Finite State Transducer Modification by Examples

Model Transformation By Example (MTBE) is a new branch of model driven software development. Transducers (automata with output) can be used to abstract model transformation. This form of abstraction makes possible to consider applications of grammatical inference algorithms. In this paper we investigate whether an effective inference procedure can be developed to derive a modified transducer from examples of desired input/output pairs instead of infering such a transducer from scratch. The paper starts with the description of our motivational example. Then we propose an algorithm to infer modifications of the transducers. Finally, we discuss metrics to evaluate the quality of such an inference.