【作者】 杨静；

【导师】 陈叔平；

【作者基本信息】 浙江大学 ， 运筹学与控制论， 2005， 硕士

【摘要】 本文在总结历史上群体决策理论研究的基础上,探讨了群体决策过程优化问题。着重研究了群体决策中专家评审这一典型问题的过程优化:采取何种决策过程使得在限定的时间内得到尽可能公正的决策结果,即同时兼顾决策效率与决策质量。 一般群体评审决策程序是由被评审对象或申报人按要求提供材料,派出一组专家评审这些材料,最终给出群体决策结果。但是当材料很多时,每个专家全部仔细读完所有待评审的材料并做出认真评判就会耗费过长的时间。若限定评审时间,又会影响决策的准确性,因为时间少了,他无法在评审过程中对每一指标进行细致的评判。这时通常采用分组的方法减少上述矛盾,达到在限定的时间内兼顾准确性和客观性。在分组的过程中,倘若分组过少,每个专家需要评审的材料的数量依然较大,在限定的时间内评审数量庞大的材料准确性依然不够;倘若分组过多,每份材料被审的专家人数过少,则客观性会不足。我们的目标是:合理分组使得评审在给定的时间内既客观又准确。本文建立了模型刻画评审的准确性和客观性,并将两者结合起来建立了整体绩效函数,求出合适的分组数使得整体绩效达到最大化。本文建立的模型对优化“限制决策时间和专家人数但备选方案个数很多”这一群体决策中的典型问题进行了有益的尝试。更多还原

【Abstract】 The study focused on the previous research on Group Decision-Making theories. We discussed how to optimize the process of Group Decision-Making and especially lay stress on a typical problem in Group Decision-Making field which is Expert Evaluation problem. We would provide a way to optimize the process of Group Decision-Making affected by a time limit and by human resource limit.In general, Expert Evaluation problem describes such a situation: the proposers provide materials → the leader asks for a group of experts to evaluate these materials → the experts provide the final Group Decision-Making result. However, when there are too many materials, what will happen? If every expert reads every material and gives an evaluation result, to finish the evaluation process will cost quite a lot of time. If the time is limited, the accuracy will be reduced. Thus we provide a way to solve the conflict: separate the whole group of experts into several small groups. If we separate the whole group into very few small groups, every expert still needs to read a lot of materials. Then the accuracy is still not high enough. If we separate the whole group into too many small groups, every material will be evaluated by very few experts. Then the objectivity is not high enough. Hence, we will try to find the reasonable number of the small groups such that the group performance which shows both the accuracy and the objectivity can reach the maximum.更多还原