Semantic Web Meets ILP: Unconsumated Love, or No Love Lost?

Abstract

Even at first glance, ILP and the Semantic Web have much in common. Both are about large volumes of data, both make use of background knowledge , and both use computationally tractable forms of logic. Nevertheless, the actual intersection of the two research areas is very small. In this talk, I will first give a birds eye overview of the Semantic Web programme (its goals, its methods, its achievements to date, and the important open challenges). I will then consider the most obvious use of ILP for the Semantic Web: could ILP be used to learn the ontologies that are such a crucial ingredient in the Semantic Web story? However, as with any result from Machine Learning, such ontologies will not be fully correct and complete. This will require that, from its side, the Semantic Web community must learn how to deal with such partially incomplete and incorrect ontologies. I will present the most recent work in this direction, the efforts to build LarKC, the Large Knowledge, a platform for infinitely scaleable distributed incomplete Semantic Web reasoning. Could the Large Knowledge Collider be the place where ILP and the Semantic Web finally meet?