On fuzzy recognizers

In this paper, we define and examine the concept of a fuzzy recognizer. If L(M) is the language recognized by an incomplete fuzzy recognizer M, we show that there is a completion M of M such that L(M) = L(M). We also show that if A is a recognizable set of words, then there is a complete accessible fuzzy recognizer M_A such that L(M_A) = A. We lay groundwork to determine rational decompositions of recognizable sets.