[Edu-sig] intell adoption problem

[portal]

Kirby,

I know Dr. Bruce Lund at In-Q-Tel.  I have worked for various agencies,
Offices of Declassification, in the past.  Bruce got his PhD at Georgetown
in 1992 in computational linguists while I was Director of the Neural
Network Research Facility at GU.  Methods we call knowledge technologies
proposed by SAIC/OntologyStream was deemed fundable by NIMA's  Novelty
Detection in Massive Databases program last year, and yet no funding
occurred.  We are in the final round of the DARPA REAL competition.

The adoption problem is that the types of technology we have is transparent,
simple, powerful and fast; and does not need J-Boss or .NET infrastructure
to work.  We do not need Cobra, or net services.  We have the perfect Python
set up, in that a number of powerful engines need to be binary executables
and to exist in an environment where a command line can bring these into the
execution space.  So very simple interfaces to the engines exist in all
cases.  Everything is ASCII or very simple XML.

The knowledge technology is simply not understood by those who can block
funding decisions.  Many do understand the technology and advocate for
funding, but there are a few who somehow feel that our technology commits a
moral sin.  This is what we have focused on the National Project:

http://www.bcngroup.org/area2/KSF/nationalProject.htm

The increased power to latent semantic indexing engines by my invention of
an ontology lens

http://www.bcngroup.org/procurementModel/to-be/lens.htm

is completely ignored because of the absence of a business model.

This is a problem that will last until my group produces a success in the
business world.

The ontology lens is only one of many un-adopted deep technologies that has
been developed by a world wide community (Core Talk from California,
poly-logics from Germany, quasi-axiomatic theory from Russian, natural
language generation from ontology work in Australia, etc.)

But the business proposition is not there at this time, and thus In-Q-Tel
can do nothing by charter.

I also know Brian Moon (Friends of the IC) and may attend his upcoming NIST
workshop.

The Orb work uses Python, Perl and the Berkeley DB (the hash table
management part of this "DB").

http://www.inorb.com/services.html


However my number theoretic I-RIB (In-memory Referential Information Base)
infringes on a powerful patent (www.primentia.com) and I use a version of
this invention (Bjorn Gruenwald's) to create what is actually a key-less
hash with no need for empty hash containers and no possibility for
collisions.  Bjorn is not happy about this, but I am not using it
commercially, only to teach about scatter-gather methods.

(The key-less has table can be demonstrated... it exists and does make
sense.  The hash function simple takes an ASCII string and "regards" the
string as a base 64 number (this is the invention that PTO awarded to Bjorn
in 2003).  Then some magic happens.

The current use of the Berkeley DB rather than the I-RIBs is in preparation
for a commercial venture.