Finite Eilenberg Machines

Eilenberg machines define a general computational model. They are well suited to the simulation of problems specified using finite state formalisms such as formal languages and automata theory. This paper introduces a subclass of them called finite Eilenberg machines. We give a formal description of complete and efficient algorithms which permit the simulation of such machines. We show that our finiteness condition ensures a correct behavior of the simulation. Interpretations of this condition are studied for the cases of non-deterministic finite automata (NFA) and transducers, leading to applications to computational linguistics. The given implementation provides a generic simulation procedure for any problem encoded as a composition of finite Eilenberg machines.