ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Semantic Web shortcomings [was Re: ANN: GoodRelation

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ron Wheeler <rwheeler@xxxxxxxxxxxxxxxxxxxxx>
Date: Tue, 19 Aug 2008 22:06:27 -0400
Message-id: <48AB7C23.60305@xxxxxxxxxxxxxxxxxxxxx>
Patrick Cassidy wrote:
> Ron,
>   You are still missing the point of the foundation ontology.  It provides
> the conceptual vocabulary to specify the meanings of type of things like
> geographical entities.  It will not be necessary to have an exhaustive list
> of instances, only a few well-known instances of each category is needed to
> make the meaning clear.
>       (01)

>    The methodology would have to be agreed to by the design team before the
> main work starts.  That should take, I estimate, a week for a preliminary
> workshop open to all participants, assuming preparatory discussions.
> Disagreements are settled by voting.  Anyone who can't agree to the adopted
> method can drop out of the project.
>   
And start their own project, which might be superior to the original 
group's proposal. Which takes us back to the current state.
Why should the taxpayer back a losing group? How can the government pick 
a winning group at this point in such a new field?
>    I don't doubt that specialized ontologies can be developed by communities
> of interest as well as by individuals.  But unless they are mapped to a
> common foundation ontology, they will not be able to interoperate with
> accuracy.  Mappings that are only used for highly error-tolerant tasks such
> as search, presenting a human with a list of possibilities, will be easier
> to create but can be useful only in such limited applications.  For
> automated reasoning in error-intolerant applications, post-hoc mapping to
> achieve the required level of accuracy will be more costly than building a
> new ontology in the same domain.  And one will still be left with the
> n-squared complexity of creating multiple such mappings, as opposed to order
> of n for creating new interoperable ontologies by reference to a common
> foundation ontology.
>
>   
> Pat
>
> Patrick Cassidy
> MICRA, Inc.
> 908-561-3416
> cell: 908-565-4053
> cassidy@xxxxxxxxx
>
>
>   
>> -----Original Message-----
>> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
>> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Ron Wheeler
>> Sent: Tuesday, August 19, 2008 8:56 AM
>> To: [ontolog-forum]
>> Subject: Re: [ontolog-forum] Semantic Web shortcomings [was Re: ANN:
>> GoodRelations - The Web Ontology for E-Commerce]
>>
>> You are describing an open source project where each member "who is
>> interested" puts in his/her 2 cents and the core group decides what
>> gets
>> committed.
>> This is difficult to fund with taxpayers' money since there is no one
>> who can be held accountable and no organization who can commit to
>> delivering a pre-defined deliverable.
>>
>> I suspect that while we wait for some rich, disinterested party to come
>> up with $1,000 per item (number tossed about earlier) with no concern
>> for the scope, budget, timeline, qualification of participants, etc. we
>> will produce a lot of subject specific ontologies that have to be
>> integrated.
>>
>> The GoodRelations ontology will likely be very helpful to me at some
>> point and I am going to have to accept what compromises are part of the
>> bargain just as I do when I select an open source package as the basis
>> of an application that I am building.
>>
>> I know that if I pick up a "foundation ontology" that includes a
>> geographical section, I am going to have to expend some effort on it if
>> I want to include every town in Canada where my client has a store. If
>> it was prepared by a European or Asian group, I may have more to do
>> than
>> if the project leader is from Transport Canada or CN Rail.
>>
>> I doubt if the GoodRelations ontology and anyone's geographical
>> ontology
>> is going to cause me a lot of integration problems.
>>
>> OTOH I have done enough with open source to understand that the
>> selection of components has a direct bearing on the amount of work that
>> you have to do and you have to do your due diligence carefully.
>> We will depend on the groups preparing the ontologies to do their jobs
>> well and to form alliances with other groups to deal with conflicts and
>> boundary conditions.
>> We will depend on forums like this to do the peer review and to provide
>> references of combinations that work and ones that do not.
>>
>> Eventually, "ontology stacks" will arise similar to LAMP, where there
>> is
>> a general agreement that the ontologies are "compatible" to some well
>> understood level.
>>
>> At least I hope that this will be true.
>>
>> Ron
>>
>> Patrick Cassidy wrote:
>>     
>>> Ron -
>>>    I thought the point was clear, but I will clarify anyway:
>>> [RW} > I can see how this group would collapse into jurisdictional
>>>       
>> disputes.
>>     
>>>> Who gets to define the medical ontology - drug companies, medical
>>>> equipment companies, HMOs, hospitals, WHO, etc.?
