Semantic Integration of Structured Data Powered by Linked Open Data

Abstract

Recent advances in open data have resulted in vast amounts of tabular datasets containing valuable, actionable information to several stakeholders. However, information pertaining to any given entity is fragmented across several arbitrarily structured tables. There is a pressing need for semantic integration of such disparate datasets to enable deeper forms of inference and intelligence. This task is challenging because not only such datasets have no overarching schematic framework, there is also no overarching thematic framework. The datasets need not be about any one specific topic or theme. Hence, there is no one specific ontology onto which the datasets can be mapped. In this work we address the issue of mapping arbitrarily structured tabular data to one or more existing ontologies from the Linked Open Data Cloud (LOD) or abducing a new ontology around subsets of such tables. The overall objectives of this work called "Inferencing in the Large" aims to go further than this, to enrich mapped ontologies with inferencing rules and enable the use of semantic reasoners.