Implementing “Case Based Reasoning”

advertisement
 2001 UICEE
4th UICEE Annual Conference on Engineering Education
Bangkok, Thailand, 7-10 February 2001
University,
Erol Inelmen
Implementing Bogazici
“Case
Based Reasoning” in
Bebek, Istanbul, Turkey
Engineering Management Education
ABSTRACT. “Case Based Reasoning” (CBR) is emerging as a new Artificial Intelligence tool. This new tool can be used in design
activities where a large number of already completed “cases” can help the experts in finding solutions to similar new problems. The
success in the implementation of this challenging technique has prompted the author to seek new areas of application. In a previous work,
project management was selected as a new area for implementation for CBR. The possibilities of extending the experienced gained to
education management is presented in this paper. Several authors have recently voiced the view in favour of a more soft approach –using
less mathematical models- in engineering management. This approach hopefully is more in touch with the realities in the practical realm.
INTRODUCTION
There is no doubt that as computer applications become more
popular, research work on Engineering Management is
becoming even more challenging. Although there is a vast
number of newly developed managerial techniques [1] it is
necessary to admit that many solutions have failed to meet the
expectations of many customers in the practical realm [2]. As the
world economical, social and technological conditions change at
a very high speed we argue here that experience gained from
earlier completed managerial work can help in dealing new
managerial projects more efficiently [3].
This paper attempts to bring past experience in perspective with
the hope that students involved in Engineering Management will
be able to provide better solutions to real life problems of
customers [4]. As suggested recently by [5] modern management
should consider the fact that the actual process of “modelling” an
activity is as important as the activity itself. After presenting the
results obtained previously in the implementation of “Project
Scheduling”, in the light of the new developing ideas, a general
and novel scheme for “Engineering Management” is here
suggested. The creation of a “case-based expert system” for
management based on past experience is the end product
envisaged by the author.
.
CASE BASED REASONING
Since management aims at coordinating the available resources
–men, material, machinery, money and information- in the
realisation of the objectives of a given task [6] the most valuable
asset of a company is without doubt the experience gained in
previous years. If the collective experience of the company could
be gathered in a “repository” of past cases, the possibility of
success in future activities can be enhanced. As cases must be
abstracted into symbolic form so that they can be manipulated in
the future [7], convenient parameters must be selected for each
factor that would characterise each individual case. This method
is used currently in engineering design as a tool to replay and
modify past experience [8, 9], provide the theoretical and
practical background in the creation of a “Case Based
Reasoning” repository.
Following the recent work on “Case Based Reasoning”
–summarised in the previous paragraph- an attempt was made to
create a repository of knowledge based on previous project
management experience obtained from a local corporation.
Using the outline given by Chapman and Ward, the input factors
relevant in a project were grouped under the following headings:
parties, motives, design, activities, resources and timetable [10].
In this repository attributes were carefully selected and values
for two cases, named simply A and B where elucidated during
three interviews made by the two authors [11]. Interviews took
place in the company from where the data was collected at
intervals of two weeks. During this period, the authors had the
opportunity of reflecting separately on the progress made. The
values for a new case –named as Case P, now in the phase of
realisation- have been predicted based on past experience. The
results obtained presented at a workshop held in Istanbul, were
found to have a potential in the field of management if enough
cases were compiled.
ENGINEERING MANAGEMENT
The warm welcome in the audience to the “Case Based
Reasoning” repository - described in the previous section and
presented at the workshop held in Istanbul- suggested the
possibility of extending the technique to other operation
management applications. The challenging work by Fortuin, van
Beek and Van Wassehove [12] encouraged the author of this
paper to classify operation management activities as shown in
the Appendix. Knowledge based engineering –an emerging
discipline- promises to provide new tools in helping experts in
“inferring” new ideas from previously acquired data [13, 14]
The work presented in this paper is by no means complete. We
must accept that the attributes selected may not all be
representative and the values given and each attribute cannot be
always fully assessed. The intention is to encourage the parties
involved to follow the steps that have been laid in the first trial in
project management. Since “Case Based Reasoning” is currently
been successfully implemented in engineering and architectural
design we argue that it is possible to transfer this experience to
the field of engineering management. Major contributions in the
improvement of overall performance of new projects can be
expected.
FUTURE WORK
REFERENCES
As highlighted in the previous section this paper’s aim is to
introduce the new Artificial Intelligence technique known as
“Case Base Reasoning” in the realm of Engineering
Management. When the attributes that distinguish an operational
problem are adequately indexed it is possible to identify a new
case and recall a method that can be used to find the possible
solution. Using the experience of a commercial software
provider (see www.promodel.com) new operation management
tools can be implemented.
