W3C home > Mailing lists > Public > public-esw-thes@w3.org > March 2004

Re: Modelling 'term-to-term' relationships in SKOS

From: Douglas Tudhope <dstudhope@glam.ac.uk>
Date: Thu, 4 Mar 2004 19:30:32 -0000
Message-ID: <026d01c4021f$29ec24c0$92754d51@unet.com>
To: <public-esw-thes@w3.org>

>Types of synonyms and parts of speech:

The standards mention types of equivalence relationships. Most of these are
not currently explicitly represented. It is possible that they might be
though in the future by specialisation.
Some thesauri do in fact explicitly represent types of equivalence (see
below).

From Aitchison, Gilchrist & Bawden, 2000
(the new draft thesaurus standard is similar)
F1.1.1 synonyms
popular - scientific names
common names - trade names
standard names - slang
current - obselete
etc
F1.1.2 Lexical variants
variant spellings
abbreviations - full names
singular - plural
(perhaps in some cases noun-adjective - I think it is useful to have the
option of augmenting thesauri in the future with some basic linguistic
relationships. )

In some Thesauri, different types of equivalence are explicitly represented
eg (I've also seen examples concerning trade names and types of scientific
name in other thesauri)
from http://www.multites.com/conference03.htm, CAB Thesaurus presentation
(James Brooks)
CAB Thesaurus in MultiTes - Equivalence Relationships
CNP - Common Name PT vs SNN - Scientific Name NPT
CSN - Chemical Standard Name vs CTN - Chemical Trade Name
FFT - Full Form vs ABB - Abbreviated Form
SEN - Senior Scientific Name vs JUN - Junior Scientific Name
SNP - Scientific Name PT vs CNN - Common Name NPT
UK - British Form vs US - American Form

The question is whether the concept plus labels route would make these
distinctions harder to make? One way I suppose would be to introduce more
types of mark-up in the labels as for the language tags? I think it is
important to try and cater for common (and maybe extensible for future)
subtypes here - could this be done by specific tags?


> A label may be used as a label for more than one concept.

Homographs are common in natural language and can arise in controlled
languages. Some thesauri do have more than one concept associated with a
non-preferred term (eg AAT). My concern was that only having labels in SKOS
would make it more difficult for a retrieval application to deal with this
sensibly - having to process all sets of labels for a thesaurus as opposed
to directly looking up a term in the entry vocabulary. However on
reflection, I guess that applications would populate a dictionary or set of
data structures from the SKOS information for a particular KOS and work off
that.
Is this how you see it, Al?

Doug
Received on Thursday, 4 March 2004 14:35:29 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:51 GMT