One of the specializations of my companies is the development of Knowledge
Networks (tm) and Knowledge Servers(tm) for various clients.
The Knowledge Servers have, as their core, an "Agent Based" artificial
intelligence tool which permits users to codify knowledge into Agents.
More imortantly, because the Agents are internally-equipped by the
software to interface *autonomously* with SQL databases, CAD drawings,
flat ASCII files, plant machinery I/O points, etc. the Agents *themselves*
are able to learn.
Agents may be launched across the intranet to find information or
relationships previously unknown to the user. Working with discrete event
simulators the Agents have discovered knowledge previously-unknown to
*anyone*. When this occurs the Agent passes the new knowledge to the
Knowledge Server in the form of an expert system rule or set of fuzzy
logic statements or as a neural net. In any case the Knowledge Server
converts the above formats to English- language logic statements and
submits them to the appropriate person (human) for
approval/disaproval/modification and ultimate incorporation into the
enterprise knowledge base.
Interestingly, the machine learns from people and other people learn from
the machine - a human-machine "partnership" results which is worth much
study.
Hal Popplewell
IntelliSys Group
GaltJohn22@aol.com
--Learning-org -- An Internet Dialog on Learning Organizations For info: <rkarash@karash.com> -or- <http://world.std.com/~lo/>