Extending Temporal Models in Timed Influence Networks

摘要

An extension to the formalism of Timed Influence Networks, including time-varying and cyclic influences is presented. The proposed time-varying influence functions allow modeling of influences whose strengths vary with time. A cyclic influence, on the other hand, provides a provision for a self- promoting set of influences. The two extensions will provide TIN models with the capability to depict PMESII (i.e., political, military, economic, social, infrastructure, and information) aspects of informal and uncertain domains, in the quest for the evaluation of various courses of action. An earlier mapping from a TIN model to a Time Sliced Bayesian Network is revisited and redefined for the new temporal extension. The temporal approach is illustrated with the help of a TIN model, used earlier for an effects based operations application.