FW: unique altLabels

Leonard has addressed this question very thoroughly, pointing out the
confusion that can arise if any label (whether preflabel or altlabel) is
non-unique.  The argument is entirely contingent on the assumption that
the KOS we are dealing with is a thesaurus, that is to say a vocabulary
complying with ISO 2788 or ANSI/NISO Z39.19 or BS 5723. If this is the
model, then the convention must be followed that every term, whether a
preferred term or a non-preferred term, must be unique.

However, the preamble to the SKOS document says, "SKOS-Core 1.0 is an
RDF schema for representing thesauri and similar types of knowledge
organisation system (KOS)". What do we mean by "similar types"? If a
classification scheme is included, you move to a different model. For
classification schemes, every concept has a unique notation. The
concepts are generally known as 'classes'. Each class has a caption, and
sometimes also a descriptive note. The captions do not have to be
unique. For example, the meteorology schedule of the scheme could have a
class ABC56, with caption Depression; at the same time the mental health
schedule could have a class XYZ13, with caption Depression. Obviously
these concepts are very different, but there is no confusion because the
caption is never used to represent the concept.

All this is leading up to a suggestion that it may be important to set
out more background as to where and how the SKOS schema is intended to
be used. If it is intended to apply to situations other than that of an
ISO2788-compliant thesaurus, then these should be described, together
with advice on how to deal with the features that differ from the
ISO2788 model. 

In the short term, I would urge limiting the applicability of the schema
to situations where terms may be used to represent concepts, and in this
circumstance any term used to represent a concept should be unique.  I
also agree about the need to limit it to a single vocabulary.  Once this
simple basis is established, it may become possible in future to extend
to more situations, but it is risky to try to move too fast to a very
sophisticated model.
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
*****************************************************



-----Original Message-----
From: public-esw-thes-request@w3.org
[mailto:public-esw-thes-request@w3.org] On Behalf Of Leonard Will
Sent: 15 March 2004 12:37
To: public-esw-thes@w3.org
Subject: Re: unique altLabels



In message
<350DC7048372D31197F200902773DF4C04944166@exchange11.rl.ac.uk> on Mon, 
15 Mar 2004, "Miles, AJ (Alistair)" <A.J.Miles@rl.ac.uk> wrote
>Leonard wrote:
>[Quote from SKOS-Core guide:]
>"It is perfectly reasonable, however, to assign a concept a preferred 
>label that is also an alternative label for some other concept."
>>
>> This is contrary to thesaurus practice and standards, and would cause
>>problems. Labels should be unique, and are made so by the  addition of

>>a qualifier in parentheses if necessary.
>
>Leonard could you outline exactly the problems this would cause?

When you enter a vocabulary with a term, you don't know whether it is 
preferred or non-preferred. It should still lead you unambiguously to a 
single concept. This allows for automatic substitution of preferred 
terms for non-preferred terms when indexing or searching (or automatic 
linkage of documents to the appropriate concepts, if the actual terms 
are not used).

If you use a term that is ambiguous, such as "seals", then there is 
normally an initial process of "disambiguation", which presents options 
from which the user has to choose, e.g.

seals (closures)
seals (mammals) USE Pinnepedia

In this example the first of these is a preferred term and the second is

a non-preferred term (or "alternative label" if you wish).

If you look up "seals" in an alphabetical sequence the two entries above

are immediately evident. If a machine is accessing the thesaurus, there 
may need to be some provision for it to match the term without the 
parenthetical qualifiers and take steps to determine which of the 
options is appropriate. This seems to be a separate step, though, and 
does not imply that the unqualified term is an alternative label for 
either of the concepts.

Chaals raises the issue of merging different vocabularies. I think that 
that is a different issue and raises many other complications, but as he

says the extra verification or disambiguation step that I have described

above applies there too.  As the present SKOS draft doesn't cover 
questions of mapping between vocabularies, I think we should keep it to 
what is necessary and desirable for a single vocabulary.

Leonard
-- 
Willpower Information       (Partners: Dr Leonard D Will, Sheena E Will)
Information Management Consultants              Tel: +44 (0)20 8372 0092
27 Calshot Way, Enfield, Middlesex EN2 7BQ, UK. Fax: +44 (0)870 051 7276
L.Will@Willpowerinfo.co.uk               Sheena.Will@Willpowerinfo.co.uk
---------------- <URL:http://www.willpowerinfo.co.uk/> -----------------

Received on Monday, 15 March 2004 09:26:34 UTC