>>>>
>>>>         
>>>    Any member of the project who is interested.  Membership in any
>>>       
>> working
>>     
>>> group should be fully open - no one can feel 'left out' of anything.
>>>    You also seem to focus on a lot of specialized concepts, but that
>>>       
>> is not
>>     
>>> the purpose of developing a common foundation ontology.  A foundation
>>> ontology will contain representations of the most basic primitive
>>>       
>> concepts
>>     
>>> with which all the specialized concepts can be created as
>>>       
>> combinations of
>>     
>>> the basic concepts.  That is the path to interoperability - agree on
>>>       
>> a basic
>>     
>>> set of concepts with which you can specify all the specialized
>>>       
>> concepts of
>>     
>>> interest to you, and in that manner everyone can define the
>>>       
>> specialized
>>     
>>> concepts however they consider it appropriate, using the primitives
>>>       
>> that are
>>     
>>> agreed on.  By using the same set of primitives it does not matter
>>>       
>> how much
>>     
>>> people disagree on the logical representation of any specialized
>>>       
>> term; it is
>>     
>>> the logical representation that specifies the intended meaning, and
>>> different representations of a specific term by different groups are
>>>       
>> not the
>>     
>>> same concept in a merged ontology of two specialized groups: they
>>>       
>> will have
>>     
>>> different names (different namespace prefixes for those terms) and
>>>       
>> different
>>     
>>> logical structure.  The relations between the two - similarities and
>>> differences -  will be immediately evident from a comparison of the
>>>       
>> logical
>>     
>>> structures.  If the logical structures are different, the terms mean
>>> different things.  If they are intended to represent the same
>>>       
>> instances,
>>     
>>> then the two different definitions constitute different theories of
>>>       
>> that
>>     
>>> type of entity, but this does not create a logical contradiction
>>>       
>> within the
>>     
>>> merged ontology, because the different theories are isolated within
>>> different contexts in the merged ontology.  The two systems can still
>>> communicate, but in communication the fact that they have different
>>>       
>> theories
>>     
>>> will have to be taken into account, and the differences will need to
>>>       
>> be
>>     
>>> isolated in order for comparisons to be made.  If this sounds
>>>       
>> complicated,
>>     
>>> it is only what is absolutely necessary for accurate communication:
>>>       
>> the
>>     
>>> automated systems are *forced* to do what people *should* do when
>>> communicating - precisely define terms and account for different
>>> assumptions.  It is also a mechanism for recognizing different
>>>       
>> assumptions,
>>     
>>> and can lead to one or both disagreeing groups to change their
>>>       
>> definitions.
>>     
>>>   Ontologies that are developed separately cannot be merged into a
>>>       
>> single
>>     
>>> consistent ontology without an effort comparable to the effort of
>>>       
>> building a
>>     
>>> new similar ontology from scratch.  This is a consequence of the need
>>>       
>> for
>>     
>>> the merging agent (person or machine) to actually **understand** the
>>> meanings of the terms to be merged.  Machines can't, so it is
>>>       
>> necessarily a
>>     
>>> labor-intensive task for one or more experts.  From observing
>>>       
>> ontology work
>>     
>>> for over a dozen years, it is abundantly clear to me that waiting for
>>>       
>> a
>>     
>>> common ontology to emerge from multiple specialized ontologies is
>>>       
>> hopeless
>>     
>>> unless a properly funded task finds the common basic terms and
>>>       
>> relates the
>>     
>>> different ontologies through those terms.  Funding a common
>>>       
>> foundation
>>     
>>> ontology is unavoidable, and the only question in my mind is how many
>>> trillions of dollars will be wasted before some agency finally takes
>>>       
>> that
>>     
>>> insignificant financial risk.
>>>
>>> Pat
>>>
>>> Patrick Cassidy
>>> MICRA, Inc.
>>> 908-561-3416
>>> cell: 908-565-4053
>>> cassidy@xxxxxxxxx
>>>
>>>
>>>
>>>       
>>>> -----Original Message-----
>>>> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
>>>> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Ron Wheeler
>>>> Sent: Monday, August 18, 2008 7:22 PM
>>>> To: [ontolog-forum]
>>>> Subject: Re: [ontolog-forum] Semantic Web shortcomings [was Re: ANN:
>>>> GoodRelations - The Web Ontology for E-Commerce]
>>>>
>>>> Patrick Cassidy wrote:
>>>>
>>>>         
>>>>> John,
>>>>>   Concerning your point:
>>>>>
>>>>>
>>>>>           
>>>>>> The best designs are developed by small groups.  After they have
>>>>>> proved their value on at least one important application, a
>>>>>>
>>>>>>             
>>>> committee
>>>>
>>>>         
>>>>>> can evaluate them, note missing or inadequate features, and polish
>>>>>>
>>>>>>             
>>>> up
>>>>
>>>>         
>>>>>> the details.
