Abstract for: Modeling of Network Dynamics: From Dynamic Nodes to Dynamic Structure

The paper presents a conceptual framework for modeling of dynamic systems with variable structure. A practical motivation comes from analysis of value networks – complex systems of stand-alone business entities that bond together, more or less tightly, through exchange of goods, services, and money. The existence of bonds between certain businesses, and their strength, are determined by the relative performance of individual businesses in terms of a value they add along the network. The collection of effective bonds defines the structure of a network. As the performance of businesses changes over time, so does the network structure. Better performing nodes are more likely to get bonded, and nodes with stronger bonds are better positioned to further improve their performance. Having an operational model of the value network behavior, with the capability to predict the changes in the network configuration as a result of the changes in the individual node performance, is a crucial prerequisite for effective management of the network performance. We propose a stochastic model of the dynamic behavior of a value network, which combines probabilistic graphical modeling with stochastic extension of system dynamics to model a network configuration and individual node performance, respectively.