Message 12123 of the SUO list

Subject: Re: Model of Activity and Action in SUO ontology
Date: Fri, 12 Dec 2003 12:19:17 -0500

From: Philippe Martin <phmartin at phmartin dot info>
In reply to: msg12086 by Richard Cooper and msg12107 by Patrick Cassidy


Most of the Lifecycle Integration Ontology (LIS; refered to by Matthew West
two weeks ago) is now integrated in the Multi-Source Ontology of WebKB-2.
See http://www.webkb.org/kb/top/lis.html  for the documentation, the 
representations and as a way to begin the browsing of the categories.
As detailed in this documentation a choice had to be made regarding the
representation of LIS relationships. If I am asked to, I will use
the alternative choice and also check that I have fully integrated the
LIS ontology. If I am not asked, I'll assume that this work
is sufficient and/or that no one is interested in it.


Richard Cooper wrote:
> Is there anything in CGs, WordNet, SUO, OpenCyc,
> ... that corresponds to this very simple concept
> of activity and actions on objects?

As noted in previous answers, this is the notion of process,
or more generally, of situation (state or process), both of which are present
in most ontologies. Here is a small excerpt from the ontology of WebKB-2
that gives some inter-connections (categories from WordNet begin by 
"wn#"; "__" separate alternative names (the 1st name being the key name)).
For more details, see 
http://www.webkb.org/bin/categSearch.cgi?categ=pm%23situation&recursLink=%3E&depth=2&hyperlinks

pm#situation  (^something that "occurs" in a real/imaginary region
                of time and space^)
 / pm#entity  pm#spatial_object  dolce#abstract  dolce#endurant
 < pm#thing suo#physical dolce#abstract_or_perdurant dolce#endurant_or_perdurant,
 = dolce#perdurant  suo#process
 > pm#state  (^situation not changing and not making a change during
               a given period of time^)
   >  wn#state  (^the way something is with respect to its main attributes^)
   >  wn#feeling (^the state of experiencing affective and emotional states^)
   >  pm#state_playing_some_role
 > pm#process  (^situation that makes a change during some period of time^)
   >  pm#event  (^process considered instantaneous from some viewpoint;
                  classification under this category is application-dependant^)
   >  pm#problem_solving_process  (^cognitive activity to solve a problem^)
   >  #cognitive_process__process__operation__cognitive_operation__act
   >  wn#unconscious_process__process  (^a mental process that you are ...^)
   >  wn#human_action__act__human_activity  (^something that people do ...^)
   >  pm#process_playing_a_role

This ontology also includes a quite hierarchy of 70 types of basic 
relations from a situation (e.g. those to a time measure, those to
another situation, and the common thematic relation); their names are quite
intuitive and such relation types are very useful when representing natural
language sentences. For details, see
http://www.webkb.org/bin/categSearch.cgi?categ=pm%23relation_from_situation&recursLink=%3E&hyperlinks


Patrick Cassidy wrote:
> Since I have severe skepticism that any kind of universal agreement will be
> reached on anything in an upper ontology, I doubt that any progress can be
> made without voting and of necessity displeasing some people. We can aim
> for a lattice of theories, but if it is not to be a mere list of 
> incompatible ontologies, and is to represent some kind of consensus
> among more than two different viewpoints, at least one ontology
> must be built which represents some agreement between different
> viewpoints.  That of necessity requires choosing, which (given
> the demonstrated inability to reach universal agreement) means
> that some issues must be resolved by voting.

Yes and no. Some choices must sometimes be made (such as the one I cited above
because LIS was apparently not designed with a knowledge representation
language but with an object-oriented or entity-relationship language/approach)
but that does not mean that people must agree or be displeased in order to reach
a unique integrated ontology. The ontology in WebKB-2 seems a good example (or
counter-example of the quoted statement), and any Web user can add new 
categories or add names to them if she thinks that the existing ones are
not enough. She cannot remove what she has not created but such removals
would not actually make sense (as explained in my message of December 2:
http://www.webkb.org/doc/onSUOlist/m031202Schoening.html).
Thus, voting is NOT required. However, since not everone is allowed to add
new categories with the prefix "suo#", if the SUO group really wants those
categories to have the most intuitive key names, voting may be a solution.


> With a modicum of goodwill on the part of participants, I think this will
> actually work quite well. I am convinced that nothing will be achieved
> unless such a method is adopted.

That seems achievable for a small number of categories, but not for a 
thousand. Voting for links between categories (between the SUO ones or with
categories from other users/sources) would be more important than voting for
their names, but would require more involvement from the participants, and
then more efforts to synthesize the results.
I doubt this will be done and, as noted above, I do not see why this is
necessary.

Philippe