Conditional Utility, Utility Independence, and Utility Networks

TL;DR

This paper introduces a new interpretation of conditional utility and utility independence, aligning them with probabilistic concepts, and develops utility networks that efficiently represent utility distributions similar to Bayesian networks.

Contribution

It presents a novel interpretation of utility concepts and introduces utility networks, enabling compact representation of utility distributions akin to Bayesian networks.

Findings

Utility distributions mirror probability distributions structurally.

Utility networks provide a compact representation of utility distributions.

The new interpretation aligns utility notions with probabilistic counterparts.

Abstract

We introduce a new interpretation of two related notions - conditional utility and utility independence. Unlike the traditional interpretation, the new interpretation renders the notions the direct analogues of their probabilistic counterparts. To capture these notions formally, we appeal to the notion of utility distribution, introduced in previous paper. We show that utility distributions, which have a structure that is identical to that of probability distributions, can be viewed as a special case of an additive multiattribute utility functions, and show how this special case permits us to capture the novel senses of conditional utility and utility independence. Finally, we present the notion of utility networks, which do for utilities what Bayesian networks do for probabilities. Specifically, utility networks exploit the new interpretation of conditional utility and utility independence to compactly represent a utility distribution.