W3C home > Mailing lists > Public > public-esw-thes@w3.org > July 2008

Re: SKOS comment: change of namespace (ISSUE-117)

From: Simon Spero <ses@unc.edu>
Date: Fri, 25 Jul 2008 12:22:26 -0400
Message-ID: <1af06bde0807250922t679f4f7w69c8ac60c4afaec0@mail.gmail.com>
To: "Alistair Miles" <alistair.miles@zoo.ox.ac.uk>
Cc: "Laurent LE MEUR" <Laurent.LEMEUR@afp.com>, public-swd-wg@w3.org, public-esw-thes@w3.org
Can we clarify some  crucial point here:

1) Is it correct to say that  the new documents specify that one should only
assert broader, and broader transitive can always be inferred, but should
not be asserted, and cannot be trusted,

2) Must broaderTranstive always be valid?

3) Is the *sole purpose* of this change to provide imperative control over a
generic reasoner without using reification or other general mechanism for
provenance? Would it better not to usurp those functions for what is
indended to be an application profile rather than a fundamental technology.
 If not, could *broader* and *narrower* return to their standard semantics,
a new  *directlyAssertedBroader* property be define as a sub property of *
broader* to convey the extra meta-semantics? That would allow the old
namespace to be reused, and preserve standard semantics.

4) If a vocabulary wishes is standards compliant, must it add an explicit
broader:2008 relationship for every pair of elements for which broader:2004
holds?

5) If a vocabulary does not assert explicit broader:2008 relationships
wherever broader:2004 would be inferred, does that mean that the vocabulary
provider is stating  that *they do not support standards compliant
hierarchical relationships*.


   1. The domain of discourse for controlled vocabularies is the set of all
   Documents.
   2. The standard *BT* relationship requires that the subordinate concept
   be entirely contained within the superordinate concept.
   3. That is, the extension of the superordinate term must be a  proper
   superset of the extension of the subordinate term.
   4.  A relationship that does not satisfy this requirement over all
   domains of documents is *not  hierarchical* in Z39.19, and is
*only*admissible in ISO 2788 where the domain of discourse is
restricted to
   situations in which this  contion must always be true.
   5. The relationships that the standards permit  as  hierarchical are
   restricted *soley* to those that must universally apply.
   6. *Parrots* *BT* *Birds* is always admissible, because  *I[**Parrots** ]
   *  ⊂ *I[**Birds**]*
   7. *Parrots* *BT* *Pets* is admissible *only* *where* for subdomains
   where all parrots *must be* pets - for example, a corporate thesaurus for
   Pet Smart*, δPS* , *I[Parrots,δPS] ⊂** I[Pets**,δPS**]*


Where
      *BT* = broader as defined in 2004  (The KOS standard Broader Term
relationship)
       *B* = broader as defined in 2008 (Somehow more general).
       *BT** = transitive closure of *BT*
       *BTT* === broaderTransitive = *B**

   *  I[C,Δ]* = {extension of concept C in the universe of all documents *Δ*}

   *  I[C]* ≡  * I[C,Δ]*
     *I[C, δ ⊆ Δ]* = {extension of concept C in* δ* }

     A BT B ⊨ I[A] ⊂ I[B]
                ≡ ∀a. a ∈ I[A]  → a ∈ I[B]

     NOT(A BT C) ∃a. a ∈ I[A] ^ a ∉ I[C]

   *   A BT B ^ B BT C ^ NOT (A BT C)   -> ∃a. a ∈ I[A]  -> a ∈ I[B] -> a ∈
I[C] ^ a ∉ I[C] , a contradiction
Received on Friday, 25 July 2008 16:23:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:39:00 GMT