Data Profiling for Semantic Web Data

摘要

Lots of RDF data have been published in the Semantic Web. For human users it is often rather difficult to get the big picture of a large RDF data exposed by Semantic Web applications. How to understand a large and unfamiliar RDF data becomes very important when the data schema is absent or different schemas are mixed. In this paper we describe a tool which can induce the actual schema, gather corresponding statistics, and present a UML-based visualization for the RDF data sources like SPARQL endpoints and RDF dumps. Experimental results, using six data sets from the Linked Data cloud, compare our approach and ExpLOD. The evaluations show that our approach is more efficient than ExpLOD.