A Syntax-Directed Model Transformation Framework Based on Attribute Grammars

Abstract

Model transformation is a key aspect of model-driven software development because it enables the automatic derivation of different interpretations of a system model. In many scenarios (e.g., design of domain-specific languages), models usually have implicit identifiable primary tree-like syntactic structures, on which additional secondary relationships are imposed to yield the final model graphs. Therefore, in these scenarios it seems natural to address the processing of these models on the basis of their underlying syntactic structure. For this purpose, we have developed AGT, an experimental transformation framework based on attribute grammars, which takes full advantage of the underlying syntactic structure of source models. For models in which this structure is clearly identifiable, the approach could result more natural and easier to use and maintain than other more conventional model transformation approaches (e.g., those based on more standard model transformation languages).