Conservative and Creative Strategies for the Refinement of Scoring Rules

摘要

In knowledge engineering research the refinement of manually developed intelligent systems is still one of the key issues. Since scoring rules are an intuitive and easy to implement knowledge representation they are suitable for the manual development of knowledge bases. In this paper, we present a conservative and a creative strategy for the refinement of scoring rule bases adapting existing rule weights and inducing new rules, respectively. The usefulness of the approach is demonstrated by a case study with a reasonably sized rule base from the biological domain.