ontolog-forum
[Top] [All Lists]

Re: Representation - KIF vs Protege [was Re: [ontolog-forum] Personas

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>, "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Adam Pease <adampease@xxxxxxxxxxxxx>
Date: Thu, 07 Aug 2003 21:21:16 -0700
Message-id: <5.0.0.25.0.20030807211851.024b7008@xxxxxxxxxxxxxxxxxx>
Hi Peter,    (01)

At 02:13 PM 8/7/2003 -0700, Peter P. Yim wrote:    (02)

>Adam Pease wrote Thu, 07 Aug 2003 09:48:57 -0700:
> > Peter,
> >   If the model is created in KIF and then translated to Protege this
> > might work, but if the model is created in Protege and translated to > 
> KIF you'll have the problems I mentioned analogous to writing C
> > style in C++.
>
>Sounds good. I was, in fact, looking forward to more discussion on "how" 
>we can end up with both KIF and Protégé. Therefore, what you point out 
>here is most relevant indeed.
>
> >   It's good that Protege is familiar, and has a large user
> > community, but that line of argument could be extended to XML.
> > More people are familiar with XML.  It has a larger user base
> > and tool set, so why not just create XML rather than worry
> > about ontologies?
>&
> > if the argument for a frame representation is that it has a larger
> > user base, or more tools (which is the argument I believe that
> > Peter was making), than that same argument could be made ad
> > absurdum to argue for just sticking with XML, and not trying to
> > create an ontology.
>
>I was using that argument to support my case that "Our solution will need 
>to be adopted and applied, if it were to even come close to being 
>successful", not to be confused with condoning that Protégé (or KIF, or 
>XML) is totally adequate, or optimal, on its own.    (03)

ok, thanks for clarifying    (04)

>My consideration extends beyond the ability to represent the knowledge ... 
>and is supposed to cover diffusion, user acceptance, and other user 
>adoption related "soft" (social, psychological, ...) issues as well.    (05)

yes, that's important too    (06)

>As mentioned, we would need to
>    ... "tackle both the science, as well as coming up with the 
> engineering solution that is "good enough" ..."
>
>I am sincerely hoping that we we can achieve the precision (as in the 
>"science") as well as being able to compromise, optimize, and DELIVER a 
>standard business ontology to the masses/industry (as in the "engineering").
>
>The approach to achieving one or the other may be different, but that 
>should not preclude us from considering both, and to make an honest 
>attempt to achieving both at the same time. It's going to be a challenge 
>... but then, we've got some *very* good people in the community too!  :-)    (07)

I think we can accommodate both by creating a formal model in KIF and then 
translating out what's expressible in Protege and XML.  If we work from 
more expressive to less expressive, I think we'll wind up with a good 
model.  If we work the other way around my belief is that there will be a 
lot of wasted effort, and a poor result.    (08)

Adam    (09)

