RE: Scheme versioning & change management

Stella

I appreciate very much your bottom line

> ... the network of relationships may be a more reliable indicator than the URI.

Food for thought in this domain can be found in old debates in Topic Maps community around
the notion of "subject identity" [1]. Which properties should be used to infer that two
"subjects" (read : "concepts") are identical? A fundamental Semantic Web axiom is that
subject identity can/should be captured through a single property (a URI). But more than
whatever is declared by a URI (in its very structure, and/or whatever "Resource" it
"Identifies"), it's indeed the way(s) the URI is used, otherwise said its network of
relationships (for example in a concept scheme, but not only ...) which carry better its
identity.

[1] See e.g.
http://www.isotopicmaps.org/pipermail/sc34wg3/2003-November/001909.html

Bernard Vatant
Senior Consultant
Knowledge Engineering
Mondeca - www.mondeca.com
bernard.vatant@mondeca.com


> -----Message d'origine-----
> De : public-esw-thes-request@w3.org
> [mailto:public-esw-thes-request@w3.org]De la part de Stella Dextre
> Clarke
> Envoye : vendredi 13 aout 2004 13:34
> A : 'Miles, AJ (Alistair) '; public-esw-thes@w3.org
> Objet : RE: Scheme versioning & change management
>
>
>
> These lines in an earlier message caught my eye:
> "As a convention for managing URIs for concepts, I suggest that once a
> concept URI has been published, preference should always be given to
> deprecating and replacing with a new concept, rather than altering the
> concept.
>
> Usually, when a concept is dropped from a scheme, another concept or
> combination of concepts is added to replace it. "
>
> Just a little health warning: concepts are not so cut and dried as you
> might expect. They are slippery customers, and it is often impractical
> to decide when a concept has been "changed" or when the thesaurus entry
> has changed without changing the concept. Here are some of the changes
> that can happen to the record for a concept:
> - A new scope note is added. (Has the concept changed, or has it simply
> been clarified?)
> - An existing scope note is changed. (But the editor would argue, the
> intended concept has not changed, only been clarified because some
> people were misinterpreting it)
> - New relationships are added. (No change to the original concept, only
> an amendment when additional concepts were added to the thesaurus)
> - New relationships are added. (The concept looks the same, its
> preferred term is the same, but it never did have a scope note and in
> fact the new hierarchical relationship presents it in such a way that
> the concept will be perceived differently and effectively will have
> changed.)
> - The preferred term and one of its non-preferred terms are swapped.
> (But the concept has not changed.)
> - The spelling of the preferred term is changed (Just a correction, so
> the concept has not changed.)
> In any case, concepts tend to creep slowly and subtly over the years as
> people gradually shift their patterns of terminology use.
>
> It is often quite subjective deciding whether a thesaurus change amounts
> to a change in a particular concept. The ID assigned to a particular
> concept (or term) is often applied automatically without the editor even
> seeing it. And that ID is often the basis for the URI. When making the
> changes, the editor may do them by deleting a record completely and
> starting afresh EVEN THOUGH the concept has not changed, and in the
> process deriving a completely new ID and URI, or may just modify an
> existing record, retaining the same ID and hence URI, but maybe this
> time the concept really has changed!
>
> Sorry to go on a bit. I am just trying to point to the danger of relying
> on automatically assigned identifiers to show where concepts have or
> have not changed. ( And if the IDs were added humanly, they would still
> be subject to errors and subjectivity).
>
> Also, I'm not sure I agree with the assertion that "Usually, when a
> concept is dropped from a scheme, another concept or combination of
> concepts is added to replace it." I suspect that what this intended to
> describe was the dropping of terms rather than concepts. Concepts are
> sometimes but not often dropped from schemes. Terms may be dropped, or
> rather they may become non-preferred terms pointing to a different
> preferred term, and during this process the original concept may be
> retained or subtly modified. It is useful to track the original concept
> if you can, and the network of relationships may be a more reliable
> indicator than the URI.
>
> Stella
>
> *****************************************************
> Stella Dextre Clarke
> Information Consultant
> Luke House, West Hendred, Wantage, Oxon, OX12 8RR, UK
> Tel: 01235-833-298
> Fax: 01235-863-298
> SDClarke@LukeHouse.demon.co.uk
> *****************************************************
>
>
>
>

Received on Tuesday, 17 August 2004 08:52:58 UTC