Expert systems implementation:. interviews with knowledge engineers
| DOI | https://doi.org/10.1108/02635579510101447 |
| Published date | 01 December 1995 |
| Pages | 3-7 |
| Date | 01 December 1995 |
| Author | Terry Anthony Byrd |
| Subject Matter | Economics,Information & knowledge management,Management science & operations |
Expert systems
The deployment of expert systems (ES) in public and
private organizations appears, if anything, to be
accelerating[1]. ES are systems that consist of, minimally,
a knowledge base, an inference engine, and a user
interface. The knowledge base include facts, heuristics
(e.g. hunches, experiences, rules of thumb, gossip,
opinions, views, judgements, predictions, algorithms),
and relationships usually gleaned from the mind of at
least one expertin the relevant domain. ES provide a
standardized methodological approach to solving
important, fairly complex problem tasks that normally
require human expertise[2].
Many researchers and writers have written on the benefits
that ES bring to businesses (e.g.[1,3-5]). Although these
benefits have been overstated, there is some empirical
evidence that the effects of ES are indeed positive (e.g.[6]).
ES are becoming common decision- making tools in many
organizations. Typically, ES are not able to take over large
complex decision-making processes and problem-solving
tasks completely. However, they are helpful in fairly
standardized situations where there are a moderate
number of choices to consider and the evaluations of these
choices are extremely difficult.
In this article, insights on the nature and effect of ES
implementations in real organizations, gained from
interviews with knowledge engineers (KEs), are reported.
Since the information reported here is from KEs, it is
important to understand who they are and what they do.
KEs are concerned with identifying the specific
knowledge which an expert uses in solving a problem[7].
Contrary to popular belief, a KE does not use mystical
techniques to “mine” knowledge nuggets from the minds
of experts. Actually, a KE uses various communications
techniques (often referred to as knowledge acquisition
(KA) techniques) to elicit data and information from the
expert[8]. The use of these techniques calls for the KEs to
have good communications skills and a working
knowledge of psychology and cognitive science. A good
overview of these KA techniques is given in[9].
Once the KE has acquired the relevant data and
information, the KE then interprets the data and
information (more or less skilfully) to draw conclusions
on what might be the expert’s underlying knowledge and
reasoning processing[8]. The conclusions drawn by the
KE are used to drive the development of a model
commonly implemented in an artificial intelligence (AI)
language (e.g. LISP or PROLOG) or ES shell (e.g. VP-
Expert, KEE, KnowledgePro).
An interactive process is carried out by the KE and
expert as the ES model evolves into a functional
system[8]. An ES usually goes through continual updates
so a KE is normally available to maintain the system over
its life. These updates are required for a few reasons.
First, minor and major adjustments must be made to the
knowledge base as “bugs” are uncovered. Second, new
knowledge must constantly be added to the system as
changes occur over time and new ideas, concepts, and
knowledge become realities.
Because of this continuing relationship with the ES they
help develop, KEs are in an excellent position to discuss
the nature and effect of ES in their organizations. They
are intimately involved with the experts that help develop
the ES and also with the introduction to and training of
the users. From this position, the KEs typically observe
first-hand the implementation process and all its
implications. It is for this reason KEs were chosen to be
interviewed about the changes going on within their
organizations resulting from ES implementations. The
KEs had recently attended a conference on ES in
production and operations management.
Research method
The author telephone interviewed 28 KEs who had
responded to an earlier mailed questionnaire and had
consented, through a question on that questionnaire, to
the interview. Of the KEs 74 had answered and returned
the questionnaire. These KEs had attended a conference
that focused on the development of ES in production and
operations management. Statistical tests found no
difference (at the 0.01 significance level) between the 28
KEs who agreed to an interview and the remaining 46
respondents in demographics or in other variables on the
3
EXPERT SYSTEMS IMPLEMENTATION: INTERVIEWS WITH KNOWLEDGE ENGINEERS
Expert systems implementation:
interviews with knowledge engineers
Terry Anthony Byrd
Some of the benefits attributed to expert systems are actually being realized in today’s organizations
Industrial Management & Data Systems, Vol. 95 No. 10, 1995, pp. 3-7
© MCB University Press Limited, 0263-5577
To continue reading
Request your trial