Semantic Overflow

Question

1

I'm hoping to define what I think is a useful expression of a conference program in RDF (based on the ISWC "dogfood" work on their own ontology). What I'm thinking of doing is describing it as much like XML as possible to keep from freaking out people who just don't care about the underlying tech.

What my aim at is to build some basic expectations which other conferences and events can copy, and will want to because there's some tools which can consume data in that format. My idea is to define a simple core structure, which it must have, plus some properties each object can have, and then it can be extended any which way, but those triples will be ignored by tools expecting my defined structure.

I'm thinking of something like RSS but for describing the program of structured events. Is there any wisdom in the community about such things, yet?

flag	asked Dec 2 at 2:11 Christopher Gutteridge 529●8

Bill Barnhill · Answer 1 · 2009-12-03 14:48:32Z UTC

1

I'd suggest taking a look at AtomOWL. I've found it to be simpler to understand than RSS and more extensible. There are also fewer versions of it around. AFAIK there's only one AtomOWL, where with RSS there is RSS 2.0, RSS 1.0, etc.

answered Dec 3 at 14:48

Bill Barnhill
320●4

	You can also check out the following for a good comparison of different RSS versions: rss-net.sourceforge.net/documents/… – Bill Barnhill Dec 3 at 14:49
	I meant a standard XML serialisation of a set of triples, like RSS. Not a format for expressing a list of resources. – Christopher Gutteridge Dec 4 at 8:23
	Note, RSS1.0 IS RDF+XML, but then a standardised way of laying it out so non RDF aware tools can just use it as an XML file. – Christopher Gutteridge Dec 9 at 9:16

billroberts · Answer 2 · 2009-12-08 09:31:22Z UTC

I understand your wish to describe your system in common familiar technologies, but I think it will ultimately be a bad idea to invent your own syntax and you should just bite the bullet and use OWL and RDF. To help non-specialists understand it, provide lots of diagrams and explanations and examples.

If you try to make your own serialisation of your RDF model, you will probably end up re-inventing something similar to an existing format, but it won't be a standard and will probably have some inconsistencies that have been ironed out of the standard formats through years of reviews and comments.

(I know your main question was about how to communicate the data model rather than how to design it, but if you haven't already you might take a look at the 'Scholarly Event Ontology' (SEDE) http://www.eventography.org/index.php/Scholarly_Event_Ontology_%28SEDE%29 - maybe some good ideas there).

Rob Vesse · Answer 3 · 2009-12-04 10:03:39Z UTC

If you're looking for an XML serialization that is easy to read and understand how about using TriX which has the advantage of being able to encode multiple graphs in one file: http://swdev.nokia.com/trix/trix.html

e.g.

<TriX xmlns="...">
<graph>
  <triple>
    <uri>http://example.org/subject</uri>
    <uri>http://example.org/predicate</uri>
    <plainLiteral>object</plainLiteral>
  </triple>
</graph>

Sorry this isn't a longer example but it's hard to type XML on a phone even with a full keyboard

Also TriX has an interesting extensibility mechanism based around XSLT so you could use a domain specific XML serialization that looked nothing like this and just transform it to TriX using a stylehseet at which point it can be parsed by any TriX parser. in fact a good TriX parser should apply any stylesheets specified in the XML before actually attempting to parse which means you'd just have to add the appropriate stylesheet processing instruction to the file.

How do I communicate an RDF structure to non-linked data developers?

3 Answers