[foaf-dev] Suggestion for WOT vocab - PGP Word List fingerprints
danbrickley at gmail.com
Tue Jul 29 20:41:17 BST 2008
Briefest of comments for now. I'm in transit and not 100% sure this
mail will get through (and yes, Dreamhost IMAP/POP is wedged again so
danbri at danbri.org doesn't work).
OK, to pop up to the 'is this redundant' level:
We *do* have some redundant vocab in FOAF. Specifically, a few
inverses. In general I like to avoid that, but with foaf:depicts and
foaf:depiction there was enough benefit to folk who cared about the
XML encoding of the RDF that I thought the redundancy bearable. Note
that this was also designed before RDF Core finished (including before
rdf:nodeID existed in the XML syntax), and when there were no non-XML
syntaxes for RDF around (nor any GRDDL, RDFa, SPARQL in the works).
So in general I cringe at adding clear redundancies; I don't want to
have to think up a name for every property's inverse. Nor get those
translated (although that's an interesting project).
This kind of redundancy is a little different, since the equivalency
(like with :mbox_sha1sum) is hard to express in universally understood
machine form. Perhaps in a few years everyone will have common .js,
SPARQLScript, or RIF-based approaches to expressing these data
patterns. But for now, they don't. It's also a little related to the
:age vs :dateOfBirth discussion.
The fundamental distinction that can help us here is that of FOAF
document versus the FOAF vocabulary.
We have this colloquial notion of FOAF files, things published in the
Web (usually in public, but increasingly mediated via openid/oauth
We have the FOAF vocabulary; a dictionary of terms. The core 'thing
that is FOAF'.
These two are often conflated, but they're not quite the same. I think
we can be liberal about what goes into FOAF, while conservative about
raising expectations for what people publish and find in typical FOAF
files. It might be that the FOAF vocab for eg has :dateOfBirth, yet we
only publish an age in years on the public Web.
If we're too restrictive about what goes into RDF vocabularies, I fear
it's a little like 1984; controlling what is say-able.
If we publish too many variant properties and patterns in the public
Web, the data will be harder to consume by a general consumer.
There's a balance to be had there. I'm happy putting this kind of data
into the FOAF vocabulary if it can be shown that there are producers
and consumers of it already in the Web community. This doesn't mean
that it is prudent to use the property in all circumstances. Different
species of FOAF file/document may or may not make use of the
vocabulary. But we need more subtle ways of managing expectations
about what appears in FOAF files than simply keeping the properties
from existing in RDF at all.
Hope that makes some sense. Gotta run for a train again, oof...
Earle, how much of this data is out there in the word list form? Tools
that produce and consume it? Evidence that people actually use this?
(rather than it being a high class geek code?).
More information about the foaf-dev