协同工作下角色-任务匹配与互动多智能体模拟研究  

Agent-based Simulation for Matching and Interaction between Roles and Tasks Under Collaborative Work Environment

作　　者：蒋国银[1,2] 胡斌[1] 

机构地区：[1]华中科技大学管理学院,武汉430074 [2]湖北经济学院信息管理学院,武汉430205

出　　处：《管理科学》2009年第6期56-63,共8页Journal of Management Science

基　　金：国家自然科学基金(70671048)~~

摘　　要：利用多智能体模拟方法对角色-任务匹配和互动关系进行研究,构建基于Agent的角色-任务匹配的多目标评价模型,并设计最小匹配度和最大匹配度两种匹配度算法,基于Repast,在Eclipse上用Java实现角色-任务互动模拟系统。模拟实验结果表明,与最大匹配度算法相比,最小匹配度算法能减少任务接口通讯费,有效提高员工能力利用率,最大化完成任务。该算法能较好地用于协同工作环境下角色任务分配,增加学习率,有助于员工能力增长;在一定的协同学习率下,任务量越饱和,员工各项能力提高程度越大;任务的能力需求期望变化小于方差变化时体现的动态性与能力增长负相关,而任务的能力需求期望变化大于方差变化时体现的动态性与能力增长正相关。This paper employs the agent-based simulation method to study the matching and interaction between roles and dynamic tasks. Firstly, a multi-object evaluation model based on agent is established to demonstrate matching between roles and tasks, and then, the minimization matching and the maximization matching algorithms are designed to solve the model. As for different numbers and dynamics of tasks, some organization performances are analyzed by simulation experiments. On the basis of Repast, roles and tasks interaction model system is realized with the use of Java on Eclipse software Results show that the minimization matching algorithm is a better matching algorithm than the maximization matching algorithm for dynamic tasks allocation in business organization, and under a certain collaborative learning rate, the heavier the number of task becomes, the greater work ability of employee grows, and dynamic of tasks have an impact on employee learning. When the change of task ability expectation is smaller than the change of variance, the dynamism is negatively related with the advance of ability, however, when the change of task ability expectation is larger than the change of variance, the dynamism is positively related with the advance of ability.

关 键 词：多智能体模拟 匹配 动态任务 协同学习率 

分 类 号：C931[经济管理—管理学]