Re: [SKOS] "Mapping" vs "standard" relationships

Alistair, Antoine,

On Wed, Jan 21, 2009 at 11:41:40AM +0000, Alistair Miles wrote:
> > I don't see a real opposition here, unless Tom did really intend to discuss evolution and enrichment in the Primer: to me the core of his proposal lies in the paragraph:
> >
> >>> > The argument could run as follows: Ideally, we should be able
> >>> > to tell from provenance information who said what, but in
> >>> > practice, Semantic Web data is often merged in simple ways that
> >>> > obscure the origins of assertions.  The distinction between
> >>> > "mapping" and "standard" relationships is one of etiquette --
> >>> > directly asserting "standard" relationships sends the message
> >>> > that the asserter considers herself qualified to define the
> >>> > relationship in a standard way.  For everyone else, the polite
> >>> > thing is to assert a "mapping" relationship.
> 
> Well, I don't like the idea that "mapping" vs. "standard"
> relationships should carry any connotation such as "the asserter
> consider herself qualified...". This was not what I had in mind at any
> stage of the design process.
> 
> Generally, I think it is a bad idea for properties to carry any
> connotation regarding authority or provenance. This includes a
> connotation such as "...the asserter considers herself
> qualified...". I discussed this in an email I wrote back in Feb last
> year [1] (see section "My Position").

Very nice email [1]!

> How about something like...
> 
> """ The SKOS mapping properties skos:broadMatch, skos:narrowMatch and
> skos:relatedMatch are used to represent links that have the same
> intended meaning as the semantic relation properties skos:broader,
> skos:narrower and skos:related. However, by convention, mapping
> relationships are only asserted between concepts that belong to
> different concept schemes. """
> 
> I actually think this is all the Primer needs to say. 
> 
> If you wanted to provide some additional context, you might add
> something like...
> 
> """ Mapping properties are expected to be useful in applications that
> use multiple, conceptually overlapping KOSs; for example where an
> application is providing integrated search across bibliographic
> metadata from several different sources, each of which has used a
> different KOS to index items. """
> 
> If you really wanted to go into more detail about the SKOS data model
> for mapping properties, you could say something like...
> 
> """ Advanced SKOS users may note, however, that there are no formal
> constraints in the SKOS data model restricting the use of SKOS mapping
> properties to link only concepts in different schemes. For further
> discussion see section 10.6.1 of the SKOS Reference. """

For the record, I think the text above would be fine, but see below.

> > Such a paragraph is I think rather agnostic towards promoting KOS evolution and enrichment as full-fledge scenarios. Note that like you, I would be actually uncomfortable with extensively discussing KOS evolution and enrichment in the Primer.
> > As a matter of fact, the only mention of these in the Primer is:
> >
> >> However, the use of mapping properties might also be appropriate in cases where someone other than its owner needs to enrich the semantic relationships within a particular concept scheme.
> >
> > Do you think that sentence is too strong? For me it's a relatively open sentence, in line with what you propose -- it is not *discussing* the scenario. IMHO, doing more than using this "might" (even if to say explicitly that this is an open scenario) could actually puzzle the reader by giving more importance to these potential "cases" than what is required.
> 
> I don't have a problem with that sentence, but see also below.

However it was in fact this enrichment scenario which motivated
me to propose an alternative formulation in the first place.

If the point is that "someone other than its owner" is using
mapping properties to say how concepts within one concept
scheme relate to each other, then this seems to be at odds
with the idea that "mapping relationships are only asserted
between concepts that belong to different concept schemes".
I proposed the alternative formulation (see below) in order
to make explicit what I thought the example already implied.

Basically, I think that if we agree that the enrichment
scenario above is a good example for using mapping properties,
we should consider an alternative formulation along the
lines I propose.  If we are not confident about the enrichment
scenario, I would be happy with the minimal commitment proposed
by Alistair.

But in that case, my question would remain: Do we think that
the convention about choosing between mapping or standard
semantic relationship properties has _anything_ to do with
provenance?  Alistair thinks it is a bad idea for properties
to carry any connotation regarding authority or provenance,
but in last week's call I thought we were also implicitly
acknowledging that the "convention" had something to do with
provenance when Ralph suggested: "note that we recognize a need
for standard ways to communicate provenance in the Semantic
Web and when we have such mechanisms, this question of what
one thesaurus provider says versus what others say about the
thesaurus will become more explicit" [2] -- in other words,
to acknowledge that the "convention" has something to do with
provenance, even if we emphasize that "using the SKOS mapping
properties is no substitute for the careful management of
RDF graphs or the use of provenance mechanisms" [3].

Tom

[1] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0095.html
[2] http://www.w3.org/2009/01/13-swd-irc#T17-01-26
[3] http://www.w3.org/2009/01/13-swd-irc#T17-03-59


>>> > The argument could run as follows: Ideally, we should be able
>>> > to tell from provenance information who said what, but in
>>> > practice, Semantic Web data is often merged in simple ways that
>>> > obscure the origins of assertions.  The distinction between
>>> > "mapping" and "standard" relationships is one of etiquette --
>>> > directly asserting "standard" relationships sends the message
>>> > that the asserter considers herself qualified to define the
>>> > relationship in a standard way.  For everyone else, the polite
>>> > thing is to assert a "mapping" relationship.

-- 
Tom Baker <tbaker@tbaker.de>

Received on Wednesday, 21 January 2009 13:29:37 UTC