[Person-ontology] Methodology using an upper ontology

[Aldo Gangemi]

That's it. Thanks for the moderate answer, just a few clarifications:  
I was in a hurry

Il giorno 08/nov/07, alle ore 18:14, Pat Hayes ha scritto:

>> Pat, it is a bit unfair that, since you have so much time to  
>> criticize reusable ontologies
>
> Reusable =/= upper, in general. Im all in favor of reusability.   
> And BTW, I have VERY little spare time, but dashing off emails  
> (like this) is quick, and I can often do it while attending  
> meetings (as I am now) which I don't have to listen too all of  
> (like now). And, regrettably, I can't get funding to support the  
> development of ontologies or even ontology frameworks in this  
> country at present. No US government agency (AFAIK) is supporting  
> such work. (I have been trying for years.)
>

This is sad indeed, given that you have been one of the initiators of  
good knowledge engineering. Sure in the next project involving  
serious modelling I'll try to involve you in some way, even remotely.

>>
>> The same argument you make can be made at any level of detail,  
>> therefore, your (and Lenat's) claim is just a slogan.
>
> Well, Lenat certainly speaks from experience, and I from  
> observation. The upper levels are usually taken to be more  
> "authoritative" than the lower ones, is my point. Not perhaps by  
> the people (like you) who actually helped create them, but by many,  
> perhaps most, of those who have more limited scopes in mind. So if  
> they (the upper levels) clash with important intuitions at the  
> level at which information really needs to be represented, those  
> intuitions are modified rather than the upper-level concept  
> modified or abandoned. Im not saying this is always bad; but I  
> think its often is, and the harm is ignored or under-rated by  
> proponents of universal upper-level ontologies as a guide, or even a
>

Yes, you told this many times, and I basically agree; on the other  
hand, I see your point as a methodological issue ("avoid distortion  
induced by a reusable component"), rather than an obstacle to the  
reusability of foundational, upper, or similar ontologies.

For me, each ontology must have an associated task at design time  
(similar to artifacts in general). Foundational ontologies make no  
difference: their task is to provide modelling solutions as  
invariants ('knowledge patterns') across different modelling problems.
Consider that this is the same task all ontologies try to help  
achieving (together with their proper one, e.g. aiding expert  
searches or finding new biochemical pathways), but at different  
levels of detail: foundational, domain-generic, domain-specific,  
cross-domain, etc. Someone may want to be "guided" by foundational  
patterns, some others by domain-generic, someone else by domain- 
specific, or even by nothing at all. That's life.

IMHO, the real issue here, maybe the only one worth discussing is:  
can we find an ordering among modelling-problem-solving patterns? For  
example, is there any similarity, representable e.g. in a partial  
order, between: "having a workplace address in a period of the year"  
and "being located in a place at a time"?

DOLCE contains several of these invariants with a very general scope,  
but not all possible ones, and some of them can of course conflict  
with other invariants encoded in other ontologies, as you point out,  
for example the continuant-occurrent distinction, the transitivity of  
partOf (but we also have intransitive partOf, called componentOf), etc.
I am aware of that, and I'm convinced that DOLCE and other  
foundational ontologies are "coherent containers" (or "frameworks" as  
you said time ago) of several patterns, and their actual usage cannot  
be prescriptive, or authoritative in the bad sense, while they are  
best used when someone takes appropriate, non-conflicting pieces, and  
reuses them in projects by specializing, composing, etc. Distortion  
is much easier to avoid if the reused piece is small.

> Quite. Please don't let me stop you doing your excellent work.  
> (Really, no sarcasm!)
>

Precious comment, thanks Pat :)
Aldo



_____________________________________

Aldo Gangemi

Senior Researcher
Laboratory for Applied Ontology
Institute for Cognitive Sciences and Technology
National Research Council (ISTC-CNR)
Via Nomentana 56, 00161, Roma, Italy
Tel: +390644161535
Fax: +390644161513
aldo.gangemi at istc.cnr.it

http://www.loa-cnr.it/gangemi.html

icq# 108370336

skype aldogangemi



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://idcommons.net/pipermail/person-ontology/attachments/20071109/c9b66760/attachment.htm