Modeling Search in Group Decision Support Systems

Abstract

Groups using group decision support systems (GDSS) to address particular tasks can be viewed as performing a search. Such tasks involve arriving at a solution or decision within the context of a complex search space, warranting the use of computerized decision support tools. The type of search undertaken by the groups appears to be a form of adaptive, rather than enumerative, search. Recently, efforts have been made to incorporate this adaptation into an analytical model of GDSS usage. One possible method for incorporating adaptation into an analytical model is to use an evolutionary algorithm, such as a genetic algorithm (GA), as an analogy for the group problem-solving process. In this paper, a test is made to determine whether GDSS behaves similarly to a GA process utilizing rank selection, uniform crossover, and uniform mutation operators. A Markov model for GAs is used to make this determination. Using GDSS experimental data, the best-fit transition probabilities are estimated and various hypotheses regarding the relation of GA parameters to GDSS functionality are proposed and tested. Implications for researchers in both GAs and group decision support systems are discussed.