CUI Networks: A Graphical Representation for Conditional Utility Independence

Proceedings of the AAAI Conference on Artificial Intelligence, 21

Volume

Technical Papers

Uncertainty in AI

We introduce CUI networks, a compact graphical representation of utility functions over multiple attributes. CUI networks model multiattribute utility functions using the well studied and widely applicable utility independence concept. We show how conditional utility independence leads to an effective functional decomposition that can be exhibited graphically, and how local, compact data at the graph nodes can be used to calculate joint utility. We discuss aspects of elicitation and network construction, and contrast our new representation with previous graphical preference modeling.