Re: URIs / URLs

This seems to be a never-ending story. So, I'll just speak out my opinion that has helped me in some senses to move "forward". (I apologize in advance for using a perhaps unusual terminology.)

In the following I consider the URI-space to be the Web, and I will use the term entity to represent a resource that can exist outside the Web (e.g. people, file), and the word Web-resource to represent a resource that only exists on the Web (this is done to clarify the difference, even if I might be a bit confusing).

A Web-resource is a pair: (URI, mapping). The Web-resource is identified by the URI. The mapping (abstract or formal) is used to represent entities on the Web. Since we can't move things into the Web we use Web-resources to represent them by mapping the Web-resource to the entity.

There are two types of mapping that could occur: The mapping could be encoded in the URI string, or the mapping could be abstract. In the case of a URL (the mapping is actually to an algorithm which could be used to e.g. retrieve the entity) it is encoded in the URI string. The only mapping that could be used by a computer to retrieve the entity that the Web-resource represent, e.g. a file, is the formal one. 

A Web-resource that has a URN as its identifier (and does not use a resolution service) has an abstract mapping (i.e. in somebody's head, or on a paper) to the entity.

The only thing that is different from a Web-resource that has a URN as an identifier from one that has a URL is the mapping.

The only resources that are "on" that Web are Web-resources. A Web page is not on the Web - a Web-resource representing that page is.

Persistency is related to the mapping. A Web-resource is "considered" persistent if the entity that the Web-resource maps onto is persistent. Thus, there is absolutely no way that a URN could be more persistent - it the entity that the Web-resource maps to that should be persistent. The reason for this (URN persistency) belief is that the mapping is abstract so no one will notice if the entity it maps onto does not exist or has changed.

This view has helped me somewhat (though it might be wrong). It fits the RDF (should perhaps be Web-Resource DF) world. With RDF we attach properties to Web-resources that represent some entity. We do not attach properties to the entities that they represent (not directly anyway).

There is a bunch of loopholes in the above. A longer version exists on
Thanks for your attention!


Received on Wednesday, 11 April 2001 13:17:02 UTC