Abstract: The decision-making experts in the decision-making group have a wide range of sources, with different knowledge backgrounds and emergency experience, and the use of the crowd wisdom knowledge of the decision-making group is more conducive to the solution of emergency incidents. However, emergency events have high complexity, uncertainty, and dynamic mutation, and how to make good use of the advantages of large decision-making groups and properly deal with hesitation risks and conflicts of opinion is an urgent problem that needs to be solved in emergency event decision-making.
In the literature on the risk of large-group decision-making, most of the literature does not consider the consensus reaching process, in the context of emergency decision-making, there is a lot of unknown information in the early stage of decision-making, and the consensus reaching process is also a process of exchanging opinions and modifying preferences among experts, and the lack of consensus will lead to a decline in the quality of decision-making. In order to obtain results with high decision-making validity, two conditions need to be met: first, the final ranking of alternatives needs to be recognized by most decision-making experts, that is, the level of consensus of large groups is high enough; Second, the degree of uncertainty of group decision-making information is low.
Both consensus and risk affect the quality of decisions in large groups of emergency decisions, while fewer articles consider both consensus and risk. In order to obtain more robust decision-making results, firstly, a decision-making validity measurement method is proposed according to the consensus level and hesitant risk level. Then, a large group is dynamically divided from the two dimensions of consensus and hesitant risk, and four basic subgroups are obtained: core subgroup, bias subgroup, risk subgroup and invalidity subgroup. Then, with the goal of improving the validity of decision-making, the idea of “divide and conquer” is used to build corresponding management mechanisms according to the characteristics of subgroups, so that the decision-making results are more stable.
After the outbreak of the epidemic, the government immediately organized a number of relevant government departments with a total of 20 experts to discuss the plan to control the epidemic, according to the epidemic prevention situation to formulate a blocking strategy, influenza strategy, reduce the flow of people, appropriate measures to increase social distance, according to the method proposed in this paper to obtain the optimal plan for the blocking strategy.
Compared with research that only considers the level of consensus or the risk of hesitation, this paper aims at decision validity, and proposes a subgroup identification and management method, which can obtain higher quality decision-making results and can be applied in a wider range of decision-making situations.
In the context of emergency decision-making of risky large groups, there are still many issues worthy of in-depth study on how to define decision-making validity and improve decision-making validity, such as considering the impact of minority opinions on large-group decision-making, the impact of social network relationships on large-group decision-making, etc., and future research will further expand the management mechanism of subgroups.

Key words: large group, risk, emergency decision, dynamic subgroup, decision validity

摘要： 针对大群体风险应急决策问题,提出动态子群体识别和管理方法。首先,提出决策效度测度方法;然后,从共识和犹豫风险两个维度对大群体进行动态划分,得到核心子群体、偏差子群体、风险子群体和低效度子群体四个基本子群体;接着,以提高决策效度为目标,根据四个子群体的特征分别构建不同的子群体管理机制。该机制不仅能提高大群体的共识水平,同时能降低大群体的犹豫风险水平,保证了决策质量。案例分析验证了所提方法的可行性。

关键词: 大群体, 风险, 应急决策, 动态子群体, 决策效度