Practical RDF

Volume 6, Issue 128; 11 Dec 2003

I pitched a Town Hall for XML 2003 in the hopes that I could get a handful of folks up in front of an audience to talk about practical things that you can do today with RDF. I think I succeeded.

I pitched a Practical RDFThis is also the name of Shelly Powers excellent RDF book. Apologies for any confusion I’ve caused. Town Hall for XML 2003 in Philadelphia in the hopes that I could get a handful of folks up in front of an audience to talk about practical things that you can do today with RDF.

I succeeded better than I could have hoped. I talked Dan Brickley , Dan Connolly , Edd Dumbill , and Eric Miller into the exercise. Unfortunately, Eric was unable to come to the conference.

My motivation for this town hall was to encourage people to think about RDF as a tool for solving problems outside the “semantic web” vision. I don’t know if I believe in the semantic web vision or not, but I don’t have to believe to use the tools.

There are two features of RDF that I find particularly practical:

Aggregation. Metadata comes from essays, photographs, my Palm, and other files. Expressing the metadata in RDF lets me pull it all together and address it uniformly.

Inference. I hate duplication of information. For example, the fact that Dan was born and has not yet died, implies that he has a birthday. Last tuesday, most recently, in fact. Recording Dan’s birthday in one place and the repeating appointment to remind me of his birthday in another place is wrong. Instead, I use rules:

# Handle birthdays and births and deaths
{ :p a palm:Contact .
  :p p:birthday :o }
 log:implies { :o a palm:Appointment ;
                    palm:category xdb:_10 ;
                    palm:description "Birthday";
                    palm:datebk4 [
                       rdf:type palm:DateBk4 ;
                       palm:category "10" ;
                       palm:icon "10" ];
                    palm:repeat [
                       rdf:type palm:Repeat ;
                       palm:frequency "1";
                       palm:type "Yearly" ] } .

{ :p a palm:Contact .
  :p p:birthday :r .
  :p p:born :o }
 log:implies { :r palm:begin-date :o } .

{ :p a palm:Contact .
  :p p:birthday :r .
  :r palm:repeat :t .
  :p p:died :o }
 log:implies { :t palm:end :o } .

(Yes, I admit, this is a trivial, anal-retentive example. But the principal applies none-the-less.)

My list of practical tools includes the following:

cwm: I use cwm (“closed world machine” ) to aggregate and apply inference rules to my data.
jpegrdf: I use jpegrdf to store metadata about photographs inside the photographs.
Jena: Jena is a Java library for accessing RDF graphis.
XSLT/RDFTwig: I use RDFTwig, a set of extension XSLT extension functions, to make it easier to style RDF data for presentation.
My own Palm stuff: As I’ve said before, my Palm Pilot is my “personal information appliance”. I store as much data in there as I can. I have some tools for syncing that data with my computer and building RDF from it. I’ll get around to republishing the conduits eventually.