W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > March 2007

Re: [Fwd: Re: cell types, brain regions mentioned in gensat]

From: Pat Hayes <phayes@ihmc.us>
Date: Mon, 5 Mar 2007 14:52:31 -0600
Message-Id: <p0623092fc2123338548d@[10.100.0.26]>
To: Bijan Parsia <bparsia@cs.man.ac.uk>
Cc: "Kashyap, Vipul" <VKASHYAP1@PARTNERS.ORG>, public-semweb-lifesci@w3.org

>On 5 Mar 2007, at 17:37, Kashyap, Vipul wrote:
>
>>
>>Thanks for the response, Pat, That's quite a relief to hear.
>>
>>________________________________
>>
>>>>EquivalentClass(X, intersectionOf(Y, Restriction(locatedinBrainRegion
>>>>allValuesFrom(Z))
>>>>
>>>>Is this what you had in mind?
>>>>I guess, one could define macros (using the lisp backquote mechanism) where
>>>>these variables could be plugges in at run time and then sent to the OWL
>>>>reasoner.
>>>>
>>>>As long as these variables are instantiated, you are OK, but without
>>>>instantiation you end up in higher order
>>>>logics, which might be beyond OWL full.
>>>>So that will not be feasible.
>>
>>Whoa. This isn't higher-order, even with
>>variables. In fact its within OWL-DL, which is a
>>tractable subcase of first-order.
>>
>>[VK]... The fact that X, Y, and Z range over classes (and laziness 
>>:)) is what
>>made me think that.
>>Glad to heat that this is not the case.
>
>If those do range over classes, then the above expression is, 
>indeed, at least syntactically second order and definitely not in 
>OWL-DL.

Whoops. Yes, of course. I thought those were what one might call 
metasyntactic variables, not quantified. I reacted too quickly. 
Thanks for the correction, Bijan.

Pat

>
>Depending on exactly what you wanted to do with that, especialy if 
>the procedural account above is what you want, then this is fairly 
>straightforward.
>
>E.g., is the above a *query*? As long as the variables range over 
>atomic classes (i.e., named classes), it's pretty easy. You do have 
>to loop over all the tuples and check if the resultant expression is 
>entailed (i.e., you are treating it as a certain sort of axiom 
>schema). You can do a better by being clever about your testing 
>(e.g., if replacing X with C1 is entailed for some given value of Y 
>and Z, and C2 is known to be disjoint with C1, then you don't need 
>to do a full test iwth C2 for X for those Ys and Zs.) Inside the 
>reasoner, you could make use of other optimizations which are used 
>in classification.
>
>Cheers,
>Bijan.


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Monday, 5 March 2007 20:53:28 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:52:30 UTC