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

Hi,

To me the most important point in Tom's argument was the "not consider herself qualified". So this can be at first sight related to provenance. But fundamentally, it's a matter of authority, though maybe not in the usual Semantic Web sense [1]
Actually I'd like to have SKOS send the message that the mapping links are just more dubious to re-use than the standard relationships. The latter are part of the *definition* of the concept and supposed to be usable by any application that will access the concept scheme, while the former are not. 

Just have a look at the results of automatic mapping tools. Or even what is produced humans trying to build a mapping themselves, or fixing such an automatic mapping. My experience is that this is hugely error-prone. And our fate is that many mappings will be produced by automatic tools or by non-expert people [2]

So at some point I expect users will just want to know that something was said, which is reliable from a conceptual perspective. And that's essentially quite distinct from the problem of who said what. Antoine Isaac can publish nice SKOS conversion of existing validated schemes he got from his library, and crap mapping links he's just using for a demo.

Cheers,

Antoine

[1] And even if it was the case, I would be reluctant to say that SKOS should not deal with authority, as it should not deal with provenance. Unless you can point me to something that would solve the issue and that we could refer to as Dan does for SPARQL named graphs. That is, something that is established and already giving birth to credible applications or demos that we know of.
[2] Note that I don't say this is per se a problem, these mappings are clearly better than nothing, and in fact can be useful for many applications!


> On 22/1/09 19:07, Alistair Miles wrote:
> [...]
>> I would not use SKOS mapping properties to "enrich" a concept scheme
>> published by someone else. I would use skos:broader, skos:narrower and
>> skos:related. If you are concerned about whether these extra triples
>> were asserted by the owner of the concept scheme or by someone else
>> (e.g. me), you should keep track of where your triples came from.
>>
>> Hence I am suggesting a minimum commitment approach, because unless
>> everyone agrees with what I just said, we won't have consensus.
> 
> FWIW I agree with what you just said. Since SPARQL arrived with named 
> 'graph' support, it has become quite easy to super-impose SKOS data from 
> different sources. While lcsh.info was alive, I started to make a demo 
> of this kind of overlay here - 
> http://wiki.foaf-project.org/SanfordBergmanLCSHScorecard.
> 
> I think the idea as far as possible with RDF/RDFS is to avoid the 
> meaning of words getting tangled up with the identity of the party using 
> them. Sometimes there are exceptions, and the mapping stuff may be in 
> that category, but in general, if we can keep the terms meaning the same 
> thing (incl. having the same truth conditions) regardless of the user, 
> life's simpler all round.
> 
> 
>>> 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].
>>
>> The sentence "using the SKOS mapping properties is no substitute for
>> the ... use of provenance mechanisms" means exactly what it says. SKOS
>> mapping properties tell you nothing about provenance (who said what).
> 
> I'm in agreement with Alistair and Attributed-to-Alistair above.
> 
> cheers,
> 
> Dan
> 
> -- 
> http://danbri.org/
>

Received on Tuesday, 27 January 2009 14:21:48 UTC