Efficient algorithms for semantic unification

The problem of matching vague terms must be addressed in knowledge-based systems involving uncertainty. Fril is a logic programming language extended by an integrated set of features for handling uncertain knowledge, based on the theoretical foundations provided by Baldwin’s mass assignment theory. This combines probabilistic and fuzzy uncertainty into a single framework. One of the fundamental operations in Fril is an extension to unification in which both terms are fuzzy and execution requires the calculation of a support for the conditional probability of the match. This is known as semantic unification. We outline algorithms for the calculation of interval and point value supports in the discrete and continuous cases.