基于知识观的原始性创新多Agent仿真研究

Multi-agent Simulation Research of Original Innovation Based on Knowledge View

通过构建基于知识观的原始性创新多Agent模型,剖析原始性创新支撑知识、发起与参与主体以及主体合作网络之间的联系,进而完成主体行为描述及其适应性规则设计,并且借助Swarm平台实施仿真。结果显示学科研究方向总数极小的系统,其原始性创新绩效最优。对于学科研究方向总数中等乃至稍多的系统,采用优先适应其他主体原始性创新的策略或者强强联合的策略都可以显著提升其原始性创新绩效,二者相比不存在绝对的最优。

It proposes a multi-agent model on the original innovation, and analyses the relationships among the supporting knowledge system, its participators, and participators collaboration network, design agents＇ behaviors as well as their adap- tion rules, and simulate the model on Swarm platform. It shows that the system with very few research directions tends to have high innovation performance. As for systems with middle or higher amount of research directions, adopting the strategy to cooperate with the original innovations proposed by others with priority and the strategy to induce cooperation between ex- perts may both improve the system innovation performance. These strategies both have their individual advantages, and can＇ t conclude the best one.