Re: [ontology-summit] [Making the Case] Elevator Pitch

["Brian K Lucas" <lucasb@xxxxxxxx>]

 What an interesting thread of conversation so far!  I've been following, but too busy to compose my thoughts until now.  Allow me to add a few observations from my own perspective:

1) Some of the conversation appears to refer to something more like Master Data Management and singular, integrated data models.  As a newcomer to ontologies, that is not my impression of what ontologies offer - nor the reasoners that operate on formal ontologies, like OWL-DL types.  While you can use an ontology to produce a model, I do not consider that their "highest and best use".  I do not believe that it is possible (or desirable) to pursue a single model anyway (oil well examples notwithstanding).  But that is a much larger philosophical conversation.

2) To be different (and therefore address an unmet need), ontologies must offer solutions to things that other methods do not.  Since I can only speak for myself, I'll mention what attracted me to Semantic-web-based ontologies - the ability to describe awareness of the world from a particular perspective, in whatever local terms I use, and then, employ a reasoner to tell me whether (a) two things I've described are really the same (based on the ontology structure) and (b) what other awareness is discoverable once I express a concept in a given ontology.  To achieve (b), I need all relevant ontologies to cross-reference equivalent concepts to extend my awareness beyond what I've collected in my own perspective, and access to the other expressions.  Some better tools for capturing awareness and cross-referencing ontologies would help here - but that's just a tooling issue.  There are many other classification methods that I know that can solve the terminology and classification problems, by forcing their definitions into use (e.g., any data modeling tool).  Ontologies are the first one I've encountered that accept the data for what it is, and apply rules to reconcile and extend it.

3) In my limited use of ontology languages so far, it has made me call into question the anti-polymorphism imbedded in many data models and ontologies.  We frequently begin by declaring some concepts as metaclasses, then require the user to declare instances of these classes before we can operate on them.  By definition, this forces the user to pre-classify the thing that they are trying to describe, removing the ability of the ontology to discover what it really is (including the possibility of being multiple things).  The data that is expressed by instantiating the ontology is already "muddied" by the perspective of the collector.  To refer back to a previous example in the thread, the problem may not be that finance and engineering refer to oil wells and mean different things - it may be that the ontology of each perspective forced then to classify things as oil wells, which were not in fact the same concept.  Two possible solutions are (a) extensively cross-referencing local ontologies or (b) building ontologies that do not force me to decide which class an object belongs to before instantiating an object in the first place.  On a side note, I believe that this imbedded irreconcilability is the major problem with the programming paradigms of today.

Having said that, for what it's worth, allow me to offer a rough draft of an elevator speech, aimed at the audience of "anyone who wants to understand more".  I'm sure that we could edit this to be more direct and "pithy", but I wanted to get the full thought out first.

The long version:
Every human has a different perspective on the world.  When they express those perspectives, they use models of the world that they carry in their heads, and language derived from their individual life experiences.  If you are trying to tame the complexity caused by these perspectives, you have two choices: (1) try to alter the way people express themselves to align with an external model - which is virtually impossible - or (2) allow people to express themselves in local terms, then reconcile those terms with all others, to discover similarities and differences.  This heightened awareness will help you make better decisions.  This works for conversations between people as well as between computer systems.  Ontologies, and the software behind them, help you to do this.

The short version: 
Ontologies can help you tame increasing complexity.

Brian K. Lucas
Sponsor, Worldwide Institute for Organization Ontologics

-----Original Message-----
From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Peter Yim
Sent: Wednesday, February 02, 2011 12:28 PM
To: Ontology Summit 2011 discussion
Subject: Re: [ontology-summit] [Making the Case] Elevator Pitch

Yuri, John, Anders and All,

I think we have sufficiently exposed the key aspects of this issue from different vantage points, and each of your positions are clear enough.

In the solicitation for "sound bites" and "elevator pitches," we are hoping to get them from a variety of perspective anyway. There is no desire (at least for now) for consensus on *one* sound-bite or *one* all-around elevator-pitch that everyone can use in the future.

Therefore, feel free to keep your positions, and let's move on, to tackle other related issues.

So, everyone ... send in your input, either at the

wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011_MakingTheCase_CommunityInput

or, via the
online form: http://ontolog.cim3.net/file/work/OntologySummit2011/Survey/Survey01_Soundbite-Pitch/survey01.html

Thanks & regards.  =ppy
--

On Wed, Feb 2, 2011 at 8:51 AM, John F. Sowa <sowa@xxxxxxxxxxx> wrote:
> On 2/2/2011 11:10 AM, Yuriy Milov wrote:
>> We don't understand the nature of the mouse brain but a chess
>> computer wins grandmasters.
>
> Yes, but there are only two kinds of things that computers do better
> than humans (or even birds and mammals):  storing large volumes of
> data exactly and computational things like arithmetic and chess.
>
> For go, the computational methods can beat amateurs, but not the
> lowest level of go masters.  That's because humans use perceptual
> mechanisms that no computer today can come close to matching.
>
> Robots can do extremely precise detail, but they can't do things that
> are trivial for humans and other animals:  e.g., wash dishes the way
> people do or build a bird's nest the way birds do.
>
>> Court trials is just a textual games by clear rules where a judge is
>> similar to a game server, and if legal disputes go online it would be
>> a great Elevator Pitch.
>
> Please go do your homework on the state of the art of AI today.
>
> An elevator pitch for something that nobody knows how to do would be
> worthless.
>
> John
 
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2011/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011
Community Portal: http://ontolog.cim3.net/wiki/

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2011/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011  
Community Portal: http://ontolog.cim3.net/wiki/     (01)