Determining the best way to publish knowledge graphs is not trivial: some prefer very specific APIs that can only answer well-standardized questions, while others prefer to work with a data dump to answer any question on their own machine. Our approach is slightly different: we research Web APIs where everyone can ask any question by fetchin the right fragments of the data just in time.

Interested? Get in touch with Pieter to learn more!

Knowledge graphs are published on the Web in different ways. In order to extract facts from these knowledge graphs, intelligent querying algorithms are required that can handle the heterogeneity in terms of knowledge graph publishing interfaces. Furthermore, since the whole Web can be seen as a decentralized knowledge graph, we need techniques that can handle federated querying over large numbers of sources. Instead of requiring all of this data to be indexed beforehand, our approach involves exploiting the decentralized nature of the Web, and moving querying intelligence from server to client.

Interested? Get in touch with Ruben to learn more!

Data in knowledge graphs can be represented in different ways using differen vocabularies or simply different constructs. Facts often imply other facts. If we for example know that Bart is the son of Homer and that Homer is male, we can derive that Homer is the father of Bart. This generation of new knowlege based on existing knowledge is called reasoning. Reasoning can be performed based on different techniques and approaches. In our lab we mostly focus on rule-based reasoning. That means, possible kinds of derivations are expressed in logical rules which can be applied subsequently in order to gain new insides about data but also to, for example, translate between one vocabulary to another.

Interested? Get in touch with Ruben to learn more!

Determining the best way to generate knowledge graphs.

Interested? Get in touch with Pieter to learn more!