Re: Soliciting feedback for Open GUID

[Peter Ansell]

2008/9/30 Jason Borro <ja...@openguid.net>

Hello,

I recently released http://openguid.net in an attempt to establish a global subject identifier for the Semantic Web.

Open GUID consists of two things:

1. A lightweight ontology that links RDF resources based on human perceived identity.
2. A set of services to query and manipulate subjects.

Please forward any criticisms or suggestions.

Thanks,
Jason Borro

From my view the overall idea is to specify a common meaning with a GUID (which can be anything, but in this case it is a unique value that is unrelated to the subject matter).

How do you plan to cope with the idea that meaning is derived at least in part by how something is used?

What vocabulary are you planning to use to represent the "tag(ged) relations"?

How do you plan to link between a deprecated term and two alternative terms which form the disambiguation when people decide that a particular term is vague?

What has been discussed about the administrative overhead when the system is opened up to the community to edit and finetune the seeded sets of guids?

Why aren't Wikipedia (and hence DBpedia) URL's suitable for your goals? You could take advantage of the already existing community system more easily than duplicating it IMO.

Are GUIDs what people want considering the total lack of recognisability prior to resolving it?

What does this project offer over the existing UMBEL/OpenCyc/DBpedia/Wikipedia general knowledge URI creation implementations?

What do you mean by query and manipulate with respect to the services? Will the project provide a SPARQL endpoint that people can perform arbitrary queries on?

(Just a few questions to think about ;-) )

Cheers,

Peter

[ja...@openguid.net]

Hi Peter, a few answers below:

> From my view the overall idea is to specify a common meaning with a GUID
> (which can be anything, but in this case it is a unique value that is
> unrelated to the subject matter).
>
> How do you plan to cope with the idea that meaning is derived at least in
> part by how something is used?

One of the principals of Open GUID is to say nothing meaningful about
a subject, just identify as concisely as possible. Classes and
instances across ontologies can then be linked without implying rigid
semantics. Reasoners and data miners can take a local view when
appropriate, but be able to change views and still keep context.

>
> What vocabulary are you planning to use to represent the "tag(ged)
> relations"?

Tags are just for navigation, they have no semantics.

>
> How do you plan to link between a deprecated term and two alternative terms
> which form the disambiguation when people decide that a particular term is
> vague?

This should be less of an issue than it is with wikipedia because
terms are not used in the identifier.
But if an initial description is vague, it's description can be
changed to point out this issue, and tags can be added to navigate to
more specific concepts.
Any identical relations to the original GUID can be moved to more
specific ones if appropriate.

>
> What has been discussed about the administrative overhead when the system is
> opened up to the community to edit and finetune the seeded sets of guids?

A few ideas have been discussed about the democratic process of
handling changes. Nothing has been concluded; no major roadblocks
have surfaced.

>
> Why aren't Wikipedia (and hence DBpedia) URL's suitable for your goals? You
> could take advantage of the already existing community system more easily
> than duplicating it IMO.

Wikipedia has notability requirements for it's topics. Even I can
have an Open GUID :)

>
> Are GUIDs what people want considering the total lack of recognisability
> prior to resolving it?

I agree it is slightly suboptimal for people who like to read bedtime
stories in RDF. Computer should have no problem. Also, most existing
linked URIs are unrecognizable to the non-english speaking world.
Tools will help.

>
> What does this project offer over the existing
> UMBEL/OpenCyc/DBpedia/Wikipedia general knowledge URI creation
> implementations?

Unlimited scope. No semantics. Just pure linkage.

>
> What do you mean by query and manipulate with respect to the services? Will
> the project provide a SPARQL endpoint that people can perform arbitrary
> queries on?

OpenGUID is not a fact repository so no SPARQL. The services will let
you find and edit existing GUIDs in order to link ontologies or
content. Editing is limited to adding more keywords, fixing
descriptions (while not changing identity), adding tags, and
registering identical relationships.

>
> (Just a few questions to think about ;-) )

Thanks for the challenge!

Jason