W3C home > Mailing lists > Public > public-powderwg@w3.org > June 2007

Re: [SKOS] ACTION on relating skos:Concept to a foaf:Person

From: <aisaac@few.vu.nl>
Date: Thu, 21 Jun 2007 14:35:54 +0200
Message-ID: <1182429354.467a70aaa3c49@www.few.vu.nl>
To: Kjetil Kjernsmo <kjetilk@opera.com>
Cc: public-swd-wg@w3.org, Public POWDER <public-powderwg@w3.org>


> On Wednesday 20 June 2007 23:27, aisaac@few.vu.nl wrote:
>> Both problems are similar in the sense that they are concerned about
>> how to link a concept with some real-world-anchored reference, be it
>> an object or a set of objects. But they have subtle differences.
>> Following and extending what is written in the SKOS core guide about
>> dc:creator, you can say that the dc:creator of the concept
>> mythesaurus:KingOfEngland is Antoine Isaac and that Antoine Isaac is
>> also the creator of the class myontology:KingsOfEngland. However if I
>> can be the dc:creator of mythesaurus:henryVIIIConcept, I cannot be
>> the dc:creator of myontology:henryVIIIPerson, because the creators of
>> henry VIII are his parents.
>
> Yes, indeed, they are different, and the problem is that I have tried 
> and failed to create a reasonable UI to allow people to select the 
> predicate (the code is running on http://my.opera.com/ so if somebody 
> wants to have a look into how it performs, I'd encourage you to get 
> account), so that they could clearly state the semantics of that link. 
> That is not to say it can't be done, better UI designers than me could 
> do it one day. But there is an immediate need that needs to be met. 
>
> In fact, I think this would be the need for most sites that uses tags 
> (del.icio.us, flickr, youtube, etc), and given that quite a lot of 
> annotation is available as tags, I would think that not supporting it 
> would be unwise.

Are you sure this is so much a need? When tagging I just basically need, well,
tags and documents. With this you can describe, and then search. What you
mention seems to be a more sophisticated feature, where you want to link a
subject to the ontological representation of a 'real' entity. What for? Do you
want to search for documents by putting instances of foaf:persons in your queries?

>
> So, what I need is a property where the exact meaning is not as clearly 
> defined as was intended with skos:it, I think. So, yeah, it is probably 
> not quite skos:it I want. 
Could you point to a reference page/mail where skos:it and skos:at are defined?
To me "it" is not that clear ;-)
>
> It is a better-than-nothing property. For the property I need, the exact 
> semantics of the relationship between the concept and the thing is not 
> defined, but that doesn't mean that the somewhat fuzzy relationship is 
> not useful, as the uptake of tags has shown the amount of annotation 
> people are willing to do creates a useful amount of data, and you can 
> at least with some probability infer that a picture that has some 
> relationship to the concept if it is tagged that way. It is more useful 
> than not having any annotation at all.
>
> The skos:it, or similar, could be a refinement that clearly defines this 
> meaning, so that when a better UI designer comes around, users that 
> wants to specify it can do so.

So you need more fuzzier than a link between a concept and its reference? Do you
have examples of such fuzzier links? Because the case of a concept and the
person this concept refers to is finally quite straightforward a case of skos:it
(if this 'skos:it' is the link between a concept and its reference)

Cheers,

Antoine

PS: by the way if you are urgently interested in the POWDER thing it's perhaps
better if you anser the second part of my previous mail ;-)
Received on Thursday, 21 June 2007 12:36:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:42:11 GMT