Résumé
Model transformation by example (MTBE) aims at defining a model transformation according to a set of examples of this transformation. Examples are given in the form of pairs, each having an input model and its corresponding output transformed model, with the transformation traces. The transformation rules are then automatically extracted from the examples. In this paper, we propose a two-step approach to generate the transformation rules. In a first step, transformation patterns are learned from the examples through a classification of the model elements of the examples, and a classification of the transformation links using Formal Concept Analysis. In a second step, those transformation patterns are analysed in order to select the more pertinent ones and to transform them into operational transformation rules written for the Jess rule engine. The generated rules are then executed on examples to evaluate their relevance through classical precision/recall measures.