W3C home > Mailing lists > Public > public-esw-thes@w3.org > December 2007

RE: [SKOS]: [ISSUE 44] BroaderNarrowerSemantics

From: Sini, Margherita (KCEW) <Margherita.Sini@fao.org>
Date: Mon, 17 Dec 2007 13:00:40 +0100
To: Leonard Will <L.Will@willpowerinfo.co.uk>, public-swd-wg@w3.org, public-esw-thes@w3.org
Message-id: <BA453B6B6B217B4D95AF12DBA0BFB669029DAE07@hqgiex01.fao.org>

Hi all,

I also agree with Joe and Leonard.   In my opinion:

- BT/NT is thesauri can somehow be considered like the "is a" in an ontology
(even if this is not true all the times).
So in my opinion skos:broader and skos:narrower should be transitive.  If
this is not true in the thesaurus is may probably be that in the thesaurus
itself the relationship is wrongly assigned, (maybe was a RT instead of
BT/NT) or  it may be that the meaning of the applied BT in the thesaurus is
not "is a" but something like "has generic broader term"... But also in this
case I can see that if I say 
	A "has generic broader term" B
	B "has generic broader term" C
Then I can conclude that A "has generic broader term" C... So seems to me
that BT should be transitive...
I think the same apply for NT.

For the classifications schemes, I think that in that case the BT/NT
relationships can actually be considered as "has super-category" or
"sub-category of"... And in my opinion, they should also be transitive...

In fact, if in the thesaurus we ends up with inconsistencies such as the one
between Doorbells and Mammals, in my opinion just means that somewhere in the
tree a BT/NT was not properly used... and an RT should have been somewhere
used instead...

So, the inconsistencies pointed out by Leonard, should have been fixed
because if we consider BT as transitive they appear not correct. If we
consider BT non-transitive then we can leave as they are and the samples may
be true (and not inconsistent)... But I vote for the first case.

So I would suggest that  (not ">" for me means "BT"):

	non-alcoholic cocktails > cocktails > alcoholic beverages

becomes in my opinion:

	alcoholic beverages     > beverages
	non-alcoholic beverages > beverages
	cocktails > beverages
	alcoholic cocktails     > cocktails
	non-alcoholic cocktails > cocktails
	alcoholic cocktails > alcoholic beverages
	non-alcoholic cocktails > non-alcoholic beverages
	(I used double parenting...)
	So it would ends up to be:
	non-alcoholic cocktails > cocktails > beverages
	non-alcoholic cocktails > non-alcoholic beverages > beverages

And for the other example I suppose we may find solutions... ?

Hope this helps

-----Original Message-----
From: public-esw-thes-request@w3.org [mailto:public-esw-thes-request@w3.org]
On Behalf Of Leonard Will
Sent: 14 December 2007 21:55
To: public-swd-wg@w3.org; public-esw-thes@w3.org
Subject: Re: [SKOS]: [ISSUE 44] BroaderNarrowerSemantics

On Fri, 14 Dec 2007 at 13:40:47, Ed Summers <ehs@pobox.com> wrote
>The filenames are reflective of the "seed" concept that I used to walk
>about a certain radius from. Well and it's Friday... I'd be interested 
>to hear where you see the broader relationship not working.

Ed -

I agree with Joe Tennis that you will have difficulties in avoiding 
inconsistencies if you take the BT/NT relationships in LCSH as 
equivalent to the way they are applied in thesauri developed in 
accordance with standards. (To be fair, I believe that the LCSH people 
are well aware of this, and are working gradually towards standard 
conformity, but they have such a load of existing material to deal with 
that it will take some time.)

You asked for any invalid relationships in your diagrams. I have not 
checked all the XML, but I take it that the arrows in the diagrams are 
intended to point from a narrower concept to a broader one. Many of the 
relationships do indeed appear to be valid, but a few invalid ones I 
spotted are:

non-alcoholic cocktails > cocktails > alcoholic beverages

malt > beer
[is malt really a kind of beer? If this is a part-whole relationship, it 
would be acceptable only if malt occurred only as a part of beer, which 
is not the case.]

malt-extracts > beer
malt-extracts > malt > beer
[as above, and the second of these is redundant and would not be 
accepted in a thesaurus if the first was present]

malt > brewing
[these concepts are in different facets, "material" and "activities" and 
so cannot be hierarchically related]

wine and wine making
[LCSH often combines terms from an activity facet with terms from 
materials or objects facets like this. They really need to be separated 
out in order to create valid thesaurus hierarchies.]

WebTV (trademark) > World Wide Web
[you might get away with WebTV being _part_ of the WWW, and perhaps the 
qualifier "trademark" is just the legal "TM" symbol spelled out, but if 
the concept is the trademark, as it appears, it is not a true 
hierarchical relationship. Part/whole relationships are best kept to a 
few specific situations, spelt out in the standards.]

semantic web > semantic integration (computer systems)
[mixed facets - a system and an activity or property]

semantic integration (computer systems) > integrated software [mixed facets -
an activity or property and an intellectual product or 

It is not possible to be definite about some of these; I have not 
checked all the scope notes, where they exist, but there do appear to be 
prima facie problems.

I suppose we are looking only at the relationships at the moment - many 
of the terms are not formatted as thesaurus standards would have them, 
but that is another issue, e.g.

cereals, prepared
[thesaurus terms are not inverted like this]

motion pictures and rock music
[a compound concept, presumably meaning "motion pictures in combination 
with rock music", not just the whole of both concepts viewed separately]

gays and rock music
[another compound of two concepts]

piano - studies and exercises (rock)
[a combination of three concepts: instrument, form and style]

I cannot comment on most of the popular music examples, as I would not 
recognise most of them if I heard them. I wonder if anyone can write 
clear scope notes to distinguish one from another!

Your examples do show the need for some grouping into arrays with node 
labels to show characteristics of division. These would organise sibling 
terms more clearly, e.g. (music by instrument), or (plant products by 
source plant). There are also some

Thanks for the interesting Friday exercise!


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, 17 December 2007 12:01:27 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:45:45 UTC