Re: [foaf-protocols] ACL Ontology and Discussion

Joe Presbrey wrote:
> On Tue, Apr 20, 2010 at 2:49 PM, Story Henry <henry.story@bblfish.net> wrote:
>> I think you can get what you want by using owl reasoning
> 
> authz_webid can't currently reason about OWL.  I currently get this
> done with roqet+rapper in a Makefile near my ACLs and FOAF using
> CONSTRUCT.  See http://presbrey.pastebin.com/wyPSdU36 if you're
> interested.

ty for the pointer, and likewise can't reason with owl (and in many ways
don't want to, want ACL to be as quick and painless as possible for both
man and machine.

> I run 'make' after I change my FOAF (adding or dropping foaf:knows)
> and my FOAF and ACLs are always up to date.  I've also toyed with
> activating the script from crontab, new user signup, pubsubhuhbbub
> push, and others.
> 
> I had an agentASK-like predicate in an early version of authz_webid.
> It was the ugliest thing: a url-encoded SPARQL statement supporting
> $ENV substitutions but it was awesome stuff.  The expressiveness of
> SPARQL ASK and boolean result made it seem like the obvious Right Way
> to delegate authz to a full/custom semweb stack that CAN reason (and
> cache!).

very good idea, I like that :) [re ASK]

>> This does however bring up the issue of how to say "read access for
>> anybody with a webid" - is there any wild card syntax that could be used
>> for acl:agent or suchlike?
> 
> This is precisely what I use acl:agentClass to do (namely: []
> acl:agentClass foaf:Agent) assuming that you can't be an Agent without
> having a WebID and don't care about the distinctions between client,
> server, agent, and person.

Snap (picked up the example from your paper!) was more thinking about
staying clear of owl reasoning inferencing etc, but as henry has pointed
out using restrictions etc is a decent way of making things happen -
even if I can't personally support it at the minute.

Nathan

Received on Tuesday, 20 April 2010 19:37:23 UTC