Re: non opaque primary topics

On 9 May 2013 02:28, Tim Berners-Lee <timbl@w3.org> wrote:

>
>
> On 2013-05 -07, at 17:08, Melvin Carvalho wrote:
>
> > the standard model in sem web seems to be
> >
> > <> foaf:primaryTopic <#relative-uri>
> >
> > I've seen 4 relatively common relative URIs used here and there
> >
> > 1. #me -- common in foafs
> > 2. #i -- often used by timbl and tobyink
> > 3. #this -- used by kingsley as per the 'this' keyword in OO
> > 4. # -- used by facebook and others
> >
> > It's a very common case that a document will contain one primary topic
> and I want to standarize all of my pages (mainly robots) into one of these
> 4.
> >
> > It seems to me that (4) is the most sensible choice, after "it doesnt
> matter".
> >
> > Does anyone have any preference here?
>
> A URI takes the form
>
>           <document identifier>  #  <local identifier>
> or
>           <document identifier>
>
> The point about this is that all kinds of languages can talk
> about the same thing in different ways.   To get interopabiliity,
> though, you have to stick to a limited set of local identifiers which
> will work in any language.
>
> So having the empty string as a local identifier does
> in fact work in RDF.  But it is a pain.
>
> You can't use qnames with an empty localid:
>
> > 1. :me -- common in foafs
> > 2. :i -- often used by timbl and tobyink
> > 3. :this -- used by kingsley as per the 'this' keyword in OO
>
>
> Ooops! facebook's empty localid isn't allowed as a localid.
>

Ah Thanks.  I was unaware of this difference between qnames and ntriples.
It does seem that the empty fragment could be unsafe.

However I think that qnames are predominantly (maybe always?) used in the
predicate position, so this makes facebook profiles still OK?


>
> That means also when machines later re-serialize your data,
> they won't be able to use the qname syntax.
>
> Of course lots of other languages have restrictions that
> identifiers have length  > 0.
>
> So, while for facebook I'm not complaining that they
> are using turtle, I wouldn't advice anyone to use an empty localid.
>
> Of the other 3 forms, take your pick.   I must say I  used
> :i but I find it less easy to explain than :me. Another possibility
>  for a FOAF file is #timbl  which might have
> read better when explaining to a class.
>
> But as someone else said the string you use does not
> is opaque in the system and so :x or :it   is fine.
> I quite like :it if you are generating stuff automatically.
>
>
>
> Tim
>
>
>