Scalable Enterprise Systems: Emergent Organizational Structures in Multi-Agent Scalable Enterprises

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

This grant provides funding for the development of models of emergent enterprises that capture the independent behavior of each decision agent involved, as well as the effect of the interactions among agents. The modeling approach entails several steps. First, a representation scheme for the characteristics of individual participants will be developed. This representation will incorporate task information, organizational relationships, local and system level goals and possible changes in the environment. Second, the agent representation will be transformed into a state space representation that is better suited for the overall study of the dynamics of the system. Third, in order to incorporate the interactions between various agents, the problem will be formulated as an interacting particle system. A renormalization scheme will be used to reduce the system to a manageable size while maintaining an adequate level of accuracy. Different levels of aggregation will be tested to examine the tradeoffs between computational complexity and accuracy. Finally, the modeling approach will be validated using data collected from leading companies in the food industry. These data will be used to construct agent representations and then to verify the performance estimates based on real world observations. Large scale multi-agent simulations will be used to empirically estimate the performance of a given organizational configuration and the results will be compared with the predictions of the analytical model. If successful, the results of this research will allow accurate prediction of the dynamics of emergent organizations and the performance of the system in the long run. In addition, these models will provide insight into understanding the mechanisms that result in effective alliances and organizational designs. These models can be used by decision agents in an extended enterprise to estimate the performance of the enterprise resulting from different local decision strategies.