- From: Graham Klyne <GK@ninebynine.org>
- Date: Fri, 23 Feb 2001 23:10:33 +0000
- To: "Sean B. Palmer" <sean@mysterylights.com>
- Cc: <www-rdf-interest@w3.org>
What's the problem here? Given incomplete knowledge, you cannot infer what you need to know. But then you dig a bit further, discover more information, and then you can infer what you need. Seems quite right and proper to me. #g -- At 05:13 PM 2/23/01 +0000, Sean B. Palmer wrote: >Stop me where I'm wrong... > >Let's say we have the following snippet of easily understandable N3:- > > @prefix : <http://infomesh.net/2001/02/ns1/> . > @prefix myns: <http://webns.net/2001/02/ns2/> . > :x :property1 :y . > :x :property1 myns:z . > >Now... :x is linked somehow to :y with :property1. No need to think of >examples, that's all we need to know. Let's also say that it is >*essential* for the machine that uses this schema to know that :x can >take on the property1 of *both* :y and myns:z. Otherwise it won't >work. >Now, what if I take out the second line, and move it elsewhere, but >replace it with another bit of information. > > @prefix : <http://infomesh.net/2001/02/ns1/ > . > @prefix myns: <http://webns.net/2001/02/ns2/> . > :x :property1 :y . > :x :property1 :q . > :q = myns:p . > >My system is now completely broken. It needs to know, for some reason, >that ":x :property1 :y, myns:z .", and this cannot be worked out from >the information above. > >However, what if the Schema at http://webns.net/2001/02/ns2/ contained >the following:- > > @prefix : <http://webns.net/2001/02/ns2/> . > :p = :z . > >In other words, it needs to know that ":x :p1 myns:z". At the moment >it knows that ":x :p1 :q", and also that ":q = myns:p". If my system >could just work out that "myns:p = myns:z", then the thing would work. > >How can it get this knowledge... well (take a deep breath), if my >program dereferenced the URI for the second schema (shock horror), it >finds that it understands what the URI *means* in this context (it >explicitly trusts that server, because I programmed it to). It did >need to understand the HTTP, and the N3 needed to derive that fact, >but at least it now correctly finds out that ":x :property1 :y, myns:z >.", and the dang thing works at last. All because it dereferenced a >URI. > >-- >Kindest Regards, >Sean B. Palmer >@prefix : <http://webns.net/roughterms/> . >[ :name "Sean B. Palmer" ] :hasHomepage <http://infomesh.net/sbp/> . ------------ Graham Klyne GK@NineByNine.org
Received on Friday, 23 February 2001 18:14:05 UTC