中文摘要：

为探索交互记忆系统对知识型团队中人群与工作互动的影响，交互记忆系统通过基于多智能体建模被引入到模拟研究中。用互动网络模型描述任务、个体、技能的复杂联系，用交互记忆系统中任务执行、沟通、学习规则驱动此模型，考虑经济型和社会型两种智能体在交互时受环境因素的不同影响，设计通过数据库作为中间介质的人一物并行机制，并采用甘特图描述物系统中的任务计划和任务分配。用Repast和Java Swing实现模拟系统，并用定性验证的方法确认该系统的可信性，最后给出一个应用分析。模拟实验研究表明，交互记忆系统可以提高任务完成质量，并缩减任务完成时间；一个团队在完成任务时并不是参与人越多越好，人在足够满足任务完成需要的情况下与工作有效交互可以发挥较高的效率。

英文摘要：

To research the interaction between group and task in knowledge-based team, interactive memory is introduced to computer simulation over agent-based modeling. AMTMGTI （agent-based model of interactive memory for group-task interaction） is proposed to describe the mechanism in group-task interaction. There are many relationships among tasks, individuals and skills of a project team in complex team network. We use some action rules related to task execution, communication and learning in interactive memory system to drive this AMTMGTI model. To combine task sub-system with person sub-system, the database is acted as media to communicate. The Gantt graph analysis is used to represent task schedule and task allocation in task sub-system. In implementation, we use open source Java Repast and Java Swing for programming. Then, through qualitative validation methods, we can verify the validation of this model. Finally, we give analysis of an application. Simulation study shows that due to the transactive memory system, the quality of accomplishment can be improved and the time spent on the accomplishment can be cut down. The norm that the more people join the better for accomplishment isn＇t to the point, it can exert a better efficiency to interact with the job availably when one can meet the needs to complete tasks adequately.