Dear Chris, (01)
Thanks for your interest. The code is in Delphi
7, and is presently organized to process patent
texts. The regular old standard And/Or algorithm
I chose was Nils Nilsson's in his first book
(about 1972) on Artificial Intelligence. I wrote
the And/Or graph, search and explanation
components in Pascal a long time ago, and have
ported them into D7 into object oriented form. (02)
But the code is embedded into my product, and is
not available to public users at this time because
I have not fully completed the reference design,
which will become a product when I do, or which
may be a license reference design if the business
climate is more conducive to licensing at that
time. The patent specification, along with
standard textbooks (I still recommend Nilsson 1972
as the most readable) which are taught in computer
science departments everywhere, and so the
hypothetical Person of Ordinary Skill in the Art
(Posita in patent terminology) can construct
everything needed from the patent specification. (03)
There are lots of intersections with my other code
classes for the lexical, syntactic and pragmatic
parts of the program, which is why I don't make
the And/Or graph and search algorithms available
by themselves. . There are many such publicly
available implementations of the algorithms to be
found in various languages with a Google search
though, if you are interested. (04)
The database interfaces in D7 are the ones I
prefer - straightforward ADO standards for
emitting SQL from the client programs to the
server of your choice, touching off stored
procedures, and retrieving selected columns and
tables from the massive databases that can benefit
most - for example, medical records databases,
observations from experiments, and business
transaction databases. (05)
I also have a proprietary memory resident
relational database, also implemented in D7, which
handles the metadata database shown in the first
few figures in the patent. Performance is better
that way; it can interpret the storage level
representations in an organized way like that, I
believe, than doing EVERYTHING in SQL, That
approach also provides a higher degree of database
independence because so many DBMS products have
highly peculiar metadata tables, and therefore it
doesn't require a lot of rehosting to move to (for
example) Oracle, SQL Server, Sybase, Firebird, or
other SQL dialects so long as they are not too far
deviant from the SQL '92 standard for the less
exotic SQL statements. The variety in metadata
SQL implementations can become very confusing in
the code, mixing SQL with Pascal. So I recommend
you consider the code level interpretation of the
metadata database as I did if you decide to
implement something similar. But that isn't
necessary; its simply a useful technique if you
want portability of DBMS products, which may or
may not even be of interest, depending on the
application. (06)
Another implementation pragma is that there are
many more elegant structures that can be built
around the metadata database - symbol tables,
directed graphs, the And/Or search algorithms,
inverted index manipulations, lexical analyzers,
recursive descent parsers, etc, that can be
constructed around a metadata database than using
SQL directly, but it can be done with straight SQL
if you prefer. That direction is simpler, just
less flexible. (07)
HTH,
-Rich (08)
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2 (09)
-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On
Behalf Of Christopher Menzel
Sent: Tuesday, August 16, 2011 9:51 AM
To: [ontolog-forum]
Subject: Re: [ontolog-forum] Self Interest
Ontology (010)
On Aug 16, 2011, at 11:35 AM, Rich Cooper wrote:
> Dear John,
>
> Re your ISS slides, slide number 65, you wrote:
>
> > Any formal theory can only express one version
at a time.
>
> That is why I chose the And/Or search algorithm
class for my database context discovery methods.
Using different heuristic metrics, multiple
viewpoints of the same factual structures
(represented in the graph) can be visited
sequentially as the solution subtree switches
among the alternative interpretations. Each such
solution subtree is itself consistent, and the
forest of subtrees is based on the same facts and
rules which establish any one of the subtrees. (011)
Do these "database context discovery methods"
actually exist in code, in a form people can
actually use for doing real-world knowledge
engineering, or are you just sketching an
architecture that you think can be coded and that
you think would work if it were? If the former,
where is the code and is it accessible? (012)
Chris Menzel (013)
__________________________________________________
_______________
Message Archives:
http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr:
http://ontolog.cim3.net/mailman/listinfo/ontolog-f
orum/
Unsubscribe:
mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join:
http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePa
ge#nid1J (014)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (015)
|