[rdfweb-dev] foaf:depicts and parts of an image

Dan Brickley danbri at w3.org
Fri Oct 10 15:00:48 UTC 2003


* Masahide Kanzaki <post at kanzaki.com> [2003-10-10 22:21+0900]
> Hi Libby, thanks for prompt suggestion.
> 
> >And we decided to add regionDepicts to the foaf vocabulary. I can't
> >actually see it in the specification though:
> 
> Hmmm... regionDepicts doesn't seem, eh, very good for me ...

OK, I'll hold off from adding it (tracking these discussions in
Bugzilla, or perhaps the Wiki, seems good though).> 

> If foaf:regionDepicts is a subproperty of foaf:depicts, then
>  X foaf:regionDepicts Y. => X foaf:depicts Y.

I wasn't anticipating that., since we've never claimed (yet) that a 
region of an Image is an Image.

> by definition of subproperty. This means we can use foaf:depicts for
> region, and does not solve the current question.

Yup

> If it's not a subproperty of foaf:depicts, we have two unrelated (at least
> for machine) terms, whose meanings are basically the same or closely
> related (that's why those names share English verb 'depicts'). I'm afraid
> this sort of new term might hurt the generality and usability of FOAF
> vocabulary...

I'd support a separate vocabulary for image content description, I'm 
sure there will be other things beyond regions that will prove
interesting to describe.

> 
> A reckless question: should foaf:Image be a subclass of foaf:Document ? (as
> I asked in IRC several months ago.) I think a region (part) of an image

I think it should be. We currently say it is subclassof
wordnet:Document, which suggests this is just an editorial oversight.

> could be a foaf:Image, and if so, foaf:depicts solves those problems.

That seems more controversial.

In practice, I suspect many people have code and tools which assume that
the kinds of images we reference with depicts/depiction are bitmaps 
addressable on the Web. I can imagine this practice extending to SVG
as browser support for SVG grows, since a entire SVG image is a
self-standing thing. I'm not sure the concept extends naturally to 
cover fragments of images, since the likely processing model for 
dealing with values of this property will vary.
> 
> >I actually think depicts would be fine with a bit of tweaking.
> 
> I agree. Would you tell how you think to tweak this ? It would have a
> better solution.

Yup, I'd be happy to see a worked through proposal from someone.

cheers,

Dan



More information about the foaf-dev mailing list