Towards a Principle-based Framework for Repair Selection in Inconsistent Knowledge Bases

This paper investigates a general principle-based framework for retrieving preferred repairs from inconsistent knowledge bases under a broad family of strategies. To begin with, we define a set of principles that ensure rational behaviours of repair selection strategies. Then, we classify the strategies into two basic categories: (i) comparing repairs without requiring formula information; and (ii) comparing repairs based on formula information. Based on this classification, we present several novel repair selection strategies and show that our framework encompasses various existing popular strategies. Through a systematical analysis of these selection strategies using the proposed principles, we conclude that our principles allow for effective discrimination among the strategies. Finally, preliminary experimental results are presented to show the feasibility of our approach.