Ambiguous Decision Support by AHP
MATSUDA, Noriyuki IPPS, University of Tsukuba, Japan
AHP (Analytic Hierarchy Process), proposed by Saaty ( 1980), is one of the most popular decision support tool in many areas. It derives relative priorities for a given alternative set from the weights accorded to the hierarchically arranged decision items. Its user-friendliness stems from the localized evaluations within a hierarchy, since our natural judgments are mostly kanseic rather than rational. The contrast does not mean that kansei is irrational. Instead, they both constitute human intelligence, but differ in orientation as follows ( Matsuda, 1997, 1999):
Rationality: Intelligent capacity oriented toward unambiguity, precision, rigor, consistency,...
Kansei: Intelligent capacity that allows partial deviations from such standards.
In short, kansei enables us to live with practical efficiency when relevant information is partially available and/or optimal procedures are not completely known in the ever changing environments. Within reasonable limits, however, kansei leads us to attain, to a satisfactory extent, what rationality strives for.
Taking advantage of the hierarchical structure of ABIP, its user can concentrate on localized or partitioned judgement, leaving systematic information integration to the tool. Nevertheless, we believe that the standard procedure of pairwise comparisons are too fragmented and repetitive for the user to maintain consistency even in the partitioned classes. As a solution, we suggest simultaneous multiple comparisons, making use of interactiveness of graphical computer interface ( Matsuda , 1999).
To further enhance the practicality of ABP, we propose here an application of GA (Genetic Algorithm) in expectation of reaching a set of priorities among awfully many possibilities arising from ambiguities inherentin ordinary judgment. A procedural tool that bridges between GA and AHP is called a wave model