Re: Making Rules Out of RDF-Graphs (Re: What is an RDF Query?)

>  > >You might well Skolemize it like that internally (you'd have to if you
>>  >were giving it to a Horn logic engine), but if you require users to do
>>  >it, then the rule-conclusion language needs to have functions in it.
>>  >I'm suggesting that we try making RDF rules out of less-expressive
>>  >stuff (to use the technical term) than that.
>>
>>  Why? (Not that there might not be valid reasons, but I wonder what
>>  yours are. )
>>
>>  I think there has been too much emphasis on sticking to very
>>  inexpressive languages. More expressive languages are often *easier*
>>  for users (though admittedly not for inference engines; but I really
>>  do not think that the kinds of ontologies we are going to see on the
>>  semantic web in the near future are going to tax the resources of a
>>  modern inference engine for full FOL, to be honest.)
>
>One of my main reasons may be a fantasy of this stuff being used for
>real distributed programming, where code is moving around, and having
>that code be declarative (eg pure prolog instead of java byte code).
>That's kind of off-topic here, though.

I really would suggest not mixing up logic-as-code with 
logic-as-ontology-language. In an ideal world they might coincide, 
but that's like waiting for maglev trains. (The trouble is that 
'little' hacks which are useful on one side, eg closed-world 
negation, tend to blow the other side out of the water. Its very hard 
to satisfy two masters at the same time.)

>Here, ... I dunno.   I guess ease of implementation is actually
>important.  I think techies who have never really done much with logic
>system will have an easier time thinking about Horn logic.  
>
>I suppose there's a very important question in how the different
>logics might play together.

Not just one question, an entire field of research. Eg check out 
'fibered logics' for one idea.

>  Tim B-L seems to have a vision for this
>which I don't understand.
>
>>  Well, RDF is too fixed to alter in this way now, but RDF+ could allow
>>  this extension with really minimal change either to the syntax or the
>>  inference engines that are being built. Once you have a run-time
>>  variable binder and some kind of search engine that can switch
>>  bindings between paths, adding functional terms is just a matter of
>>  extending the binding code to be recursive. Prolog hackers have some
>>  blindingly fast unifiers which do this very effectively.
>
>Yeah, I just don't know anything about the path to "RDF+".

OK, let me suggest one. Take RDF and note that it's isomorphic to a 
conjunctive/existential logic with only simple names and binary 
relations. Now, just relax the restriction to simple names: allow 
terms constructed from binary function symbols to be bound to 
variables in queries (and used in the statement of rules.) Considered 
as an assertion language this is such a small change its almost 
invisible, but it makes a big difference to what the rules can do.

Pat
-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Tuesday, 18 September 2001 12:01:12 UTC