Guess I created confusion everywhere:
Concerning IDEF and "Intelligent Agents" - John Mortimer writes:
> Sounds fascinating, I just wish I understood some of it.
> What language was it in?
and later wrote:
> I would love to find out, but I REALLY have no idea what you
> are talking about. Sounds fascinating, I just wish I understood
> some of it. What language was it in?
To which I reply:
----------------------------- NEW TEXT STARTS -------
I must be a HORRIBLE writer.
I'm speaking of a software tool which will accept and run knowledge,
codified in the form of rules, neural networks, fuzzy logic, and the like.
I define the word knowledge as "symbolic logic" in this instance.
The knowledge goes into specialized modules called "intelligent agents".
Each agent is an executable program which can run on virtually ANY
processor (eg: PC, Macintosh, Workstation, mimi, mainframe, Cray, etc.)
Any number of agents can be connected to any other number of agents via
communication links (shared memeory, LAN, modem, etc.)
Since each agent is an "intelligent" enetity running on a computer, and
since agents can communicate with one another and with outside
applications (PLC programs, accounting software, quality measurement
equipment, databases, etc.) then the agents can "reason together" about a
problem, situation, opportunity, etc.
Imagine a machine operators Man Machine Interface - a basic CRT screen and
pointing device. Running the MMI is an agent. The operator has just
finished a job and can now start any one of 15 jobs. His agent confers
with scheduling agents, finance agents, transportation agents, etc. and
then his agent recommends he start, specifically, Job 12 of 15.
The operator presses a button marked "WHY?" The agent explains that the
transportation agent reports that there is a truck waiting for exactly 17
of these widgets from Job 12, the finance agent says it will cost more if
the truck is not filled, the sales agent says the customer for this truck
will be angry if 17 widgets are missing, the scheduling agent has
rescheduled everything to allow making Job 12 early and all this is "WHY"
Job 12 should be done next.
The worker tells the MMI Agent "OK but I should make 18 because we lose 1
out of 20 at final inspection."
----------------------------------
The agent communicates and records this so that next time it will make 1
out of 20 extra.
At this point the agents have:
1. Found and recommended the optimum action for this operator.
2. They have also made coordinating recommendations to many other
personnel
3. In explaining WHY from this multi-disciplinary point of view the
worker
has gained wider knowledge of his function within the business -
in other words, the worker has received a bit of multi-disciplinary
TRAINING
4. The worker has also *trained the system* with point-specific knowledge
that the agents can now use to be even smarter, more optimum, etc.
Over the ensuing years this company's performance continues to rise, it
adapts swiftly to change, it remembers what works and what doesn't, its
workers are highly trained and have wide knowledge as well as deep,
manufacturing understands accounting better and vice versa, customers are
ecstatic. Workers' time freed-up by not having to perform the mundane
information collection and slow analysis of yester-year are busy
increasing variety and quality. Market share climbs, costs plummet,
profits rise, wages go up, shareholderget bigger dividend checks, more
capital flows into the company to improve equipment and the work
environment, and the whole cycle regenerates into ever new learning
opportunities for both the intelligent agents and the people.
Does this help at all?
-- Hal Popplewell GaltJohn22@aol.com