THE NON-DETERMISNISM PROBLEM AND FUZZY LOGIC

The notion of semantic non-determinism and its relation to the notion of fuzziness are considered in the paper. This notion is governed by the non-determinism of a domain and is formally modeled with N-matrices. These matrices include sets of truth-values, instead of a single representation. The choice of a truth-value from a certain set is random. Sematic determinism has outstanding advantages in modeling non-deterministic queries, faulty logic gates, etc. The notion of fuzziness, being fundamental for the theory of fuzzy sets, represents many-to-one relation, while the notion of semantic non-determinism - one-to-many. Fuzzy logic, which is based on fuzzy sets, handles linguistic variable "Truth" with a range of truth-values [0, 1]. Our supposition is that fuzziness - non-deterministic concept. Non-determinism in fuzziness is implicit, and we suggest alternative consideration of a grade of membership as an interval value. This consideration allows us to have a different look at the Sorites paradox. Main operations of the fuzzy logic are overridden in the terms of the intervals, and an effectiveness of the redefining is confirmed by the computer experiments.