Inductive consistency in knowledge-based decision support systems

摘要

Efforts to integrate knowledge into Decision Support Systems have lead to approaches that incorporate induction techniques to derive rules from data. Such integration efforts create a new integrity control challenge for Knowledge-Based Decision Support Systems. The challenge is recognizing and coping with change. In this paper, we analyze the inductive consistency relationship that exists between data and inductively derived rules and present a strategy for recognizing violations of the inductive consistency relationship. The recognition of integrity problems provides a first defense toward refining knowledge in a Knowledge-Based Decision Support System.