On The Use Of Epistemic Ordering Functions As Decision Criteria For Automated And Assisted Belief Revision In SNePs

In this thesis I implement belief revision in SNePS based on a user-supplied epistemic ordering of propositions. I provide a decision procedure that performs revision completely automatically when given a well preorder. I also provide a decision procedure that, when given a total preorder, performs revision with a minimal number of queries to the user when multiple propositions within a minimally-inconsistent set are minimally epistemically entrenched. These procedures are implemented in SNePS as options alongside the old belief revision subsystem, wherein revision must be done entirely by hand. I implement both prioritized and nonprioritized belief revision by adjusting the epistemic ordering function passed to the new procedures. The first procedure uses O(|Sigma|) units of space, and completes within O(|Sigma|² · smax) units of time, where Sigma is the set of distinct minimally-inconsistent sets, and smax is the number of propositions in the largest minimally-inconsistent set. The second procedure uses O(|Sigma|² · s2max ) space and O(|Sigma|² · s2max ) time. The examples provided herein demonstrate that the new procedures generalize previous work on belief revision in SNePS.