A model of decision-making with sequential information-acquisition (part 2)

摘要

While most real-life decision are of necessity made with less than perfect information, there is usually some opportunity to acquire additional information regarding the problem at hand before a final decision is made. It is, of course, the recognition of this fact which has led to the importance now attached to the field of Decision Support Systems. On the other hand, the formal analysis of the sort of decision problem for which Decision Support Systems can be useful appears to have lagged behind the developments in applications. In this paper we develope a model of decision-making in which there is available a variety of informational sources (experiments) which can reduce (though generally not eliminate) the uncertainty associated with the final decision. Since the informational sources are available only at some cost (either monetarily or in terms of time, or both), the decision-maker must solve two conceptually distinct problems: (1) developing an optimal information-gathering strategy, and (2) developing an optimal final decision strategy, conditional upon the information obtained during the information-gathering process. A theoretical framework is developed here for the analysis of this general problem, and fairly complete solutions are obtained for some interesting special cases; most notably the computer file search problem.