[foaf-dev] strawpoll: migrate from foaf:holdsAccount to foaf:account?

[Richard Cyganiak]

On 24 Aug 2009, at 21:42, Toby Inkster wrote:
> On 22 Aug 2009, at 18:37, Richard Cyganiak wrote:
>> I'll repeat my POV that the URI of an account
>> homepage -- e.g., http://twitter.com/cygri -- is a good URI for the
>> account itself.
>
> I don't agree with this. Consider:
>
> 	<#danbri>
> 		foaf:account
> 			<http://www.flickr.com/photos/danbri/> ,
> 			<http://www.flickr.com/people/danbri/> .
>
> Each triple is an entirely sensible in its own right. So how many  
> accounts does danbri have on Flickr? The two URIs represent  
> something which is logically the same account, but it certainly  
> wouldn't be right to assert:
>
> 	<http://www.flickr.com/photos/danbri>
> 		owl:sameAs
> 			<http://www.flickr.com/people/danbri> .

Flickr is a pathological example, because it's not clear which of  
these pages is the true homepage of the account. For accounts on most  
websites, this problem does not exist. Making the typical case hard to  
accomodate a weird edge case is not smart.

The Flickr case can be dealt with by including some language into the  
spec that more or less arbitrarily picks one of the candidate URIs,  
e.g. "if it is unclear which page is the main page of an account on a  
website, then it's the one that is typically linked to when the user's  
name appears throughout the site." Which would be /photos/, in the  
Flickr case.

> As another point, if foaf:account is defined to link to an account  
> homepage, then how does it differ from foaf:homepage? Why would the  
> property even be needed at all?

My (insert random Web 2.0 site I signed up for but never used) profile  
page is not my homepage. These are different concepts in most people's  
understanding of the web. Same thing with foaf:weblog, btw.

Best,
Richard



>
> -- 
> Toby A Inkster
> <mailto:mail at tobyinkster.co.uk>
> <http://tobyinkster.co.uk>
>