>>>>>>
>>>>>>
>>>>>>
>>>>>>             
>>>>>    This may well be true of foundation ontologies, though a
>>>>>
>>>>>           
>>>> foundation
>>>>
>>>>         
>>>>> ontology is different enough from other artifacts to give one
>>>>>           
>> doubts
>>     
>>>> about
>>>>
>>>>         
>>>>> any analogies.  Even if it is true, it is not inconsistent with
>>>>>
>>>>>           
>>>> development
>>>>
>>>>         
>>>>> by a large group (50-100 participants), since each part of the
>>>>>
>>>>>           
>>>> ontology
>>>>
>>>>         
>>>>> beyond the top level or two is likely to be the focus of a small
>>>>>
>>>>>           
>>>> subgroup,
>>>>
>>>>         
>>>>> and the group as a whole would serve the function of the
>>>>>           
>> "committee"
>>     
>>>> to be
>>>>
>>>>         
>>>>> sure that the work of the subgroups integrates with everything else
>>>>>
>>>>>           
>>>> and can
>>>>
>>>>         
>>>>> handle the applications of interest to the whole group.
>>>>>
>>>>>
>>>>>
>>>>>           
>>>> I can see how this group would collapse into jurisdictional
>>>>         
>> disputes.
>>     
>>>> Who gets to define the medical ontology - drug companies, medical
>>>> equipment companies, HMOs, hospitals, WHO, etc.?
>>>> Software engineering belongs to who?
>>>> What about process control - Equipment suppliers, system
>>>>         
>> integrators,
>>     
>>>> engineers?
>>>> Transportation - carriers, travel agents, shipping companies,
>>>> governments?
>>>> Homeland Security - can you imagine the FBI adopting an ontology set
>>>>         
>> up
>>     
>>>> by Scotland Yard or the KGB or the Chinese Army let alone the CIA or
>>>> Pentagon? If the FBI went along what would be the resistance from
>>>>         
>> state
>>     
>>>> and municipal police?
>>>>
>>>> How is the funding to be divided up? So much per term and
>>>>         
>> relationship
>>     
>>>> defined?
>>>>
>>>> What about the funding agencies/companies - do they get a final say?
>>>> What if the ontology does not meet their needs, will they continue
>>>>         
>> to
>>     
>>>> fund it once they realize that their needs are not being met?
>>>>
>>>> What are the language(computer and human) choices for expressing
>>>> ontologies?
>>>> Who is responsible for translation between computer languages and
>>>> between national languages?
>>>>
>>>> My ontology for process control or homeland security could (and
>>>>         
>> should)
>>     
>>>> be very different from someone else's since we will view objects and
>>>> relationships differently and will need different results.
>>>>
>>>> I think that ontologies will be developed as small packages and
>>>> application designers will have pick the namespaces that they
>>>>         
>> require
>>     
>>>> for their needs.
>>>>
>>>> Sometimes this will result in new merged ontologies being published.
>>>> Some ontologies will be abandoned as better ones appear.
>>>>
>>>> The focus should be on identifying ontologies as they emerge,
>>>> commenting
>>>> on them and providing peer review to improve quality.
>>>> The development of tools to support this process and the use of the
>>>> ontologies as they arise is a much better place for funding to be
>>>> focused.
>>>>
>>>> Ron
>>>>
>>>>
>>>>         
>>>>> Pat
>>>>>
>>>>> Patrick Cassidy
>>>>> MICRA, Inc.
>>>>> 908-561-3416
>>>>> cell: 908-565-4053
>>>>> cassidy@xxxxxxxxx
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>           
>>>>>> -----Original Message-----
>>>>>> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-
>>>>>>             
>> forum-
>>     
>>>>>> bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F. Sowa
>>>>>> Sent: Monday, August 18, 2008 2:20 AM
>>>>>> To: [ontolog-forum]
>>>>>> Subject: Re: [ontolog-forum] Semantic Web shortcomings [was Re:
>>>>>>             
>> ANN:
>>     
>>>>>> GoodRelations - The Web Ontology for E-Commerce]
>>>>>>
>>>>>> Ron,
>>>>>>
>>>>>> Just a comment about standards:
>>>>>>
>>>>>>  > My understanding is that most of the "best" standards have come
>>>>>>  > about through a consensus between the major commercial players
>>>>>>  > with the active (frequently funded) participation of the
>>>>>>             
>> academic
>>     
>>>>>>  > community.
