W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > December 2002

Re: pruning the semantics document (and "meaningless terms")

From: Graham Klyne <GK@NineByNine.org>
Date: Thu, 12 Dec 2002 09:57:17 +0000
Message-Id: <5.1.0.14.2.20021212094845.0483a190@127.0.0.1>
To: Frank Manola <fmanola@mitre.org>
Cc: pat hayes <phayes@ai.uwf.edu>, w3c-rdfcore-wg@w3.org, Patrick Stickler <patrick.stickler@nokia.com>

My eye has been off the ball recently, so my apologies if this is covering 
ground already discussed.  I think what I say here is mostly reinforcing 
what Frank said (below).

It seems we're getting hung up on issues of normative and formal 
definition.  I think its clear that a vocabulary can be defined normatively 
("intended" meaning) without being defined (or definable) formally.  True, 
without formal definition said meaning is not accessible to formal 
reasoners, but that doesn't mean it's in all respects meaningless.

So, lack of formal definition for a term doesn't mean it can't be defined 
normatively.

This begs a question of where such non-formal definitions should appear;  I 
think Frank hinted that the vocabulary document might be the right 
place:  "(actually, I think Schema covers that, at least about our 
"intent", and also does so for the other terms mentioned in the 
following)".  That makes sense to me given that the vocabulary document is 
now about more than just the schema vocabulary.

#g
--

At 04:25 PM 12/11/02 -0500, Frank Manola wrote:

>Pat--
>
>I'm sorry, but on the basis of Patrick's recently-expressed concerns about 
>"meaningless terms", I've got to temporarily object to this.  The problem 
>is roughly this:
>
>When I said:
> >>All these terms are discussed in the Primer (in some cases extensively),
> >>together with examples of their use.  In all the use cases, there are
> >>caveats expressed that describe these as "intended meanings",
> >>
>Patrick said:
> >
> > But intended by whom? If they are intended by the RDF Core WG, then
> > they should be normative. If they are intended by someone else, why
> > should we say anything about them or even include the terms in the
> > RDF vocabulary.
>
>and later said:
> > Precisely. I think that the Primer should reflect, in minimally technical
> > and accessible terms the normative content of the other documents, and
> > the examples and verbage for these terms does in fact suggest that RDF
> > is asserting meaning for these terms which it is not.
>
>So it seems to me that when the Semantics document describes the intended 
>meaning of terms from this vocabulary, like containers and collections 
>(and reification, and ...), it's a normative statement of our intent (even 
>if we can't fully define the semantics formally), and it's OK then to 
>elaborate on that in the Primer.
>
>[NB:  The Semantics document I'm referring to is 
>http://www.coginst.uwf.edu/~phayes/RDF_Semantics_finalCall.html]
>
>If anyone thinks the above is a sneaky argument, consider also the following:
>
>a.  Several applications described in Section 6 of the Primer use the 
>container vocabulary, and it seems to me they use it correctly. Although 
>we can't formally define everything about this vocabulary, it seems to me 
>we've stated it well enough in English so people can use it with a 
>reasonable degree of interoperability.  We could elaborate more on that in 
>the Semantics document (describing it as "intended" again) if people 
>insist on something normative.  But I don't understand how directing 
>people away from that vocabulary, and forcing them to individually invent 
>N different representations for containers on their own, helps achieve the 
>goals of globally-consistent meaning.
>
>b.  We added the "meaningless" collection vocabulary not that long 
>ago.  This isn't a piece of bad old legacy syntax from M&S.  Did we really 
>have no normatively-describable intent in doing that?  If so, then we can 
>delete it now without impacing anyone can't we?  If we actually mean 
>something by it (and I think we did), then let's say it.  (And I think the 
>Semantics document *does* say it, and the Primer reflects that).
>
>c.  There are some problematic aspects of the other vocabulary items, but 
>not that many really.  rdfs:comment?  Say in some normative document that 
>it's intended to provide a place to put comments (actually, I think Schema 
>covers that, at least about our "intent", and also does so for the other 
>terms mentioned in the following).  No one expects any more formal 
>semantics than that for a comment anyway do they, and why not have a 
>pre-defined way to make comments (most languages do)? rdfs:label?  Same 
>comment.  rdfs:seeAlso?  Relatively harmless.  It says there'a 
>relationship between the subject and object of the statement. Big 
>deal.  That's exactly what the formal semantics are.  Users could define 
>their own names for that relationship depending on how they use it, so we 
>could delete it, but why bother.  rdfs:isDefinedBy?  It's a subproperty of 
>rdfs:seeAlso.  Say what the intent is (which the Schema document does) and 
>forget it.  rdf:value?  We've been working on that, but I don't think 
>we've achieved closure yet.  (But there's a nice description of our intent 
>in the Semantics document.)
>
>So my current suggestion is to leave the current descriptions in the 
>Semantics document, so they're "normative", and we can continue to talk 
>about the uses of useful vocabulary in the Primer (of course, with 
>appropriate caveats about what is *not* guaranteed about uses of these terms).
>
>--Frank
>
>pat hayes wrote:
>
>>After reading through the Primer, particularly section 4, I propose that 
>>almost all of section 3 of the semantics document be simply removed. It's 
>>all in the primer section 4.
>>Semantics section 3 should just list the RDF vocabulary that isn't being 
>>given a formal meaning, say explicitly that it isn't, and refer to the 
>>Primer for the informal meaning. If anyone feels that there are any 
>>subtleties in section 3 that aren't in the primer text as it stands, then 
>>lets put them into it. I don't think there are many, if any, in fact.
>>Anyone object? Frank??
>>Pat
>
>
>--
>Frank Manola                   The MITRE Corporation
>202 Burlington Road, MS A345   Bedford, MA 01730-1420
>mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-875

-------------------
Graham Klyne
<GK@NineByNine.org>
Received on Thursday, 12 December 2002 05:09:43 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:54:52 EDT