Integrating relational data into the semantic web

By two guys who know what they're talking about.

I was sorry to miss the Semantic Web Technologies conference, but I had a very interesting time yesterday giving a talk at the New Horizons in Teaching and Research conference at the University of Virginia (so nice to go to a conference 25 minutes from my house) on Semantic Web technologies, RDF and OWL (and Linked Data). It was an audience of professors and researchers from a wide variety of disciplines who were all very open to hearing about how these standards could let them store metadata that helped them them get more out of their data.

If you follow the second link in the preceding paragraph, you'll see that "(and Linked Data)" wasn't part of the original title. I added it because I felt that it was important for them to hear about the possibilities of linking publicly available data sources with each other and with local resources in order to find new connections and patterns, whether they were interested in storing the semantics of the various terms or not. They were surprisingly receptive to the basic ideas of RDF—some audiences keep coming back to "why use a URI for something that isn't a web page where you can send your browser?"—but I still emphasized the increasing availability of non-RDF data to SPARQL queries and what this means to the growing collection of linked open data.

As an example, I showed my demo for using D2RQ to integrate two different relational databases, but that was more of a toy example. At the LinkedData Planet conference in a few weeks, we're going to hear Jim Melton and Ashok Malhotra's talk on Integrating Relational Data into the Semantic Web. They're both employed by Oracle, but their reputation for developing and implementing important standards such as SQL and XQuery goes way beyond their Oracle work. Here is their abstract:

Invaluable data is stored in relational databases, but a growing fraction is being created in non-traditional forms such as spreadsheets and PDF files. Integrating relational and XML data is well-understood with widely-accepted solutions. Relational data must be similarly integrated with data in other forms. A promising approach is to translate the underlying data into a common format (such as RDF) and the creation of a "semantic cover" atop the data in the form of an ontology (perhaps using OWL). Classes and subclasses of the ontology are mapped into queries on the underlying data. The ontology can then be queried using SPARQL and the SPARQL queries translated to queries on the underlying data.

The two of them together make for a lot of firepower on the topic, and with so much of the world's data stored in relational databases (particularly in their employer's products), their insights on how to open up such data to Linked Data tools such as SPARQL will help more people get access to more data to mix, match, and use for interesting new applications.

This is only one of many great talks we'll have at the Linked Data Planet conference; come join us!