DEVELOPMENT OF GROUP WARE BY MULTI-OBJECTIVE DECISION ANALYSIS

群件支持群体协作的一个重要功能就是支持群体决策。在群体决策支持系统的开发中，通过增进群体间的相互理解来促进群体间的共识是非常重要的。组由具有各种值范围的成员组成。多目标决策分析的目的是从全局的角度出发，通过对各种不同的取值范围进行适度的调整，找到一个满意的解。在这项研究中，我们扩展了满意的权衡方法，这是最初开发的情况下，与单个决策者，群体决策，并将其纳入一个群件。在本研究中，有介质和无介质的情况下处理。有中介者的案例是等级决策问题，没有中介者的案例是自主决策问题。我们利用满意折衷法和其它多目标决策分析方法，如ISM和AHP，开发了一个群件，该群件设计在人机界面上，以适用于许多实际问题。此外，我们还开发了一个使用晶片机的委员会机器，这是一种复杂的神经网络。它可以有效地模拟群体决策，为群体软件的开发提供丰富的信息。最后，对群决策规则进行了详细的理论分析，并强调了其实际意义。这些决策规则进行检查，从一个实际的角度来看，通过模拟纳入一个群件。

One of the most important function of group ware for supporting cooperation in group is to support grpup decisions. It is very important in developing group decision support systems to promote to make consensus among the group by making better mutual undestanding. Groups consist of members who have various value scopes. The multi-objective decision analysis aims to find a satisfactory solution from a total view point by adjusting various different value scopes moderately. In this research, we extend the satisficing trade-off method, which was originally developed for cases with single decision maker, to group decisions and incorporate it into a group ware. Both xases with mediators and ones without mediator are treated in this research. Cases with mediators are hierarchical decision problems, and those without mediators are autonomous decision problems. We developed a group ware using satisficing trade-off method and other methods of multi-objective decision analysis such as well known ISM and AHP.This group ware is devised in human-interface in order to be applicable to many practical problems. In addition, we developed a committee machine using transputers, which is a kind of complex neural networks. It can be effectively used to simulate group decisions and provides much information in developing group wares. Finally, theoretical analysis for rules of group decisions was made in detail with emphasis on practical implication. These decision rules are examined from a practical view point through simulation by incorporation into a group ware.