W3C home > Mailing lists > Public > public-esw-thes@w3.org > November 2009

Re: [Dbpedia-discussion] Using DBpedia resources as skos:Concepts?

From: Rick Murphy <rick@rickmurphy.org>
Date: Thu, 05 Nov 2009 23:40:58 -0500
Message-ID: <4AF3A8DA.2050707@rickmurphy.org>
CC: SKOS <public-esw-thes@w3.org>
John, Pat & All:

Here's a useful diagram that I use to illustrate the issues below.


I also published a piece recently that I hope you find of some value.

Here's a reference to a version slightly revised from the original.



John Graybeal wrote:
> On Nov 5, 2009, at 1152, Pat Hayes wrote:
>> There seems to be a problem in the very heart of the SKOS design. Is 
>> it talking about things, or about concepts of things? Y'all really 
>> need to get this straight before proceeding.
> Let me use this as a way to propose an answer to the question, and get a 
> bigger answer to a related point that's been bothering me.
> Let's say SKOS is talking about names for things.  This is once removed 
> from the things, obviously -- "Mrs Obama" is a name for the person, not 
> the person herself -- but not the concept.  And "person" is a name for 
> the type of thing of which "Mrs Obama" is a name for an instance.
> (There is a metaphysical discussion about the correspondence between a 
> 'name' and a 'concept', but I'd like to avoid it, as I don't think it's 
> central to your question.)

I think edge between the interpretant and the sign speaks to this 
distinction without much of the metaphysics. Assume the interpretant is 
a conception, or concept. The sign evokes the interpretant and 
conversely interpretant signifies or connotes the sign.

> So now I have said I have a name of a type of thing, and the name is 
> "person".  Can I say anything semantically that connects this name, in a 
> constructive and semantically-friendly and Pat Hayes-agreeable way, to 
> the semantic web resource that represents the concept of a human being?  
> (Pick your favorite semantic web resource for representing that concept, 
> OK?)

Pat: Since your Blogic talk, I've dusted off my copy of the CL standard.

I see that CLIF differentiates interpreted names and interpretable 
names, but CL semantics says "all names are interpreted in the same way 
... that is why there is only a single interpretation mapping that 
applies to all names regardless of their syntactic role."

If a universe of discourse in CLIF identifies both interpreted and 
interpretable names wouldn't the interpretation have to recognize these 
two distinct roles?

> In other words, what is the correct way to connect the semantic web to 
> all of these thesaurus and dictionary entities?   (Which I am here 
> calling 'names', but elsewhere are called 'terms' and other things, to 
> the resource in the semantic web that you've picked above (which for 
> many people is associated with a 'concept', but I am avoiding that term, 
> hoping that 'resource in the semantic web' is clearer).)

I think the Semantic Web needs to be more specific than the general term 
resource. The names chosen by the community need to differentiate both 
role according to the triangle and extent they cover: machine, external 
world and consciousness.

While I was at OWLED this year I had a chance to ask Peter 
Patel-Schneider for some history on how the description logic community 
came to use the term concept to denote what the OO community described 
as a class. The thinking at the time was simply to connote more of a 
logical image for the work of the community.

> I infer that you would like to say they are not concepts, which is fine, 
> but then can we all agree on a paradigm for making the connection?

I suggest an extension of good old fashioned model theory that (in the 
language of CL) differentiates interpreted names and interpretable names 
in a universe of discourse with mappings to a universe of reference that 
differentiates information objects and objects.

> John
> John Graybeal   <mailto:jgraybeal@ucsd.edu>
> Ocean Observatories Initiative Cyberinfrastructure Project: 
> http://ci.oceanobservatories.org
> Marine Metadata Interoperability Project: http://marinemetadata.org


cell: 703-201-9129
web:  http://www.rickmurphy.org
blog: http://phaneron.rickmurphy.org
Received on Friday, 6 November 2009 04:44:44 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:45:59 UTC