Re: rdf inclusion

>Pat,
>
>What you now saying is that so long as they publish the 
>algorithm/code which they use to derive their conclusions, we will 
>all be ok ... or is there something I am missing?

Not algorithm/code, but inference rules (actually, the whole 
derivation, I would prefer). It has to be checkable by the reader, 
and it ought to be easier to check than to derive, usually. The 
reader shouldn't have to re-derive it, and in any case I don't want 
to have to check their *code* for correctness.
This assumes that there is an interchange format for derivations, of 
course, which there isn't yet, but I think there ought to be :-). In 
the shorter term, a derivation could just be something like 'derived 
using RDFS from <uri of KB> assuming <uri of KB> is closed with 
respect to <uri of namespace>' which treats all of RDFS inference as 
a single inference rule. This would be acceptable since RDFS 
entailment is so easy to check.

Pat

>
>guha
>
>
>
>patrick hayes wrote:
>
>>>You know, someone will want to make temporally qualified 
>>>statements or (gasp) even defaults. Or maybe some rules to deal 
>>>with your favorite (the frame problem). At which point, all bets 
>>>are off!
>>
>>
>>All bets aren't off, provided they publish the rules they are 
>>using. They can use default reasoning, as long as they (1) say that 
>>they are using it and (2) say which 'worlds' they are taking to be 
>>closed. In summary: people can use any proof methods they like, as 
>>long as they say what they are in enough detail to enable someone 
>>else to check their conclusions (ie to check whether they are happy 
>>with the methods they used to derive the conclusions.) If those 
>>assumptions are part of what gets published as 'input', then we can 
>>allow people to use any proof methods they like, even something as 
>>patently invalid as citing an external authority.


-- 
---------------------------------------------------------------------
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 Wednesday, 29 May 2002 11:00:07 UTC