Abstract
GA seems to be a useful approach to increase the practicality of AHP suitable for evaluating a set of alternatives. To this end, we supplied the lower and upper-bound values around users' judged values by the wave model, expecting to compensate for the partility in kansei judgment and expressions. The proposed degree of prominence associated with the best altemative helped us find out desirable combinations of the elements of <i L, R, U > for each evaluation. The intitial and subsequent generations were created by employing three coding schemes at each AHP level : monotype, biased and random coding. We could obtain fairly good results in several generations.
