Title :
Any-time knowledge revision
Author_Institution :
CRIL, Univ. d´Artois, Lens, France
Abstract :
We propose and experiment a practical multi-level approach to maintain contradiction-free knowledge when some incoming additional information that can contradict the preexisting knowledge must be taken into account. The approach implements an any-time strategy that triggers successive reasoning paradigms ranging from credulous to computationally more intensive forms of skepticism about conflicting information. It makes use of recent dramatic computational progress in constraint satisfaction techniques for finite domains and Boolean-related search and reasoning.
Keywords :
knowledge representation; Boolean-related search; any-time knowledge revision; artificial intelligence; constraint satisfaction techniques; contradiction-free knowledge; contradictory knowledge representation; finite domains; multilevel approach; successive reasoning paradigms; Abstracts; Cognition; Complexity theory; Computational modeling; Distance measurement; Lenses; Standards; Artificial Intelligence; Belief and Knowledge Revision; Credulous and Skeptical Reasoning; SAT;
Conference_Titel :
Information Reuse and Integration (IRI), 2014 IEEE 15th International Conference on
DOI :
10.1109/IRI.2014.7051908