[uf-discuss] Re: A (big) problem with XFN: identity of source and target not findable

[Toby A Inkster]

Costello, Roger L. wrote:

> I don't see any solution to the problem with XFN.  As far as I can see,
> social networks using XFN cannot be processed by spiders.  Only social
> networks that use FOAF can be processed by spiders.  Bummer.
> 
> Hopefully, I am missing something.  I really like the simplicity of XFN
> and its rich set of relationships.

You're not missing much. XFN *can* be useful for robots, but it takes a 
lot of work.

Let's take a look at the RDF model. Now RDF/XML as a syntax is not too 
friendly for authors, but I'm not going to concentrate on the minutiae of 
the syntax here: just the underlying model. RDF encodes three things:

	<subject> <predicate> <object>

e.g.

	X	foaf:name	"Toby Inkster"
	X	foaf:maker	Y
	Y	dc:identifier	"http://tobyinkster.co.uk"
	Y	dc:title	"TobyInkster.co.uk"

and so on. Although these triples seem simple, virtually any information 
is capable of being represented in this manner. (Although it may not be 
the easiest or most natural representation of the data.)

If we look at where XFN fits into the model, it basically adds a bunch of 
predicates (which we can imagine to be implicitly in the "xfn:" namespace 
even if the namespace isn't explicitly declared) like xfn:contact, xfn:me 
and xfn:sibling.

It seems initially that the subject and object of these predicates is 
undefined, but really when we have:

	http://bob.example.com	xfn:sibling http://dave.example.com

what is meant is:

	X			xfn:sibling		Y
	http://bob.example.com	foaf:primaryTopic	X
	http://dave.example.com	foaf:primaryTopic	Y

In other words, the two web pages are not siblings -- the primary topics 
(i.e. people described by the pages) are siblings. 

XFN right now doesn't define a way of determining the primary topic of a 
page, and probably never will. But if you need to be able to do determine 
this, then there are ways and means. FOAF is one possibility; and there is 
work being done as part of the hCard project to determine the "definitive 
hCard" for a given URL; failing either of those solutions, heuristics can 
be used (e.g. spidering out rel="me" links until you find the information 
you need). Right now it's all a bit undefined and of limited use in 
automatic cataloguing of human relationships.

Also see my second reply (11 Mar 2008, 20:26 UTC) in the "XFN getting 
smoked by FOAF" thread.

-- 
Toby A Inkster BSc (Hons) ARCS
[Geek of HTML/SQL/Perl/PHP/Python/Apache/Linux]
[OS: Linux 2.6.17.14-mm-desktop-9mdvsmp, up 15:04.]

                              The Semantic Web
                http://tobyinkster.co.uk/blog/2008/03/09/sw/