[rdfweb-dev] FOAF IRC Meeting
bnowack at appmosphere.com
Tue Feb 22 18:43:19 UTC 2005
I'm not sure if I'll be able to attend the meeting tomorrow, but
I thought I could do some action item hijacking and summarise
some things I remember from previous meetings:
- new site
nicole has done a great job on maintaining a page in the wiki:
- date of birth
not resolved yet. I proposed 3 separate props for ease of
querying and flexibility. IIRC, one problem was LJ already
publishing other terms (foaf:birthday and/or foaf:dateOfBirth?).
- naming properties (first/last/given/...)
some stats were produced to see what's used. not sure about
- better documentation of terms' lifecycles.
some way to attach dates and/or versionInfo notes to terms.
- rdfs:Resource / NonLiteralResource issue
I did some reading and tried some things, here is the attempt of
I think the issue was raised by masahide kanzaki, noting some
terms of the foaf spec had a defined domain or range of
rdfs:Resource. The class rdfs:Resource includes rdfs:Literal,
so that such a range description is not sufficiently precise
to tell if a property is an object- or a datatype property.
Plain RDFS doesn't offer any constructs to describe a
property's range as te class of non-literal resources. One
problem is that apps/tools which e.g. auto-generate forms from
the spec could benefit a lot from a machine-readable hint
telling them to exclude literals as "allowed" values.
Several (owl-based) approaches have been proposed to solve
- using owl:Thing instead of rdfs:Resource
afaik, the owl docs say something along
owl:Thing is the class of all individuals that are
the unionOf owl:Nothing and the complement of
]]] which probably still includes literals as literals
can be treated as individuals in owl full.
- defining a blank class node with nodeID=NonLiteralResource,
described as owl:disjointWith rdfs:Literal and using that
on object properties' domains/ranges.
I implemented this approach in my vapourware vocab editor
(see NonLiteralResource, foaf:page, and foaf:topic in
I doubt that other editor developers would do that.
I also tried to load the adjusted spec with protege.
It didn't read the blank class description and the
modified ranges/domains were undefined.
- typing the properties as owl:ObjectProperties
this, again, may still not be sufficient in owl full,
where literals can (theoretically) be used with object
My personal opinion:
I'd keep rdfs:Literal but also type the props as
owl:ObjectProperty. imo, that'd make things clear
for human consumers (won't hurt to add some prose
as well), allow plain rdfs tools to still read
the spec, and shouldn't cause trouble with existing
ont editors. and I guess that app developers would
be fine with the object property hint; if they want
to implement corner cases of owl full, they can
still do that..
can't remember the details..
nice to see the foaf meetings re-appear :)
hasta manana (hopefully),
45145 Essen, Germany
On 06.02.2005 21:04:35, Christopher Schmidt wrote:
>This is a proposal for an IRC based meeting to be held on February 23rd,
>2005, at 1800Z. There are currently a number of outstanding issues
>regarding FOAF, some of which have been discussed, some of which have
>A suggested agenda:
> * Resolving Existing Issues:
> * IssueTracker 
> * Other Mailing List Issues
> * Discuss process for issue resolution in the future.
> * New website: Status update
> * Neat Things: What are people doing with FOAF.
>If anyone has any complaints about the time chosen, please respond to
>this message so that we can modify the time and date of the event: The
>goal is to interfere as little as possible with other meetings, while
>allowing resolution of as many issues as possible before thee Tech
>Plenary, which takes placee at the beginning of the next week.
>This meeting will hopefully include the primaries for the FOAF project:
>if not, I will be chairing the meeting.
>rdfweb-dev mailing list
>rdfweb-dev at vapours.rdfweb.org
More information about the foaf-dev