Dear David, (01)
Let me see if can help clarify some affairs by sharing some work I did
with my colleagues fropm Cutter Consortium: Ed Yourdon and Andre
Leclerc. I am cc'ing them only for the sake of reference and
completeness (not that they need to get onto more emails on a list in
any following responses, so please delete them unless they choose to
respond). (02)
You wrote: (03)
"Said another way... if you can distill ANY semantics from MSTR-MENSA-FL with
RDF, predicate calculus, KIF, or any other formal logic process, I'm all ears.
.... How will ontologies help with legacy systems maintenance?" (04)
We have re-engineered 4.5 million lines of COBOL, several megabytes of
text files etc.. etc... at a very, very large corporation whose systems
date back to the 1960's and did this with Ed and Andre in about 8 weeks
(all complete using methods that worked *because* we started from the
ontological perspective). In our example, Ontologies fundamentally
described a classification schema of " invariants" and measures of
structural properties. One invariant in software description can be the
notion of a "loop" or an "iteration". Now, in order to detect loops,
you need to parse Cobol but, and here is the magic we saw in
Ontologies: you then "ontologize" the "loop concept" by associating it
to measure (example, you apply the Cyclomatic Number) with applicability
rules (ie, your write a little "theory") . You can parse text and do
the same. You can purchase the paper at www.cutter.com and search for
"LegacyWorks". I am listed there as a Senior Consultant. My other
colleague Ken Orr has been using some very interesting techniques and
modern tools to reverse engineer legacy databases or to cross map the
logical views. Unfortunately, the reality is that most organizations
are not ready to have an legacy re-engineering done simply because "if
it aint broke don't fix it". (05)
On another level, with respect to the last part of the question,
Ontologies in Legacy Systems will help in the optimization and
change-reduction cycle because they help to make explicit the Business
Process (Architecture) that is usually implicit, hidden and that has
evolved over decades. Ontologies also serve to elaborate and extend
well know notions such as data dictionaries and by helping to push them
upwards from instance-value descriptions to classifications or indeed,
even to behavioral objects. An example of a behavioral object, is
"Currency". In a legacy system I may have a currency variable assigned
to a value that takes a type of currency. It may be more useful to have
a currency object that knows its own conversion rules: hence, an
ontology (which in this case is a representation of a theory of currency
and currency exhcnages) can be used. (06)
I could provide a lot more on this topic, but, as I mentioned there is a
paper available. We used an Analogical Reasoning Engine that used
ontologies and a parser, that used ontologies in order to relate Cobol
and Text files through invariant Measure (we have an ontology of 80 such
Measures). (07)
On the last note, the use of Logic greatly helps these implementation
efforts because there are ways, methods and languages that to various
extents support logic programming (for example, we ended up choosing
PROLOG to write the parsers, the Ontologies and C-Language external
modules, as per the Ontology, to do the number crunching). (08)
I hope that this provides some *hope* to you that all is not dark and
all is not lost with ontologies or the proverbial Meta-Data. (09)
Wishing you a wonderful 2006, (010)
Arun Majumdar (011)
David Eddy wrote: (012)
>John Sowa -
>
>BEFORE reading the 5 additional waiting digests in my in-box, I'm going
>to come out of lurk mode & throw my two cents on the table--at the
>extremem opposite end of the table from John Sowa I think--as to what I
>think could be a valuable contribution from this group...
>
>
>
>
>>CC> A valuable task for this group would be to collect and
>> > validate user driven requirements as well as the scenario
>> > of applying an upper ontology to those solutions. This
>> > will help nail down the set of problems we are addressing
>>
>>
>>
>
>Maybe the phrase "rip & replace is NOT an option..." is old news, but
>the first time I heard it at Mitre gig in McLean in September 2004
><http://www.topquadrant.com/conferences/sept8_2004/stgov04.htm> from
>Mike Daconta, it really grabbed my attention.
>
>For a variety of reasons I'll claim that I look at the world thru the
>very foggy & scratched glasses--very much the opposite of "rose
>colored"--of a maintenance programmer. Pick your legacy language...
>COBOL, Fortran, PL/S, Algol, Java, Ruby, PHP,...
>
>I've had a wonderful time in my career BUILDING systems... guess what?
>The green fields are all built up now. While it may be feasible to put
>$13 billion into the rats hole of Boston's "Big Dig" (thank you very
>much taxpayers of America), such massive reconstruction efforts are NOT
>likely to happen to the software infrastructure we've built over the
>past 50+ years.
>
>The challenge going forward is getting those crusty old legacy
>systems--that WORK--to talk with other systems.
>
>And that's where taxonomy, ontology & semantics comes in... not 1 in
>100,000 of the systems that our daily lives depend on has been either
>conceived, designed, built, and certainly not maintained with a
>nanosecond of thought towards the organizational principles in formal
>ontologies.
>
>Said another way... if you can distill ANY semantics from MSTR-MENSA-FL
>with RDF, predicate calculus, KIF, or any other formal logic process,
>I'm all ears.
>
>
>I recently discovered this interesting article by Dr George (Mr WordNet)
>Miller...
><http://www.kurzweilai.net/meme/frame.html?main=/articles/art0186.html>
>
>... take a simple 13 word sentence & it provides some 3.6 Trillion
>combinations. As he points out... no wonder "machine translation" has
>had such a hard time living up to it's dreams.
>
>And as far as I can tell the WordNet effort primarily only deals in real
>words.
>
>News flash... inside software applications "real words" are few & far
>between.
>
>
>What I want out of this ontological discourse is something that helps
>systems analysts & programmers to more quickly UNDERSTAND what the
>systems under their care are actually talking about.
>
>
>Bluntly... I've heard this fundamental challenge/conundrum throw up in
>the SOA (service oriented architecture) commercial space...
>
>The business problem: I'm in an insurance company that has several
>thousand applications (a universe of undefined extent comprised of
>custom built applications, heavily customized packages & box-stock
>packages...)
>
>
>
>[FYI... there was a recent blurb in WashingtonTechnology...
>
>"EA helps mind the money
>By Drew Robb
>11/07/05; Vol. 20, No. 22 Intelligence is an imperfect science. Just ask
>CIA or the 9/11 Commission. Or EDS Corp., the contractor tasked with
>wrangling thousands of legacy systems into the Navy-Marine Corps
>Intranet. When EDS started the job, the Navy thought it had about 5,000
>applications to integrate. EDS found more than 100,000."]
>
>
>
>If I'm in Silo A & know what "policy number" represents, how do I know
>that over in Silo B, I need to be looking for "contract ID"?
>
>[This is an extremely simplified example... this basic problem caught my
>attention 25 years ago at an insurance company that had discovered some
>70 names for the core "policy number" concept... I doubt if that number
>has gotten smaller in the past 25 years.]
>
>
>Sorry to be so long winded... but I vote with Denise Bedford...
>"metadata is NOT hierarchical." And the corollary... software systems
>are NOT inherently organized.
>
>
>How will ontologies help with legacy systems maintenance?
>
>_____________________
>
>- David Eddy
>Babson Park, MA
>781-455-0949
>
>_________________________________________________________________
>Message Archives: http://colab.cim3.net/forum/ontac-forum/
>To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
>Subscribe/Unsubscribe/Config:
>http://colab.cim3.net/mailman/listinfo/ontac-forum/
>Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
>Community Wiki:
>http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG
>
>
>
> (013)
_________________________________________________________________
Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Subscribe/Unsubscribe/Config:
http://colab.cim3.net/mailman/listinfo/ontac-forum/
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG (014)
|