Utility Elicitation as a Classification Problem

The majority of real-world probabilistic systems are used by more than one user, thus a utility model must be elicited separately for each newuser. Utility elicitation is long and tedious, particularly if the outcomespace is large and not decomposable. Most research on utility elicitation focuses on makingassumptions about the decomposability of the utility function. Here we makeno assumptions about the decomposability of the utility function; rather we attempt to cluster a database of existing user utility functions into a small numberof prototypical utility functions. Having identified these prototypes, wecan then effectively classify a newuser's utility function by asking manyfewer and simpler assessments than full utility modelelicitation wouldrequire.

• Type: Conference
• Year: 1998
• Where: UAI
• Authors: Urszula Chajewska, Lise Getoor, Joseph Norman, Yuval Shahar
• Keywords: UAI 1998, UAI 2008, Utility Elicitation, Utility Function, Utility Model