Expert systems. (Czech) Zbl 0626.68072

In the paper rule-based expert systems are referred to in the context of knowledge engineering techniques. It is an introductory overview of mathematical background for various problems of knowledge processing. Foundations of inexact and nonmonotonic reasoning are discussed in more detail. The algebraic approach is emphasized for constructing inference engines that adequately propagate certainty or credibility factors (regardless they are interpreted logically or probabilistically) through nets of rules that represent domain-dependent knowledge. Since present rule-based knowledge bases ought to be built as loop-free nets, it is argued that nonmonotonic inferences are necessary to gain the ability to treat more complex nets, especially nets with cycles. Mutual influence between inference engines and representational schemes of knowledge is also discussed.


68T99 Artificial intelligence
68-01 Introductory exposition (textbooks, tutorial papers, etc.) pertaining to computer science
Full Text: EuDML