[foaf-dev] Group vs Class

Dan Brickley danbri at danbri.org
Sun Jun 13 14:59:33 CEST 2010


On Sun, Jun 13, 2010 at 2:48 PM, Dave Reynolds
<dave.e.reynolds at googlemail.com> wrote:

[useful detail snipped]
> One thing you might want to to consider is having a foaf:membershipClass
> property which links a Group to a class representing the set of members.
> That lets you keep the separation between the group and the set of people
> while using OWL machinery for supplying the set if you want to. Only valid
> in RDFS/OWL Full of course.

So my question for the wider community is to ask why
foaf:membershipClass remains largely un-used...

My suspicion is that those folk who are most into doing rich
class-centric reasoning, prefer a DL-based toolset and find FOAF and
real world FOAF data too scruffy for that environment. But I'd hope it
should be possible to take data from the messy quad-centric linked
data scene and project out clean DL-friendly chunks for class
reasoning.

Natural ways of grouping/classing people include by skills, employers,
roles, linkage to some focal individual (eg. "*my* co-authors), and by
geography. On Facebook or Twitter/StatusNet for example I might often
have something like an event cross my desk that I'd prefer to pass on
to a "tech people in Bristol" or "tech people in Amsterdam". There is
no way I'll maintain by hand such a people list in those sites, but
since everyone has APIs these days it is quite feasible that an
RDF/OWL based app could sort-and-categorise my contacts locally, then
push those lists up to the social network sites where they can be
used.

Maybe these are even now still early days...?

Dan


More information about the foaf-dev mailing list