RDF Identifier Terminology


$Revision: 1.1 $ $Date: 2002/07/31 20:44:17 $

Just some thoughts.


It seems like every RDF API defines a class (or small class hierarchy) for identifiers of things described in RDF. But the terminology varies a lot.....

What to Name this Class?

This is a difficult decision. We're handling information about lots of things. Those things are identified by instances of this class.

The first question is: should we call this class Thing-Which-Is-Pointed-To or Pointer-To-Thing? The decision here depends, in part, on how much you think about variables (a different sort of identifier), triples-centric vs node-centric APIs, having multiple pointers to a thing, bNodes, and literals.

Thing-Itself Names

These are great for blurring the line between normal objects in the program and things on the semantic web. Objects can be semweb objects, instead of being identified by identifiers.

Pointer-To-Thing Names

These are great for staying explicit about the referencing and derefencing which is really going on. They do make some simple operations more complicated, because you need the explicit dereference. Do these things need to be variables, too? Or is that some other class? Maybe "Reference" allows both kinds, or something?


Personally I lean towards Ident, if I have to pick one.

Sandro Hawke
First: 2002-03-29; This: $Id: Overview.html,v 1.1 2002/07/31 20:44:17 sandro Exp $