|
|
|
| Article
Summary: |
Expert Systems |
4.30.99
Human expertise in printing can be represented by computer programs called expert systems.
If the problem is well defined with clear inputs and outputs, expert systems can be
developed. These follow a chain of reasoning between observations and solutions.
Printing issues which are candidates for expert systems include diagnosis, control,
configuration and planning. Proper and timely decisions in these areas can increase
efficiency and reduce waste. An expert system can be used as a replacement for human
decision-making, or as a training aid for employees.
Expert systems express the knowledge and decision-making process as an integrated tree of
if/then rules based on one of more inputs which are evaluated with rules like (if [A and
B] then C) or (if [A or B] then C). Condition C is can then used in further
decision-making. The rules can be traversed in the forward direction, from observed data
toward a goal. They can also be traversed in the backward direction, from an anticipated
goal back to observed facts which validate the goal. Either way, a decision to implement
the goal is supported.
Different rules and observations can be assigned a confidence factor which weighs some
higher than others. With prioritized paths through the decision tree, confidence in
multiple plausible solutions can be compared.
Authors: Souhaila Almutawa and Bader Alhajji, Kuwait University
Source: GATFWorld, March/April 1999, p. 11-15 (part 3 of 5). |
|
