Re: [cuo-wg] FW: Executable English cross domain examples online

["Adrian Walker" <adriandwalker@xxxxxxxxx>]

Hi Jim --

You wrote....

   where would you say EE falls on the Semantic 
Spectrum http://en.wikipedia.org/wiki/Semantic_spectrum?

As you may see, Wikipedia defines 6 levels on its semantic spectrum.  The 6th, "most semantic ", level in Wikipedia is "ontology". 

I'd argue that  that spectrum is inward looking towards 80s and
early 90s mainstream IT, and needs to be extended  because it does
not look outward to the users and uses of IT.   Modern IT can do much more in this direction.

As argued in [1,2,3],  what Wikipedia describes as levels 1 through 6 could usefully be grouped together as Semantics1, (aka data semantics), which captures under 50% of the operational value of the semantic techniques now available.

We can build on
 Semantics1 with Semantics2, 
which 
specifies what consequences should follow from any collection of data
and logical statements [4].  In other wordsSemantics2
specifies what a rule engine or theorem prover should be able to do
with any future collection of logical knowledge.   (This is
different from current practice in commercial rule base systems, in
which programmers figure out on a case-by-case basis whether the rules
are leading to the intended consequences.)   Perhaps with Semantics1 + 
Semantics2 we have 75% of the available semantic value.

However, as mentioned in previous posts, logical statements are
famously difficult to interpret reliably in ordinary English, a big
minus for CDSI and other tasks.   This means that it's
difficult for knowledge authors to get things right, and its difficult
for end users to trust the results enough to act on them.

We can build further on
 Semantics1 + Semantics2, 
by adding Semantics3, which automatically, bidirectionally  and accurately ties the meaning of ordinary (open vocabulary, open syntax) English sentences to the logical statements in
 Semantics2.

Now, if we can make Semantics1, 2, and 3  
work
together properly in one system , we are close to  being able to
exploit  the full operational value of Semantics, for CDSI and
other tasks.

That's what  Executable English and the Internet Business Logic system are designed to do.

One consequence of the design is that the system can explain its
results, in English, at the business level.  It can also explain
in English what the transaction will do, and why it will do it, before
the transaction is committed.

Full natural language understanding by computers is generally agreed to be an unsolved problem, and  the design
of the Internet Business Logic system goes around the problem rather
trying to solve it.  It does this via a lightweight approach to Semantics3.  The approach
has the advantage of accuracy and robustness in the face of  new
words, phrases and jargons -- e.g. government acronyms, military
English.  There is a trade off, and one reason for making
the system available online to anyone with a browser is so that they
can evaluate this. 

I hope this makes sense, and I'll be glad to discuss further .

                                                
Best regards to all on the list,      -- Adrian

[1]  www.semantic-conference.com/program/sessions/S2.html

[2]  www.reengineeringllc.com/A_Wiki_for_Business_Rules_in_Open_Vocabulary_Executable_English.pdf

[3]  Understandability and Semantic Interoperability of Diverse Rules Systems
       www.w3.org/2004/12/rules-ws/paper/19

[4]    Backchain Iteration: Towards a Practical Inference Method that is Simple Enough to be 
        Proved Terminating, Sound and Complete. Journal of Automated Reasoning, 11:1-22

Internet Business Logic (R)
Executable open vocabulary English
Online at www.reengineeringllc.com
                               
Shared use is free

Adrian Walker
Reengineering
Phone: USA 860 830 2085

On 11/22/06, Schoening, James R C-E LCMC CIO/G6 <James.Schoening@xxxxxxxxxxx> wrote:

Adrian,

    This looks like the same message you posted on the 21st at 2:47 PM.

 

Also, where would you say EE falls on the Semantic Spectrim http://en.wikipedia.org/wiki/Semantic_spectrum?

 

You may respond to the list or to me if you like.

 

Jim

 

---

From: cuo-wg-bounces@xxxxxxxxxxxxxx [mailto:cuo-wg-bounces@xxxxxxxxxxxxxx] On Behalf Of Adrian Walker
Sent: Tuesday, November 21, 2006 7:52 PM
To: common upper ontology working group
Subject: [cuo-wg] Executable English cross domain examples online

Hi Jim --

Thanks for your note.

You wrote...

    EE [Executable English] is a way to express semantics, as is OWL, Common Logic, etc.   But if two groups use the same language but independently develop an ontology, vocabulary, etc., then a computer from one domain will not be be able to process/reason/inference with data from the other domain.   In other words, a standard language does not solve CDSI.   If you disagree, let's keep discussing this point.  Yes, people could understand it, but CDSI requires computers to process inputs.