The success of the new techniques will depend on the richness of
the repository of previous cases recollected. These cases maybe
classified from the point of view of a) area of specialisation,
namely: production, service, process, medicine, handling,
transportation, assembly b) area of analysis, namely: throughput,
cycle-time, bottleneck, productivity, allocation, capacity, layout,
balancing, warehouse, buffering and c) tool used, namely:
modelling, optimisation, simulation. Using the techniques here
proposed, data collected can be efficiently classified and the best
match between new cases and old cases can be obtained.
CONCLUSION
“Case Base Reasoning” a new managerial tool that was
developed from Artificial Intelligence science and practice, can
help experts and users to make better inferences from previous
experience. Using this new technique we hope that the decision
making processes will be enhanced. Sharing past experiences in
a systematic way may eventually lead to general conclusions on
managerial best practices. A more holistic approach and process
orientation to management is suggested as a means to increase
the value added by the expert intervention.
ACKNOWLEDGMENT
The inspiration given by Dr. Taner Bilgic in the Department of
Industrial Engineering in the Bogazici University is
acknowledged.
The author wishes to express the gratitude to the UNESCO
International Council for Engineering Education for the support
given since 1998. The experiences gained in the conferences
organised since have helped the author to change the educational
approach he is implementing
1. Ormerod, R.; On the Nature of OR- Entering the Fray, Journal
of the Operational Research Society, 47:1, 1-17. (1996).
2. Ackoff, R.L.; The future of operational research is past.
Journal of the Operational Research Society, 30, 93-104.
(1979).
3. Kaynak, O. and Sabanoviç, A.; Diffusion of New
Technologies Through Appropiate Education and
Training. Diffusion of New Technologies Conference,
St.Petersburg. (1994).
4. Checkland, P.; Achieving desirable and feasible changes: an
application of Soft System Methodology”, Journal of the
Operational Research Society, 9, 821-831. (1985).
5. Pidd, M.; Tools for Thinking: Modelling in Management
Science.
John
Willey
and
Co.
(1997).
6. Bent, J.A., Humpreys, K.K. Effective Project
Management Through Applied Cost and Schedule Control.
Marcel Dekker, Inc. New York. (1996).
7. Maher, M.L., de Silva Garza, A.G.; Case-based Reasoning in
design. IEEE Expert- Intelligent Systems, 12:2, 34-41.
(1997).
8. Kim, G.J.,; Case-based design for assembly. Computer-Aided
Design, 29:7, 497-506. (1997).
9. Bilgic, T., and Fox, W. Constraint-Based Retrieval of
Engineering Design Cases. In: Artificial Intelligence in
Design ’96 ( J. S. Gero and F. Sudweeks, eds.), Kluwer
Academic Publishers, Netherlands, 269-288. (1996).
10. Chapman, C. and Ward, S.; Project Risk Management. John
Wiley & Sons, Chichester, England, 4-7. (1993).
11.Petekciler, B. and Inelmen, E.; Experience gained in coping
with the uncertainties of Project Management”, In: Sixth
International Workshop on Project Management and
Scheduling (G. Barbarosoglu, S. Karabatlı, L. Ozdamar
and G. Ulusoy, eds), Istanbul, 121-124. (1998).
12.Fortuin, L., van Beek, P. and van Wassenhove, L; OR at
Work, Taylor and Francis, Ltd, London. . (1996).
13. Morley, J. and Ormerod, R.; A language-action approach to
Operational Research”, Journal of the Operational
Research Society, 47:6, 731-740. (1996).
14. Brocklesby, J. and Cummings, S.; Foucalt plays Habermas:
An alternative philosophical underpinning for Critical
System Thinking”, Journal of the Operational Research
Society, 47:6, 741-754. (1996).
APPENDIX: Example Attributes for Operation Management Activity Parameters
Group
Item
Attribute 1
Attribute 2
Attribute 3
Character
Scope
Simple
Connected
Complex
Level
Strategic
Tactical
Operational
Environment
Static
Evolutionary
Dynamic
Goal
Single
Multiple
Fuzzy
Data
Small
Large
Huge
Relations
Simple
Connected
Complex
Expectations
Low
Middle
High
Difficulties
Small
Large
Huge
Techniques
Simple
Connected
Complex
Solutions
Single
Multiple
Fuzzy
Results
Single
Multiple
Fuzzy
Conclusions
Simple
Connected
Complex
Research
Inventory
Cost
Quality
Design
Layout
Handling
Organization
Operation
Planning
Control
Audit
Mathematics
Statistics
Optimization
Reliability
Systematic
Integration
Simulation
Reasoning
Introduction
Description
Acquisition
Selection
Conclusion
Formulation
Validation
Evaluation
User
Commodity
Interactive
Supportive
Expert
Service
Facilitator
Capstone
Conditions
Stage
Tools
Process
Application
Note: From 15 OR cases presented in Fortuin, van Beek and Van Wassehove (1996)
Download