Using Feature Value Distributions to Estimate Player Satisfaction Through an Author's Eyes

David L. Roberts, Christina R. Strong, Charles L. Isbell

We explore an approach for inferring player preferences for interactive entertainment. The goal is to be adaptive and scalable by extracting preferences through observation, and using a vocabulary understood by game authors rather than depend upon the quirks of individual players. We demonstrate our approach for a number of simulated players in two simulated games and present results from a feasibility study.

Subjects: 3.4 Probabilistic Reasoning; 1.8 Game Playing

Submitted: Sep 14, 2007