Grammar Inference consist to infer a language from a set of words contain in this language, or not. In our particular case of regular grammar inference, the problem consist in fact to infer an automaton, due to the isomorphism between regular languages and finite state machine. Thanks to Vaucanson's power expression to automata's manipulation, it becomes possible to develop a library of regular grammar inference.

We will present Mical, a young library of regular grammar inference, whose aim is to make easier the writing of algorithms in this field. After a brief view of this library, we will interest first to present existing algorithms inside Mical (with and without negative sample). Different manipulations of the library will be exposed too.

Then we will explain by which reusable way the algorithms inside Mical are implemented, showing how Mical accomplish its aim : make it easier to implement regular grammar inference.

We will finish by a general consideration of Mical's actual situation, and of its future extensions.

• max0604_mical_slides.pdf