论文信息 - Conflicts between Relevance-Sensitive and Iterated Belief Revision

Conflicts between Relevance-Sensitive and Iterated Belief Revision

The original AGM paradigm focuses only on one-step belief revision and leaves open the problem of revising a belief state with whole sequences of evidence. Darwiche and Pearl later addressed this problem by introducing extra (intuitive) postulates as a supplement to the AGM ones. A second shortcoming of the AGM paradigm, seemingly unrelated to iterated revision, is that it is too liberal in its treatment of the notion of relevance. Once again this problem was addressed with the introduction of an extra (also very intuitive) postulate by Parikh. The main result of this paper is that Parikh postulate for relevance-sensitive belief revision is inconsistent with each of the Darwiche and Pearl postulates for iterated belief revision.

[1] Adam J. Grove,et al. Two modellings for theory change , 1988, J. Philos. Log..

[2] Hirofumi Katsuno,et al. Propositional Knowledge Base Revision and Minimal Change , 1991, Artif. Intell..

[3] Norman Y. Foo,et al. Measuring similarity in belief revision , 2000, J. Log. Comput..

[4] Judea Pearl,et al. On the Logic of Iterated Belief Revision , 1994, Artif. Intell..

[5] Peter Gärdenfors,et al. On the logic of theory change: Partial meet contraction and revision functions , 1985, Journal of Symbolic Logic.

[6] Marianne Winslett,et al. Reasoning about Action Using a Possible Models Approach , 1988, AAAI.

[7] Maurice Pagnucco,et al. Dynamic belief revision operators , 2003, Artif. Intell..

[8] Norman Y. Foo,et al. Distance Semantics for Relevance-Sensitive Belief Revision , 2004, KR.

[9] Peter Gärdenfors,et al. Knowledge in Flux , 1988 .

[10] Michael Thielscher,et al. Iterated Belief Revision, Revised , 2005, IJCAI.