong and long-lasting impact of his research in Artificial Intelligence and Computer Science. 

Robust framework

The research group addresses both fundamental and applied challenges in knowledge representation and reasoning in AI, using logic-based formalisms, in particular description logic and the associated inference capabilities. This allows us to take into account semantic information, which is important for better understanding data and their meaning, and for preparing data for further processing and analysis through machine learning and artificial intelligence techniques. Overall, we seek to develop a robust framework for modeling and analysis of complex systems.

Diego Calvanese is Wallenberg Visiting Professor in Artificial Intelligence for Data Management at the Department of Computing Science, Umeå University and Professor at the Free University of Bozen-Bolzano in Italy. Calvanese received his PhD from Sapienza University in 1996. He is the author of more than 350 refereed publications, including ones in the most prestigious international journals and conferences in AI and databases, with more than 33000 citations and an H-index of 72, according to Google Scholar.

Search, find, and sort information

Central to the research is the use of virtual knowledge graphs (VKG), also known as ontology-based data access, integration, and management (OBDA/OBDI/OBDM). In a VKG system, an ontology is used to describe the different types of data and their relationships to each other. Mappings act as a “translation table” that connects the different types of data to the actual data sources. When a user formulates a request via the ontology, the VKG system translates the request into a form that can be processed directly by the respective data source. In this way, the VKG system can provide the user with access to data from multiple sources simply and efficiently. Thus, our work enables seamless access, integration, and exchange of data from different sources.

Topics currently being investigated 

• Management and coordination of specific types of data, such as geospatial features and multiple data formats, including both graph and tree-structured data. This addresses the complexity inherent in such formats;
• Representing time-bound information and reasoning and answering questions in time-bound data;
• The ability to perform updates to data sources through the provision of an ontology and a mapping, taking full account of their semantics;
• Consideration of both the structural and the dynamic aspects of data and information, and their interaction to determine the behaviour of a system;
• Definition of flexible, but rigorous, mechanisms to manage data integrity, data privacy, and personalisation of data access.