Explaining recommendations generated by MDPs

There has been little work in explaining recommendations generated by Markov Decision Processes (MDPs). We analyze the difculty of explaining policies computed automatically and identify a set of templates that can be used to generate explanations automatically at run-time. These templates are domain-independent and can be used in any application of an MDP. We show that no additional eort is required from the MDP designer for producing such explanations. We use the problem of advising undergraduate students in their course selection to explain the recommendation for selecting specic courses to students. We also propose an extension to leverage domain-specic constructs using ontologies so that explanations can be made more user-friendly.

• Type: Conference
• Year: 2008
• Where: EXACT
• Authors: Omar Zia Khan, Pascal Poupart, James P. Black
• Keywords: Applied Computing, EXACT 2008, Little Work, Markov Decision, MDP Designer