2011 | OriginalPaper | Chapter
A Method for Transferring Probabilistic User Models between Environments
Authors : David L. Roberts, Fred Roberts
Published in: Interactive Storytelling
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Chief among the inputs to decision making algorithms in narrative or game environments is a model of player or opponent decision making. A challenge that will always face designers is to specify that model ahead of time, when actual data from the environment is likely not to be available. Absent corpora of data, designers must intuit these models as best they can, incorporating domain or expert knowledge when available. To make this process more precise, we derive a theoretically grounded technique to transfer an observed user model from one domain to another. We answer the question: “How can a model obtained from observations of one environment inform a model for another environment?” We verify the accuracy of our techniques using data from previous user studies.