From W3C Wiki
Jump to: navigation, search

Back to AwwswHome

Generic Resource

The generic resource model has prominence because TimBL has advocated for it forcefully and persistently. So it is worth some effort to attempt to figure out the nature and implications of this model. JAR thinks of this as a reverse engineering problem.

Tim also applies the term "information resource" to this class but JAR finds this confusing as AWWW does not define "information resource" in the same way. Perhaps Tim's intent is that the two classes should coincide, but they are defined using different words. In addition, other authorities on the subject, and Tim at other times, have used information resource to mean yet other things. So let's stick with "generic resource".

This exercise is not an attempt to understand what "information resource" should mean. It is merely an attempt to understand one particular model.

The question is, how would one define a class having all the characteristics listed below?

Here is some 2004 email from Tim explaining what he means: (courtesy Sean Palmer)

Examples of generic resources:

  1. The Bible
  2. The Bible, King James Version
  3. The Bible, KJV, in English
  4. A particular ASCII rendering of the KJV Bible in English
  5. The US constitution
  6. The referent of
  7. 14 billion other web pages (which ones?)

Not generic resources:

  1. Anything that has mass or location (e.g. a professor, a physical book)
  2. A corporation
  3. A server
  4. A property (in the RDF sense)
  5. A number

Not clear (check with Tim):

  1. A string
  2. A file on a disk (???)
  3. Would a URI that can be used with POST (e.g. via <action>), but not GET, name an IR?
  4. The referent of the URI data:text/plain,n_body_problem
  5. The referent of the URI
  6. The referent of
  7. An XML namespace (as opposed to namespace document)
  8. An RDF graph (as in SPARQL graph <foo> {...} )
  9. A "named graph" (per proposal)
  10. An OWL ontology (as opposed to ontology document)
  11. Were there generic resources before the invention of electricity? If so did they have wa-representations at the time?

Might participate in relationships:

  1. Dublin Core properties (e.g. dc:creator)
  2. cc:license
  3. genont
  4. had or has as a representation (at some time, for some request params)

Has subclasses:

  1. foaf:Document ??
  2. genont:FixedResource


  1. One GR can be "served" at two distinct URIs - perhaps even differently at the two URIs (as long as the wa-representations you get from each are wa-representations of the resource)
  2. Two GRs can have identical wa-representations under all circumstances, and still be distinct (the time sheets example)

Other questions:

  1. Relation to Fielding and Taylor REST model
  2. Relation to Booth's FTRR
  3. Relation to FRBR
  4. Relation to Information Artifact Ontology (IAO)
  5. Examples of GRs that are not documents?
  6. Examples of web pages that are not GRs?
  7. How to explain PUT and POST if the GR is not agent