Handling Inconsistency Efficiently in the Incremental Construction of Statified Belief Bases

  • Authors:
  • Éric Grégoire

  • Affiliations:
  • -

  • Venue:
  • ECSQARU '95 Proceedings of the European Conference on Symbolic and Quantitative Approaches to Reasoning and Uncertainty
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

Most approaches to inconsistency handling in stratified propositional belief bases (KB) exhibit bad worst case computational complexity results, seemingly placing large real-life applications out of reach. One natural way to overcome this drawback consists in adopting a trade-off between the solution quality and the actual computational resources that are spent. Taking Benferhat et al.'s inclusion and lexicographic orderings as a case study, a complete revision algorithm is provided in the context of a single kernel assumption. It is shown to be a good approximation in the general case. Based on powerful heuristics about local search, this technique appears efficient for very large KBs, as illustrated by our experimental results.