ontology-summit
[Top] [All Lists]

Re: [ontology-summit] Ontology Engineering for Systems Engineering

To: "'Ontology Summit 2012 discussion'" <ontology-summit@xxxxxxxxxxxxxxxx>
From: henson graves <henson.graves@xxxxxxxxxxx>
Date: Tue, 21 Feb 2012 23:08:59 -0600
Message-id: <SNT106-DS217E1D2AD246A62D7CAAFCE4640@xxxxxxx>
Dear John,    (01)

Thanks for the email; I appreciate your comments.     (02)

You asked a question which I will interpret as why I believe one can
incorporate formal methods into everyday engineering and scientific use, and
what kind of engineering usage can one build on and extend to achieve this
outcome. I can answer this fairly quickly and will include a couple of
papers on the topic. The real evidence for the assertion takes more time and
involves personal experience.     (03)

With engineering languages such as UML and SysML (a UML dialect for
engineering) engineers can and do develop the designs for very complex
systems. These languages work in practice. There is a lot of evidence for
this. I have a lot of experience with using the languages to build and
analyze complex systems. I enclose one paper on this topic; it points to
other more technical papers.    (04)

The fact that a language such as SysML works is a relatively new
development. Their predecessors did not work until after year 2000. There
are very specific "ontological" reasons why the predecessors did not work
and the more recent ones do work.  Without going into detail it has to do
with system components, binary relations, etc. The very topics that I had
hoped to engage the ontologs with. I have engaged Matthew, but he has
engineering experience. Engineers, not the AI or ontology folks, are the
ones who have discovered good terminological primitives for building system
representations. This is actually what is underlying the dialog that Matthew
and I are engaged in. I am testing how engineering languages can represent
systems such as the distillation unit against Matthew's experience, as well
as my experience.      (05)

To build on and extend engineering usage the first thing to realize is that
a language such as SysML can be retrofitted with a formal semantics based on
type theory. A SysML model translates to an axiom set within type theory;
many engineering questions translate directly into axiom set consistency
statements. What retrofitting SysML into type theory gets you is a solution
to the primary issue that has impeded the success of formal methods.
Engineers and other mere mortals cannot build first order let alone higher
order logic axiom sets. For that matter they cannot build OWL axiom sets
very well either and that is a lot simpler. However, engineers can build
complex SysML models. The graphical syntax of a language like SysML enables
engineers to build very models which translate into very complex axiom sets
with all sorts of quantification. This solves of the problem that people
cannot build axiom sets. As the engineering problems translate into logic
problems automated reasoning methods can be applied and the answer can be
presented back to the engineer. I have use cases.     (06)

So to summarize I am pretty confident that the integration of formal methods
into everyday engineering practice will happen in due course.  Further the
reasons for success and failure reflect a lot of the 20th century progress
in mathematical foundations from Cantor, Russell, down to Lawvere and
Lambek. The reasons for success also reflect the practice of engineers in
the trenches as they have sorted out the concepts they need to represent
their systems.     (07)

Based on my experience there is little intelligent design in the
evolutionary process of developing ontology useful for engineers and
scientists. The process is slow with dead ends and little or no
understanding of what went before.      (08)

- Henson    (09)

-----Original Message-----
From: ontology-summit-bounces@xxxxxxxxxxxxxxxx
[mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F. Sowa
Sent: Tuesday, February 21, 2012 7:38 PM
To: ontology-summit@xxxxxxxxxxxxxxxx
Subject: Re: [ontology-summit] Ontology Engineering for Systems Engineering    (010)

Dear Henson,    (011)

I agree with your points:    (012)

> While (1) is true one needs (formal) specifications to judge 
> deviations from. One of my current research projects is extending the 
> kind of stuff you see there to specifications for product lines as 
> well as approximate satisfaction of an individual system  against a 
> specification.  Regarding
> (2) I believe that there is sufficiently developed engineering usage 
> in what works and that it can be used to develop precise semantics for 
> engineering languages and formal ontologies that engineers can 
> actually use. Further this can be retrofitted with the languages that 
> they use without any necessity of their changing how they work. 
> Engineering usage background is well beyond the stone age as I am sure
that you would agree.    (013)

But I'd also like to emphasize that these issues have been analyzed and
debated in AI circles since the 1960s and in the DB field since the 1970s.
In 1980, there was an ACM workshop that brought together people from three
fields:  AI, databases, and programming languages.
Following is the list of speakers and their topics:    (014)

    http://www.informatik.uni-trier.de/~ley/db/conf/sigmod/pingree80.html
    ACM Workshop on Data Abstraction, Databases and Conceptual Modelling    (015)

Pat Hayes and I were two of the participants.  Ted Codd was one the DB
people.  It was a great workshop with a lot of inspiring insights and
discussions.  I still have a copy of the proceedings in my basement, but
it's no longer inspiring.  It's disgusting to realize that we're still
talking about the same topics 32 years later without seeing much progress in
implementing and using those ideas in mainstream IT.    (016)

I like your point that this work "can be retrofitted with the languages that
[engineers] use without any necessity of their changing how they work."    (017)

The reason why the Semantic Web failed is that the designers ignored the
issues about how their tools and languages would be used.  Instead of
integrating their tools with mainstream IT practices, they developed
"proactive" standards that had no connection to what anybody was doing.    (018)

Question:  How do you propose to address those issues?  What kind of
engineering usage are you planning to build on and extend?    (019)

John    (020)

_________________________________________________________________
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/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012
Community Portal: http://ontolog.cim3.net/wiki/     (021)

Attachment: Integration 03-nov-11.pdf
Description: Adobe PDF document


_________________________________________________________________
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/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012  
Community Portal: http://ontolog.cim3.net/wiki/     (01)
<Prev in Thread] Current Thread [Next in Thread>