Re: [model-challenge] Executable English Reasoning

[Adrian Walker <adriandwalker@xxxxxxxxx>]

Hi Leo,

Thanks for your note. 

The bottom line for us is -- does the Executable English system [1] do something useful in the real world?

Now for some detail on how it goes about attempting that.

You wrote:

Also, correct me if I’m wrong, Adrian, but behind this restricted English representation is logic programming, i.e., Horn Logic, a syntactically restricted FOL.   [emphasis added]

 

Actually, restricted is probably misleading.  Perhaps surprisingly, the vocabulary is open, and so to a large extent is the syntax.  This means for example that one can freely write using government acronyms, jargon, etc.  [2] .   That's a big claim, I know, and there is a trade off.  The system can also be used to reason about controlled vocabularies.

"Logic programming" means different things to different people.

I'd label Prolog as "Logic Flavored Procedural Programming", and Executable English as "Logic Oriented Specification Seeking All Answers". 

The core inference method behind EE is different from Prolog's, and is based on [3], meaning that it extends pure Horn with a version of negation-as-failure.  (NAF is controversial in some communities, but it is essentially what SQL uses, and the world economy runs on SQL.  More recent versions of SPARQL support something similar to NAF.)

Do you use any non-Horn Logic extensions or second-order logic, etc.?

Yes, a version of NAF, as mentioned.  There are  aggregations (sum, min, max etc), and explanations, namely English versions of proofs/plans.  There are also built-in predicates for arithmetic, for checking the current date, and so on.

I hope this makes sense.  It is admittedly contrarian, but it is also live online for anyone to test. There are papers and presentations on the web site, and  we can go into deeper technical details off-list or by phone if you'd like.

                            Thanks,  -- Adrian

[1]  Internet Business Logic
A Wiki and SOA Endpoint for Executable Open Vocabulary English Q/A over SQL and RDF
Online at www.reengineeringllc.com   
Shared use is free, and there are no advertisements

[2]   www.reengineeringllc.com/rule_examples.html

[3]   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

On Sat, May 5, 2012 at 5:12 PM, Obrst, Leo J. <lobrst@xxxxxxxxx> wrote:

Also, correct me if I’m wrong, Adrian, but behind this restricted English representation is logic programming, i.e., Horn Logic, a syntactically restricted FOL.

 

Is that right, Adrian? Do you use any non-Horn Logic extensions or second-order logic, etc.?

 

Thanks,

Leo

 

From: model-challenge-bounces@xxxxxxxxxxxxxxxx [mailto:model-challenge-bounces@xxxxxxxxxxxxxxxx] On Behalf Of henson graves
Sent: Saturday, May 05, 2012 4:41 PM
To: 'Modeling Benchmark Challenge'
Subject: [model-challenge] example of reprsenting marriage in FOL

 

Adrian,

I did miss your link. If anyone else did, here it is:

   www.reengineeringllc.com/demo_agents/Marry3.agent

This is the kind of example representation that can get the discussion moving forward. Correct me if I am wrong in the following. It looks like you are using a binary or 3-place predicate Marriage(x,y,t). Presumably you have unary predicates for male, female and time. That would allow you to make assertions about Mary and John being married at a specific time. Presumably you have axioms for time. Do you have time intervals. Do you have axioms that preclude someone being married to two people at the same time. With this axiomatization there does not seem to be any concept of marriage other than the pairs of individuals which satisfy the relation. If you needed a way of identifying a specific marriage you could of course add another variable to the marriage predicate to serve as a key.

 

Any other suggestions from anyone as to how to give axioms for marriage? How does one bridge between this and Cory’s pictures?

 

Marriage was chosen as a challenge as it is probably the simplest example of a class of modeling problems which includes the detailed design description of a Boeing 747.  While natural language is important one wonders if it scales to represent all of the components and connections in a large manufacture product.

 

Henson

_________________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/model-challenge/
Subscribe: mailto:model-challenge-join@xxxxxxxxxxxxxxxx
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/model-challenge/
Shared Files: http://ontolog.cim3.net/file/work/ModelingBenchmarkChallenge/
References: http://ontolog.cim3.net/file/resource/project/ModelingBenchmarkChallenge/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?ModelingBenchmarkChallenge

_________________________________________________________________________ 
Message Archives: http://ontolog.cim3.net/forum/model-challenge/ 
Subscribe: mailto:model-challenge-join@xxxxxxxxxxxxxxxx 
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/model-challenge/  
Shared Files: http://ontolog.cim3.net/file/work/ModelingBenchmarkChallenge/ 
References: 
http://ontolog.cim3.net/file/resource/project/ModelingBenchmarkChallenge/ 
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?ModelingBenchmarkChallenge    (01)