中文摘要：

在多智体社会网络中,传统的任务分配模型一般采用直接面向任务执行者的分配机制.它们不考虑社会网络组织结构对任务分配性能的巨大影响,也很少透彻地研究不可靠社会中的任务分配.针对这些问题,开创性地研究了软/硬件合一系统的任务分配,即按递阶、分层的思想设计了协作组织模型,并基于此提出了面向社区基于社会协调“软件人”的任务分配模型.模型研究过程中,提出了基于直接信任度和社区声誉的社区信任度评估机制、基于社区信任度和社区物理能力的节点选择机制、基于负载均衡的社区内任务分配机制和基于上下文资源的任务再分配策略.实验结果表明：与常见的直接面向任务执行者和基于资源的任务分配模型相比,所提出的模型具有更优的任务分配性能,且对社会任务环境变化具有更好的鲁棒性;社区内基于负载均衡的分配机制和基于上下文资源的再分配策略也有效提高了分配性能,降低了网络中的通信密度.

英文摘要：

In the multi-agent social network, the traditional task allocation models generally adopt allocation mechanism that directly orients to task performers. They do not consider the huge impact of social network structure on the performance of allocation, and seldom thoroughly research the allocation mechanism for the unreliable society. Aiming at these problems, this paper firstly designs a collaborative organization model according to the hierarchical and layering methodology, then proposes a community oriented task allocation model for software-hardware syncretic systems. In the process of model research, the paper develops a community trust evaluation mechanism based on direct trust and community reputation, a community node selection mechanism based on trust degree and physical ability of community, a load balancing mechanism which is applied to the task allocation in interior community, and a task redistribution strategy based on the context of community resources. The results of experiments show that the proposed model has better allocation performance and robustness compared with other classical models, and also validate that the load balancing allocation mechanism and the redistribution strategy can not only effectively improve the allocation performance but also reduce the communication density of the social network.