A Bayesian approach to simultaneously quantify assignment and linguistic uncertainty

Abstract

Subject matter expert assessments can include both assignment and linguistic uncertainty. This paper examines assessments containing linguistic uncertainty associated with a qualitative description of a specific state of interest and the assignment uncertainty associated with assigning the state to a particular qualitative value. A Bayesian approach is examined to simultaneously quantify both assignment and linguistic uncertainty in the posterior probability. The approach is applied to a simplified damage assessment model involving both assignment and linguistic uncertainty. The utility of the approach and the conditions under which the approach is feasible are examined and identified.