In this paper, we describe an approach to handling partially specified probabilistic information. We propose a formalism, called Partial Probability Theory (PPT), which allows very general representations of belief states, and we give brief treatments of problems like belief change, evidence combination, and decision making in the context of PPT. We argue that the generality of PPT provide new insights in all the mentioned problem areas. More detailed treatments of these issues can be found in several papers referred to in the text.
Keywords. partial probability theory, partial ignorance, probabilistic belief change, conditioning, constraining, evidence combination, decision under partial ignorance, minimax regret, satisficing.
The paper is available in the following formats and sites:
Department of Computer Science
Related Web Sites