The Research Of Optimization Strategy Of Non-Revision Method For Propositional Beliefs

Constantly updated information as well as differences in conflict with each other often exists in commonsense reasoning process. This type of uncertainty or incomplete information called the belief. Occurred in the process of reasoning in contradiction, then we must take certain methods to resolve conflicts, and this is the key areas of research points of the logical reasoning. Belief revision theory and belief non-revision theory is aimed at solving the inconsistency of the belief set of the two main methods of coordinate maintenance. Compared to belief revision theory, belief non-revision theory avoids losing some conclusions could have deduced before deleting any belief and therefore does not want to get to the conclusion and other issues in the process of revision.The non-revision method of AND-OR clause sets have been proposed in the propositional logic, which makes the idea of belief non-revision to further expand the application. This approach limits the grammatical form of the assumptions to a set of AND-OR sentences, while the ordinary generalized resolution method as the reasoning. Proved that the method for assumptions expansion have good properties, such as consistency, closeness, extensionality, etc. The ordinary generalized resolution method, however, in the process of it and without restrictions, this leads to the generalized deduction may exist in the process of many useless clause, make the process redundancy numerous long.Is aiming at the problems above, this paper, based on the thought of the non-revision by use of optimization strategy due the process of ordinary generalized resolution to improve non-revision method of AND-OR belief set. This paper selected two optimization strategies to optimize the above method. First, the generalized deletion strategy, delete those results of generalized resolution for a tautology AND-OR sentences or implied. The second, generalized lock resolution strategy, add lock to the literal of clause and stipulate the generalized resolution selection can only choose the AND-OR clause which contains minimum lock, in order to limit its resolution in general. This paper proves that the improved non-revision method of AND-OR clause sets through these two kinds of optimization strategy, by the method of income hypothesis expansion did not change, their properties are retained. In addition, the paper also gives the algorithm of the improved non-revision method of AND-OR belief set by the two optimization strategies. The experimental results of optimization by generalized deletion strategy show that the optimization strategy is effective.