Minimal Model Method For Belief Non-Revision

Monotonic reasoning can’t satisfy people’s cognitive to the objective world, so non-monotonic reasoning is proposed. The obtained beliefs are not always true, when the new belief conflicted with beliefs in the original belief set, the new belief set would appear inconsistent, how to maintain the coordination of the new belief set is a key question of the common sense reasoning. Currently, the belief revision method is the main way to deal with inconsistent beliefs, but the shortcomings of it is that: in the process of belief revision, the belief revision method would lost some useful beliefs and produce multiple conclusions which are difficult to choose a fitting one.The minimal model method for belief non-revision is used to handle inconsistent beliefs in this thesis, its basic idea is that:some inconsistent beliefs are allowed in a belief set, the conclusion inferred from the belief set which don’t has the inconsistent beliefs is considered as the final conclusion, and the final conclusion is the most concise. The minimal model method for belief non-revision could avoid the phenomenon of losing some useful beliefs and producing multiple conclusions which are difficult to choose a fitting one. Meanwhile, it can also solve the problem that an inference engine is unable to reasoning under meeting some inconsistent beliefs.Firstly, the minimal model of belief set and the minimal entailment of belief set are defined, as well as the properties of the minimal model of belief set are discussed. Based on the researches on the preferred model and the preferred entailment, the definition of the minimal model of belief set and the minimal entailment of belief set are proposed; the properties of the minimal model of belief set are analyzed, and some good mathematical properties such as reflexive property, right weaken property, cut property are expected.Secondly, the extension set of belief set is defined and its properties are discussed. By analyzing the minimal model of belief set and the minimal entailment of belief set, the extension set of belief set is not solved by the reduced method for belief non-revision, but the minimal model method for belief non-revision; the properties of extension set of a belief set are analyzed, and some good mathematical properties such as consistent property, deductive closure property and cumulative property are expected.Finally, the belief non-revision reasoning system based on minimal model of belief set is designed and implemented. The obtaining algorithm of the minimal model of belief set is designed by minimal model of belief set and minimal entailment of belief set, and the belief non-revision reasoning system based on minimal model of belief set is designed on the basic of it, some examples in real life are used to access the performance of the system. The experiment result shows that: the belief non-revision reasoning system based on minimal model of belief set not only would not lost some beliefs and not produced multiple conclusions which are difficult to choose, but also the obtained reasoning conclusions are the most concise.