Inconsistency-Tolerant Belief Revision for Distributed Decision Support
Hendrik Decker, Rubén de Juan‐Marín · 2013 Eighth International Conference on P2P, Parallel, Grid, Cloud and Internet Computing · 2013
Decision support is the main objective of business intelligence. The facts and rules of knowledge bases at the back end of decision support systems represent beliefs that are subject to revision. Belief revision is called for whenever an inconsistency occurs. Inconsistency may be due to an update that contradicts extant beliefs, or may have remained hidden before being uncovered. In particular, inconsistency is likely to appear in decision support systems based on distributed, mutually independent knowledge sources. If consistency is modeled by business rules, in the form of integrity constraints, then inconsistency can be measured, by sizing either the violated instances of constraints or the beliefs that cause constraint violations. Measures that quantify inconsistency enable a distinction of different degrees of integrity violations, and thus a relaxation of the intolerant all-or-nothing attitude of classical logic toward consistency. They also provide a logical foundation and justification of maintaining and reasoning with beliefs that possibly are inconsistent, without risking to lead to decisions that suffer too much from contradictory sources.