Rule-based data integration with ontologies 13/12/17 in KB L.T 1.3 at 14:00
Published: Friday, 01 December 2017
Rule-based data integration with ontologies by Dr Evgeij Thorstensen on 13th December 2017 in Kilburn L.T 1.3 at 2pm with tea and biscuits in the Staff Common room from 3pm for informal discussions
Rule-based data integration with ontologies Data integration is concerned with providing improved access to data that resides in hard to understand databases when moving or refactoring the existing data sources is infeasible. Data sources can be linked to the new view of the data by defining logical rules. This approach allows for the translation of queries from the view to the sources, as well as reasoning about the system as a whole. While data sources are commonly SQL databases, the target view can be something else. In particular, we have worked on using description logic ontologies to build such views. This approach is known as ontology-based data access (OBDA), and allows for scalable automated reasoning over data. In this talk I will describe the general OBDA framework, and focus on the rules linking source databases to the ontology concepts and roles. Maintaining and debugging such rules can be treated as a reasoning problem. I will discuss recent results relating to checking consistency and redundancy for such rules, and outline the difficulties we encountered in this area.