A Method for Assigning Men and Women with Good Affinity to Matchmaking Parties through Interactive Evolutionary Computation

摘要

In this paper, we define a matchmaking party assignment problem and propose a system to solve it. The problem is to assign male and female participants to several small groups so that each group consists of the same number of men and women who have a good affinity for each other. The proposed system solves the problem based on an IEC (interactive evolutionary computation) framework, which can treat indefinable evaluation functions such as affinity between men and women by feeding back the empirically obtained values of those functions. Given each participant's attributes such as bodily characteristics, academic background, and personality, which are obtained by questionnaire in advance, the system assigns the participants to several small groups in order to maximize the number of man and woman pairs likely to begin relationships. After each groups party, the number of pairs who liked each other can be obtained as a value of the evaluation function for EC (evolutionary computation). To evaluate the system, we define the JVMax Problem assuming that there would be N good affinity patterns between men and women. Through computer simulations with N from 2 to 5, we confirmed that the proposed system could find a much better group assignment than a greedy approach.