Categorical Perspectives on Minimal Realizations of Bipolar Fuzzy Languages

This paper presents a categorical framework for the realization and minimization of bipolar fuzzy languages via deterministic fuzzy automata with bipolar fuzzy outputs. Bipolar fuzzy information allows the independent handling of positive and negative evidence, which is essential for modeling uncertainty in decision systems and computational settings. Using suitable notions of reachability and observability, we construct minimal realizations of bipolar fuzzy languages and establish a Myhill-Nerode type equivalence whose quotient yields a canonical minimal automaton. Functorial connections between the categories of bipolar fuzzy languages and deterministic fuzzy automata are developed, and the kernel and image of the observability map are analyzed to characterize state indistinguishability and language derivatives. These results extend classical minimal realization theory to the bipolar fuzzy setting.