[seek-kr-sms] growl-paper

[Serguei Krivov]

            Hi Jianting,
            I apologize  for the delayed response.
 
Mathematically, if A belongs to B then we can say that the intersection
of A and B EQUALS to A. I didn't see why equivalence relation can not be
used here, if we treat operation and parameter equally, not just
parameter along. 
 
 
           You mean that if A is a subset of B then intersection of A
and B is equal to A. Yes, but still  A is not equal to B.
            A remains to be a subset of B. Similarly   OSTOOLS  is in no
way equivalent to COPY or FORMAT.
           
 
Now let's turn to programming issues. Suppose we have a bunch of such
relations (whether we call them equivalence  or overlap), how do we
retrieve them from database?. An application might be, take Web services
for example, if one is not available, Kepler might be able to suggest
another, based on the pre-stored relationship. I was thinking about
using Jena, but the representation might be too complex to do RDF query.
Any further suggestions? 
 
 
       If we have in ontology the relation "overlaps" (I would still
look for better word) then it will transpire into set of triples:
       (X overlaps Y), where X and Y are instances of actors. Now if you
wish to retrive all actor instances that overlaps with some actor
(instance ) A then corresponding query in Jena's RDQL will be something
like:
 
SELECT ?x
WHERE(  ?x <your-namespace#overlaps> <your-namespace#A>)
 
Far more complex queries are possible of course. Here are the link to
RDQL documents:     http://www.hpl.hp.com/semweb/doc/tutorial/RDQL/
 
ciao,
serguei
         
 
 
Thanks a lot
 
Jianting
 
----- Original Message ----- 
From: Serguei <mailto:Serguei.Krivov at uvm.edu>  Krivov 
To: 'Jianting Zhang' <mailto:jzhang at lternet.edu>  
Cc: seek-kr-sms at ecoinformatics.org 
Sent: Monday, January 03, 2005 9:57 AM
Subject: RE: [seek-kr-sms] growl-paper
 
 
Let me give you an example to further illustrate my question. 
 
Think about I'm going to combine two OS commands "COPY","FORMAT" to make
a more powerful command "OSTOOL". If the first parameter of OSTOOL is
"COPY" then it does COPY and similar thing happens to "FORMAT". 
 
So it is not relation of equivalence, it is the fact that under certain
combination of parameters OSTOOLS overlaps with COPY, under certain
other combination it overlaps with FORMAT. Here we have that OSTOOL is a
procedure like the following:
 
OSTOOL(arg2) {
   if(arg1=="copy")
            return  COPY(arg2);
   else   if(arg1=="format")
             return  FORMAT(arg2);
 
}
 
 
 I think it would be not possible to express this precicely in OWL.
(otherwise we would be able to use owl as a programming language) What
is possible is to express certain relation between input parameters of
OSTOOL, COPY and FORMAT. Say the following diagram says that input type
for OSTOOLS represented by class O2-Input is exactly equivalent to class
O1-Input + one extra argument "arg1", and  O1-Input is an input type for
COPY and FORMAT:

 
 
 
 
 
 
 
 
It seems that, in your two solutions, you are separating operation and
Input/output (either one I/O is a sub-property of another or one
operation is a sub-property of another, but not combining I/O and
operation to make an equivalent relation). 
 
I believe that diagram above represents the fact that input for OSTOOLS
is a combination of input for COPY (and FORMAT) with one extra
parameter. Consider usage of operation "Intersection" of things that
have some attributes ("exist" property restrictions in OWL) when you
have such combination of parameters.  When there are different options
for intput/output types, the  Union of types would express such non
deterministic type.
 
serguei
 
Let's see whether we have other choices. 
 
Jianting
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/seek-kr-sms/attachments/20050106/96bf19c5/attachment.htm