Actually I thought this was addressed with the two examples [1,2] in a previous email:

[1]  shows how to write in EE the knowledge that people have about how to bridge between metric data (kilometers per second) in stovepipe A  and  non-metric data (miles per hour)  in  stovepipe B.  It does this by choosing a canonical pivot representation (metric), so that if there are N stovepipes A,B,C...  at most 2N conversions written in EE will be needed to make them all interoperate.

[2]  shows how to write in EE the knowledge that people have about how to bridge between the terminology used by a manufacturer A (Prof. Desktop), and the terminology used for similar items (PC for gamers) by a retailer B.  It does this by choosing a canonical pivot representation in the form of a simple EE ontology whose upper levels (Computers) are shared .   Thus if  there are N stovepipes A,B,C...  at most 2N conversions written in EE will be needed to make them all interoperate.

So, in each of the cases [1,2] the scenario is like this.  Stovepipe Z wishes to compute with data from Stovepipe B.   This will be done via canonical representation C (it's optional whether or not C is the actual representation used by any stovepipe.)  Z tells EE what it wants.  EE maps that  into the canonical representation C, and then maps it further into B's representation.  B produces what's needed, but it's still in B's representation.  EE maps it into representation C, and then into representation Z.

I hope that  folks may have time to run the examples [1,2] to see how they work.  There's nothing to install.  Just point a browser to www.reengineeringllc.com .  Preferably Mozilla or Firefox, but see also the "Browsers" page for the settings needed for IE.

  So, since a standard language, regardless of how expressive it is, cannot solve CDSI, EE should not be listed as a Candidate Technical Solution for CDSI.  Is this acceptable to you?

Well, I guess not.   Based on the discussion so far,  it seems that EE at least supplies a challenge to see if other candidates can do better.  One advantage for this purpose is that EE is an implemented system that  anyone with a browser can run to see if  it meets emerging CDSI test requirements.  In other words, "EE is useful for CDSI" is a falsifiable hypothesis (which is how science proceeds), and as such EE should not be discarded at this stage.

You did send me an email suggesting an upper model for EE.  First, I'm quite sure this isn't feasible unless you're really talking about an upper ontology expressed in the form of EE.

Yes, I believe [2] is an example that most people would view as containing part of a simple upper ontology.

Second, to get people to consider this approach, you would need to present your case  in a stronger form, such as a well-thought-out paper, or demo, or group of supporters, etc.

Glad to.  The papers [3,4] should serve as a good starting point.   As mentioned, anyone can run the examples [1,2] using the "demo" ID at www.reengineeringllc.com .    There are many additional examples, including ones based on EE-augmented OWL, RDF, and biomedical ontologies.   Folks are cordially invited to use their browsers to write and run their own examples.

If you have some examples in mind, we can see whether or not EE can cover them -- a few paras of English and some sample data should do [7].  This can be done either on the www public "demo" ID, or on an ID with restricted password access.

I hope this takes your observations properly into account.   Thanks for making me think!

If you'd like to proceed via phone, or have me do a  presentation (perhaps based on [5,6]), that would be great.

                                         Best regards,  -- Adrian


[1] www.reengineeringllc.com/demo_agents/OntologyInterop2.agent

[2] www.reengineeringllc.com/demo_agents/SemanticResolution1.agent
 
[3]  www.tdan.com/i027hy04.htm

[4]  www.reengineeringllc.com/A_Wiki_for_Business_Rules_in_Open_Vocabulary_Executable_English.pdf

[5]  www.reengineeringllc.com/Internet_Business_Logic_e-Government_Presentation.pdf

[6]  www.reengineeringllc.com/Business_Rules_and_OMG_SBVR_Presentation.pdf

[7]  For an example of a task outline, please see the last two paras on page 1 of
      www.reengineeringllc.com/Oil_Industry_Supply_Chain_by_Kowalski_and_Walker.pdf

Internet Business Logic (R)
Executable open vocabulary English
Online at www.reengineeringllc.com
                                 Shared use is free

Adrian Walker
Reengineering
Phone: USA 860 830 2085

 _________________________________________________________________
Message Archives: http://colab.cim3.net/forum/cuo-wg/
Subscribe/Unsubscribe/Config: http://colab.cim3.net/mailman/listinfo/cuo-wg/
To Post: mailto:cuo-wg@xxxxxxxxxxxxxx
Community Portal: http://colab.cim3.net/
Shared Files: http://colab.cim3.net/file/work/SICoP/cuo-wg/
Community Wiki: 
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/CommonUpperOntologyWG    (01)