2006 Volume 6 Issue 3 Pages 41-46
In this paper, we present a foundation of formulating human reasoning processes with Kansei representationbased on granularity generated by background knowledge. For the objective, we examine the role of background knowledge and granularity in several kinds of reasoning. We put emphasis on difference between objective and subjective levels of knowledge. We represent subjective knowledge by lower approximation in rough set theory under granularity generated by background knowledge.Then we examine monotonicity and non-monotonicity between the two levels with respect to deduction and non-monotonic reasoning. Next we apply the idea to abduction which gives us which premise is plausible when we are given rules and conclusions. The problem in abduction is how to select one from several possible premises and we solve it by calculating inclusion degree between possible premises and the lower approximation of the given conclusion. Thus we can introduce some ordering into the set of possible premises. Finally we apply a similar idea to conflict resolution typically dealt with in reasoning in expert systems and robotcontrol. When we have plural monotonic rules which can be applied to a given premise, if conclusions are associated with their own action or execution, then we cannot take the conjunction of all conclusions. Again we must select one of the conclusions using the inclusion degree between the premise and the lower approximations of possible conclusions. Thus we can introduce some ordering into the set of possible conclusions associated with actions.
