Melinda T. Gervasio
A crisis is characterized by three elements: threat, urgency, and uncertainty. Decision-theoretic techniques, while attractive, are particularly difficult to implement for crisis domains as they require complete domain models that are difficult to construct given the urgent and uncertain nature of crisis. However, they remain attractive for crisis domains because they provide well-founded principles for making critical decisions. Machine learning techniques have been applied in automated planning to speed up this search by learning macro-operators or control rules that result in following more correct and shorter paths to acceptable solutions. We propose the application of machine learning to achieve similar efficiency gains in the construction of decision models. Specifically, we propose the development of an adaptive assistant that uses learning techniques to tailor its interactive decision modeling activity to individual users.