>>>>>>
>>>>>> The important caveat is that committees are terrible at design,
>>>>>> but they're very good at evaluation.  There are many proverbs and
>>>>>> anecdotes about that point:
>>>>>>
>>>>>>   - Too many cooks spoil the broth.
>>>>>>
>>>>>>   - A camel is a horse designed by committee.  (This is a slur on
>>>>>>     camels, which are very well designed for their environment.)
>>>>>>
>>>>>>   - Fred Brooks' _Mythical Man Month_, in which he observes that
>>>>>>     OS/360 would have been far better designed by a group of
>>>>>>     about a dozen designers instead of 150.
>>>>>>
>>>>>> The best designs are developed by small groups.  After they have
>>>>>> proved their value on at least one important application, a
>>>>>>
>>>>>>             
>>>> committee
>>>>
>>>>         
>>>>>> can evaluate them, note missing or inadequate features, and polish
>>>>>>
>>>>>>             
>>>> up
>>>>
>>>>         
>>>>>> the details.
>>>>>>
>>>>>> A prime example is FORTRAN, which was designed by a group of
>>>>>> "academics"
>>>>>> who happened to be employed by IBM (at a time when IBM had a
>>>>>>
>>>>>>             
>>>> sufficient
>>>>
>>>>         
>>>>>> monopoly to throw money at researchers who weren't making a
>>>>>>
>>>>>>             
>>>> measurable
>>>>
>>>>         
>>>>>> contribution to the bottom line).
>>>>>>
>>>>>> There were a few programming languages implemented before FORTRAN,
>>>>>> but they were all very inefficient (at a time when computers were
>>>>>> a few thousand times slower than today's cell phones).  The
>>>>>>             
>> FORTRAN
>>     
>>>>>> group (of about half a dozen people led by John Backus) set out to
>>>>>> design a language and compiler that would produce code that was
>>>>>>
>>>>>>             
>>>> close
>>>>
>>>>         
>>>>>> to the efficiency of code produced by a decent assembly-language
>>>>>> programmer.  And they succeeded.
>>>>>>
>>>>>> After a couple of iterations by IBM, FORTRAN IV became a very
>>>>>>             
>> good,
>>     
>>>>>> very usable, and very efficient language for numeric computation.
>>>>>> The ANSI and later ISO standards bodies took over.  Over fifty
>>>>>>             
>> years
>>     
>>>>>> later, they are still producing new revisions that preserve much
>>>>>> of the original core language.  Today, FORTRAN is still the most
>>>>>> efficient and most widely used language for high-speed numeric
>>>>>> computation.
>>>>>>
>>>>>> For some related thoughts, see the "Law of Standards," which I
>>>>>> formulated in 1991:
>>>>>>
>>>>>>     http://www.jfsowa.com/computer/standard.htm
>>>>>>
>>>>>> And by the way, the original WWW was designed by a small group,
>>>>>> but the Semantic Web was designed by a very large committee.
>>>>>>
>>>>>> John Sowa
>>>>>>
>>>>>>
>>>>>> _________________________________________________________________
>>>>>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>>>>> Subscribe/Config:
>>>>>>             
>> http://ontolog.cim3.net/mailman/listinfo/ontolog-
>>     
>>>>>> forum/
>>>>>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>>>>> Shared Files: http://ontolog.cim3.net/file/
>>>>>> Community Wiki: http://ontolog.cim3.net/wiki/
>>>>>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>>>>>
>>>>>>
>>>>>>
>>>>>>             
>>>>> _________________________________________________________________
>>>>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>>>> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
>>>>>
>>>>>           
>>>> forum/
>>>>
>>>>         
>>>>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>>>> Shared Files: http://ontolog.cim3.net/file/
>>>>> Community Wiki: http://ontolog.cim3.net/wiki/
>>>>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>           
>>>> _________________________________________________________________
>>>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>>> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
>>>> forum/
>>>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>>> Shared Files: http://ontolog.cim3.net/file/
>>>> Community Wiki: http://ontolog.cim3.net/wiki/
>>>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>>>
>>>>
>>>>         
>>> _________________________________________________________________
>>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
>>>       
>> forum/
>>     
>>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>> Shared Files: http://ontolog.cim3.net/file/
>>> Community Wiki: http://ontolog.cim3.net/wiki/
>>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>>
>>>
>>>
>>>       
>>
>> _________________________________________________________________
>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
>> forum/
>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>> Shared Files: http://ontolog.cim3.net/file/
>> Community Wiki: http://ontolog.cim3.net/wiki/
>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>
>>     
>
>  
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/ 
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>  
>
>       (02)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (03)

<Prev in Thread] Current Thread [Next in Thread>