On Metrics for Information Value Quantification

Abstract

In this paper, we aggregate, analyze, and exemplify several metrics for the information value. The metrics vary by absolute value, normalizations, or full probabilistic quantification. The normalization of the information value encompasses the division by (1) the expected and maximized utility of the base scenario (usually without additional information), (2) by the system performance, or (3) by the information acquirement costs. The latter can be aligned to the return over investment ratio to support and facilitate SHM system investment decision. With a case study on malevolent attacks to a generic iconic bridge, we exemplify the information value metrics and point to the application ranges and the implications of full distribution considerations.