On Modelling Hybrid Uncertainty in Information

Numerical induction models are considered in this report to be models that aggregate lower-level information into higher-level measures for decision making. Various forms of uncertainty may be present in such models including hybrid uncertainties within the information elements being aggregated. After a review of some existing approaches for representing higher-order uncertainty in information, a new approach is presented to enable greater fidelity of uncertainty representation, and consequently more rigorous uncertainty management in aggregation operations. Several different applications then demonstrate the proposed procedures which have direct relevance to many Defence decision making models where higher-order uncertainty is ubiquitous. The overall objective of these procedures is to extract as much meaning from the input information as possible.