Partial satisfiability-based merging

Abstract

When information comes from different sources inconsistent beliefs may appear. To handle inconsistency, several model-based belief merging operators have been proposed. Starting from different belief bases which might conflict, these operators return a unique consistent base which represents the beliefs of the group. The operators, parameterized by a distance between interpretations and aggregation function, usually only take into account consistent bases, consequently some information which is not responsible for conflicts may be ignored. An alternative way of merging uses the notion of Partial Satisfiability to define PS-Merge, a model-based merging operator that produces similar results to other merging approaches, but while other approaches require many merging operators in order to achieve satisfactory results for different scenarios PS-Merge obtains similar results for all these different scenarios with a unique operator. This paper analyzes some of the properties satisfied by PS-Merge. © Springer-Verlag Berlin Heidelberg 2007.