XTech 2005: XML, the Web and beyond.
One of the most common issues encountered by RDF developers is the need for some form of constraint on their data, and particulary validation. Unfortunately the RDF schema languages are (for perfectly good reasons) unsuited for this purpose. For example, property ranges are commonly misunderstood by newcomers to RDF as restricting possible values.
The need for validation is very common. Although RDF has a semi-structured data model this does not mean processors are required to be able to deal with anything thrown at them. The utility of being able to state "This is the kind of thing I can usefully process" is obvious. Some solutions have been attempted, ranging from constrained XML serialisations (as seen in RSS 1.0 and DOAP), to new schemas created specifically for the purpose (e.g. MEG).
This paper presents an alternative approach, using a minor 'corruption' of OWL processing. OWL is a language for describing and reasoning about ontologies. It requires some expertise, and specialised processors. As a consequence its use has rarely strayed beyond its roots in the description logic community. However one notable exception is the FOAF project's use of OWL: as an annotational tool. FOAF processing does not require an OWL reasoner, but uses OWL as a way to annotate the FOAF schema, indicating which property values are identifiers, and providing a means to identify groups.
We extend FOAF's attitude to OWL, and demonstrate how OWL may be used to express data structure in RDF. We present a deployed system in which a schema is used to validate incoming data, and a simple utility which creates an XForm form from the schema. We also introduce some preliminary work on creating 'plain' java objects from RDF.
Our hope is to encourage the use of OWL in a wider world of applications, as a more literate annotation than RDFS comments. As a side effect the traditional users of OWL will hopefully have more instance data to work with.
The full paper was not available at the time the proceedings were created. Unless this talk was "Late Breaking", this means the author did not submit their paper to the organizers before the start of the conference.
Please help us by encouraging the author to submit their paper as soon as possible.
Damian Steer has been developing Semantic Web applications since 1998. A number have been completed, including:Sherch - a web page scraper using Apple's Sherlock plugins.RDFAuthor - an RDF authoring tool using a graphical interface.Brownsauce - and semantic web 'browser' which attempts to make RDF human readable.FOAFFinger - a little application tying FOAF and Zeroconf (Rendezvous) together.Treehugger - an attempt to fool XPath into working over RDF graphs. Via this one can use XSLT and XQuery over with the RDF model.He has worked with ILRT and UKOLN, and HP Labs, Bristol.