Combining knowledge with many-valued logics

摘要

During the process of knowledge acquisition from different experts it is usual that contradictions occur. Therefore strategies are needed for dealing with divergent statements and conflicts. We provide a formal framework to represent, process and combine distributed knowledge. The representation formalism is many-valued logic, which is a widely accepted method for expressing uncertainty, vagueness, contradictions and lack of information. Combining knowledge as proposed here makes use of the bilattice approach, which turns out to be very flexible and suggestive in the context of combining divergent information. We give some guidelines for choosing truth value spaces, assigning truth values and defining global operators to encode integration strategies.