2. Influence diagram
3. Decision class analysis and neural network
4. Neural network based process to build an ID
5. Interactive procedure to build an ID
6. An illustrative case example
Building an Influence diagram in decision analysis is known to be a most complicated and burdensome process. The use of neural networks to generate influence diagrams in the topological level results in a good performance, the generated ID is usually not a well-formed influence diagram. Furthermore it needs more modification to be applicable to real decision problems, especially when group decision participants are involved.
This research suggests an interactive procedure to build a well-formed influence diagram from the initial influence diagram generated from neural networks which are thought to be an approximation of experts’ (explicit or implicit) interpretation of decision problem. Our procedure is composed of two phases; one is to modify the influence diagram by each decision participant, the other one is to resolve the differences of group’s influence diagram interactively. We applied our procedure to an analogous land development and conservation problems. When the problem is complicated and group decision participants are involved, this research is expected to be more useful to inexpensively model a decision problem.
[BUN 84] BUNN D.W., Applied Decision Analysis, McGraw-Hill, New York, 1984.
[CHU 92] CHUNG T.Y., KIM J.K., KIM S.H., Building an influence diagram in a know-
ledge based decision system, Expert Systems With Applications, vol. 4, 1992, p. 33-44.
[HOL 89] HOLTZMAN S., Intelligent Decision Systems, Addison-Wesley, MA, 1989.
[HOW 84] HOWARD R.A., The used car buyer, in Readings on the Principles and
Applications of Decision Analysis. Vol. II, ed. R.A. Howard and J.E. Matheson, Strategic
Decision Group, Menlo Park, CA, 1984.
[HOW 88] HOWARD R.A., Decision analysis: practice and promise, Management Science,
vol. 34, 1988, p. 679-695.
[KIM 91] KIM J.K., A Knowledge-Based System for Decision Analysis, Ph.D. thesis,
Department of Industrial Engineering, KAIST, Korea, 1991.
[KIM 92] KIM J.K., CHUNG T.Y., KIM S.H., A knowledge-based decision system to build
an influence diagram: KIDS, Proceedings of the First World Congres on Expert
Systems, Orlando, Florida, 1992.
[KIM 95] KIM J.K., A Study on the Development of Intelligent Decision Systems Using
Influence Diagram, Journal of the Korean OR/MS Society, vol. 20, 1995, p. 77-104.
[KIM 97] KIM J.K., PARK K.S., Neural network-based decision clas analysis for building
topological-level influence diagram, International Journal of Human-Computer Studies,
vol. 46, 1997.
[KIM 98] KIM J.K., CHU, S.C., Sensitivity Analysis in the Decision Class Analysis Using
Neural Networks, Proceedings of the Fourth World Congres on Expert Systems,
[OLM 84] OLMSTED, S.M., On Representing and Solving Decision Problems, Ph.D.
thesis, Department of Engineering-Economic Systems, Stanford University, 1984.
[REE 89] REED, J., Building decision models that modify decision systems, Knowledge
System Laboratory, no. KSL-89-21, Stanford University, Stanford, CA, 1989.
[RUM 86] RUMELHART D., MCLELLAND J., Parallel Distributed Processing, vol. 1.
MIT Pres, Cambridge, Mas., 1986.
[SHA 86] SHACHTER R.D., Evaluating influence diagrams, Operations Research, vol. 34,
1986, p. 871-882.
An influence diagram based on neural networks 17
[SHA 88] SHACHTER R.D., Probabilistic inference and influence diagrams, Operations
Research, vol. 36, 1988, p. 589-604.
[SON 94] SHONNENBER F.A. et al., An Architecture for Knowledge-based Construction
of Decision Models, Medical Decision Making, vol. 14, 1994, p. 27-39.
[VOL 88] VOLKEMA R., Problem Complexity and the Formulation Proces in Planning
and Design, Behavioral Science, vol. 33, 1988, p. 292-300.
[WOO 81] WOOLLEY R., PIDD M., Problem Structuring A Literature Review, Journal
of Operational Research Society, vol. 32, 1981, p. 25-63.
[ZAH 91] ZAHEDI F., An introduction to neural networks and a comparison with artificial
intelligence and expert systems, Interfaces, vol. 21, 1991, p. 25-38.