>Regards,
>PPY
>--
>
>
>Adam Pease wrote Thu, 07 Aug 2003 10:28:15 -0700:
>>Kurt,
>>At 10:18 AM 8/7/2003 -0700, Kurt Conrad wrote:
>>
>>>At 2003-08-07 09:48 -0700, Adam Pease wrote:
>>>
>>>>More people are familiar with XML.  It has a larger user base and tool 
>>>>set, so why not just create XML rather than worry about ontologies?
>>>
>>>
>>>
>>>While I find this argument to be dismissive, it may point to the core 
>>>issues underlying the current debate.
>>
>>I didn't intend the comment to be dismissive.  The point was that if the 
>>argument for a frame representation is that it has a larger user base, or 
>>more tools (which is the argument I believe that Peter was making), than 
>>that same argument could be made ad absurdum to argue for just sticking 
>>with XML, and not trying to create an ontology.
>>
>>>The question is not whether or not to create a XML representation.
>>>That decision has been made (by another, larger group).  The vast 
>>>majority of the work has been done.  And the XML version will almost 
>>>certainly be ready before the fully-formalized ontological version.
>>>Why, because more people are familiar with XML; it has a larger user 
>>>base and tool set; and -- perhaps more fundamentally -- it is an 
>>>appropriate fit for the targeted utilization scenarios.
>>
>>I agree
>>
>>> From my perspective, it seems clear that the "UBL ontology" will take a 
>>> variety of forms.
>>
>>I agree.  In prior work that I've mentioned, I've created a formal 
>>ontology in logic and then extracted what can be expressed in a 
>>relational database.
>>
>>>   Each representation will have relative strengths and weaknesses.  In 
>>> sort, logical expressiveness is only one measure of value.  Clearly, 
>>> the XML representation of UBL is well-optimized for some uses, but weak 
>>> -- or even inadequate -- for others.
>>
>>I agree
>>
>>>  That was the basis for our decision to create a formalized ontology 
>>> based on the UBL concepts.  What other representations should be 
>>> produced?  I don't know.
>>
>>>While I can follow this discussion at a general level, I don't have a 
>>>sufficient grasp of the subtleties to have strong opinions regarding 
>>>ontological tools and languages.  In fact, about the only bias that I 
>>>have is one of creeping incrementalism: Start your semantic 
>>>formalization simply, and add complexity and richness as your 
>>>requirements, understanding, and resources warrant it.
>>>
>>> From my vantage point, I'm seeing indications that a KIF representation 
>>> might not be the best fit for all usage and interaction scenarios.
>>
>>I agree
>>
>>>  This observation does not imply that a KIF representation should not 
>>> be produced, or even how and when.  Rather, it suggests that to meet a 
>>> broad set of user requirements (perhaps, even, including the needs of 
>>> the Ontolog modeling community) work on KIF needs to be balanced with 
>>> work on other technology platforms.
>>
>>I don't see that that follows.  The work can be done in logic, in order 
>>to have a deep analysis of the domain with formal axioms, and then less 
>>expressive versions can be extracted for different implementation.
>>
>>>I guess this leads me to a corollary question: Does the decision to 
>>>produce an ontology in KIF effectively preclude working with any other 
>>>languages and/or platforms?
>>
>>No, but I believe the reverse is at least partially true.  If we choose 
>>to develop the ontology as a frame system, it will have to duplicate a 
>>lot of the content that already exists in SUMO, and then won't be 
>>transferable to a formal ontology in logic later.
>>
>>>  If so, are the tradeoffs acceptable?
>>>
>>>Or, looking at it from the requirements side: What representations are 
>>>needed?  Is KIF adequate to handle all of the expected utilization scenarios?
>>
>>KIF (or more generally first order logic) will certainly not be the 
>>appropriate "object language" for many implementations, but it's the best 
>>language available for formalizing a domain.
>>Adam
>>
>>>/s/ kwc 2003.08.07 10:17
>
>
>Adam Pease wrote Thu, 07 Aug 2003 09:48:57 -0700:
> > Peter,
> >   If the model is created in KIF and then translated to Protege this
> > might work, but if the model is created in Protege and translated to > 
> KIF you'll have the problems I mentioned analogous to writing C
> > style in C++.
> >   It's good that Protege is familiar, and has a large user
> > community,
> > but that line of argument could be extended to XML.  More people are
> > familiar with XML.  It has a larger user base and tool set, so why
> > not just create XML rather than worry about ontologies?
> >
> > Adam
> >
> > At 09:41 AM 8/7/2003 -0700, Peter P. Yim wrote:
> >
> >> I will not question the appropriateness of KIF, in terms of formality
> >> and expressiveness. On the other hand, we have Protégé, a tool
> >> familiar with, probably, by the most number of ontology practitioners
> >> (albeit small as they are already) as substantiated by our own survey
> >> (see:
> >> http://ontolog.cim3.net/forum//ontolog-forum/2002-11/msg00035.html )
> >> and Protégé's user statistics (see:
> >> 
> 
>http://purpleslurple.cim3.org/ps.php?theurl=http://protege.stanford.edu/#purp169
> 
>
> >> ).
> >>
> >> Referring, again, to both our [ontolog-forum] charter (see:
> >> http://ontolog.cim3.net/cgi-bin/wiki.pl#nid011 ),
> >> and that of UBL's (see:
> >> http://www.oasis-open.org/committees/ubl/charter.php ),
> >> we are out to contribute to solving an e-business industry
> >> standardization problem. Our solution will need to be adopted and
> >> applied, if it were to even come close to being successful.
> >>
> >> As such, I am in favor of having our work represented (at least) both
> >> in KIF and in Protégé (I actually thought that was what we had agreed
> >> upon before)  -- with both being our normative deliverable.  We need
> >> to leverage the Protégé users, developers, plug-ins, ... etc.
> >>
> >> Therefore, my suggestion is that we should focus the conversation, not
> >> on whether its should be KIF or Protégé, but rather, how can we
> >> deliver BOTH, and in so doing tackle both the science, as well as
> >> coming up with the engineering solution that is "good enough" to be an
> >> applicable industry standard -- and, be able to help the industry
> >> migrate to that solution with ease.
> >>
> >> Regards,
> >> PPY
> >>
> >> P.S. May I request that we all try to make sure that the discussion is
> >> consistent with the subject line -- this will greatly facilitate
> >> future retrieval/re-use of the knowledge, at least by humans. :-)  -ppy
> >> --
>
>_________________________________________________________________
>Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>Subscribe/Unsubscribe/Config: 
>http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>Shared Files: http://ontolog.cim3.net/file/
>Community Wiki: http://ontolog.cim3.net/wiki/ To Post: 
>mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (010)

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Unsubscribe/Config: 
http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (011)
<Prev in Thread] Current Thread [Next in Thread>