Semi-Automatically Mapping Structured Sources into the Semantic Web

Presented at: 9th Extended Semantic Web Conference (ESWC2012)

by Craig Knoblock, Pedro Szekely, José Luis Ambite, Aman Goel, Shubham Gupta, Kristina Lerman, Parag Mallick, Maria Muslea, Mohsen Taheriyan

The Linked Open Data continues to grow rapidly, but a limitation of much of the data that is being published is the lack of a semantic description. While there are tools that help users to quickly convert a database into RDF, they do not provide a way to easily map the data into an existing ontology. This paper presents an approach that allows users to interactively map their structured sources into an existing ontology and then use that mapping to generate RDF triples. This approach automatically generates a mapping from the data source into the ontology, but since the precise mapping is sometimes ambiguous, we allow the user to interactively refine the mappings. We implemented this approach in a system called Karma, and demonstrate that the system can map sources into an ontology with minimal user interaction and efficiently generate the corresponding RDF.

Keywords: RDF generation, linked data, ontology mapping, source modeling

Resource URI on the dog food server:

Explore this resource elsewhere: