Search Mailing List Archives

Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

cell component ontology

Larry.Hunter at Larry.Hunter at
Tue Nov 8 08:22:37 PST 2005

 > I agree completely with the second paragraph, and with its
 > implication that an ontology should represent classes (better
 > called 'types' or 'kinds') in reality. 

Barry, I smile happily to myself whenever you say you agree with me
completely.  Us mere mortals are not so often blessed by formal
semanticists :-)

Just to recap, for public, shared ontologies, I see no advantage to
ever including instances.  See below for what happens when we expand
ontologies to knowledge-bases, which may include representations of
specific observations (and therefore instances). 
 > However, I still have problems with the first paragraph, because I
 > still do not know how the word 'instance' is to be understood.

 > An instance is, surely, (for example) this cell nucleus visible in 
 > the Petri dish on your bench, here, now. But then to replace a 
 > representation of (the class) cell nucleus in an ontology with a 
 > representation of exactly this one single cell nucleus would not be 
 > useful for any purpose.

I am perhaps guilty of having used the term "ontology" a bit more
loosely than you would prefer, not making a strict division between
ontology and knowledge-base in that paragraph.  What I meant was that
there are situations when I might want to include assertions in a
knowledge-base that pertain to particulars, for example forming a
representation of "the nucleus visible in the Petri dish on my bench,
here, now is abnormally large".  In the overall case of BioCyc, for
example, Peter was trying to model assertions of that sort in addition
to more general assertions.  The point I was trying to make in
agreeing with him is that where one decides to draw the line and
declare a representation to be able a "particular instance" rather
than a class or type is a modeler's choice.  Choices appropriate for
some applications may not be appropriate for others.

 > If this reading of 'instance' is correct, then every class has
 > instances -- not just classes treated as leaf nodes (i.e. classes
 > with no children in the ontology), but even the various parent and
 > grandparent classes.

Yes, I agree with this.  It is clearly possible to have a use for a
representation of an instance of a "cell" as well as an instance of a



This message is from the GOFriends moderated mailing list.  A list of public
announcements and discussion of the Gene Ontology (GO) project.
Problems with the list?           E-mail: owner-gofriends at
Subscribing   send   "subscribe"   to   gofriends-request at
Unsubscribing send   "unsubscribe"  to  gofriends-request at

More information about the go-friends mailing list