RE: [ALL] editors draft of simple part-whole note ready for review

Alan thanks. I have annotated your replies below, BLUE with >>>>

 

  _____  

From: Alan Rector [mailto:rector@cs.man.ac.uk] 
Sent: Thursday, April 07, 2005 4:26 AM
To: mcdaniel@adobe.com
Cc: 'Christopher Welty'; public-swbp-wg@w3.org
Subject: Re: [ALL] editors draft of simple part-whole note ready for review

 

Bill 

Many thanks.  I'll clean up the typos, missing bits etc. mentioned.
Hopefully without adding too many new ones.  Also now that we have agreed
the examples, I'll generate all the syntaxes and complete code etc. to go
along with it. 

On the substantive issues. 

*    partOf_directly vs directPartOf.  In practice for readability in
systems, especially when there are several variants of partOf - to be
discussed in a later note - putting the "direct" first makes it harder to
read, especially if things start to run off the edges of windows etc. Also
if there are other transitive relations where one wants to have "direct"
subproperties, it means they are harder to tell apart.  I know this is
pragmatic rather than principled, but I think it is a real consideration.

>>>I Understand.  I was looking for a better way to make the names more
human understandable, but you make a good point.

 

* Page 8. "Issue of locus"/"distinguishing between kindOf and partOf.  I am
not sure what more to say here. Suggestions welcome. 

>>> I will review and see if I can come up with some thoughts

*  Reflexive_parts and the comments on the partOf relation being reflexive.
For better or for worse, the classic mereologic axioms are formulated as
reflexivity, transitivity, and antisymmetry.  We might add a caveat that
they run counter to common usage in some cases, but I think we must
reference what is standard practice for a large segment of the community.
We might also put a note on that subrelations of partOf which satisfy only
some of these properties will be discussed in a future "less simple" note on
part-whole relations. 

>>>I think adding the note would be good. Just to address the issue that
there are OTHER ways of looking at this and we will consider and discuss
them later.

*    On deriving hasPart from isPartOf or vice versa.  It is true that this
can be done, but the semantics are tricky.  It does not follow from all As
are parts of some B that all Bs have parts some A - as pointed out in the
paper. The implied partial inverse is that "some Bs have As as parts" - and
even that isn't strictly implied unless we know that some As exist.
Perhaps this point needs a bit more expansion.  (As a technical aside not
for the paper, finding the least bad approximations for these cases using
multiple models or weaker reasoners is a major issue for us at the moment. I
would be pleased to hear from anyone doing or knowing of work in this area )


*    Other relations often confused with partonomy - this section has got
moved about and references to it displaced.  Will fix. I'll also add a
sentence and ref to it in the introduction. Likewise for the "flavours of
part of". 

 

>>>I think these last two points DO need a bit more explanation.  The issue
of temporality, monotonicity and such become more important when considered
in a mereological sense. But this may not be the place to explore them
except to say that they are or are not part of the Part/Whole issue for us
and to say we will consider them in a later note.

Regards 

Bill