How to revise ranked probabilities
Emil Weydert · Max Planck Institute for Plasma Physics · 2000
In this paper, we introduce and discuss a new framework for the modeling and revision of probabilistic belief. The epistemic states encode degrees of belief topped by second-order uncertainty using special Spohn-type ranking measures over subjective probability distributions. The revision strategy, which handles incoming information translated into linear probability constraints, is based on variants of Jeffrey-conditionalization and information distance minimization.