Reverse engineering of Web pages based on derivations and transformations

The final user interface of an information system could be reverse engineered according to a model-driven engineering perspective to multiple levels of abstraction, either independently of each other or simultaneously depending on the designer's goals: concrete user interface (which is independent of the user interface toolkit), abstract user interface (which is independent of any interaction modality), and "task and concepts" (which is independent of any particular implementation). To support the user interface reverse engineering from the code level to any model level, a set of derivation rules have been implemented in ReversiXML. To support the user interface reverse engineering from any model to any other level, graph grammars have been implemented in TransformiXML. A graph grammar consists of graph transformation rules, called productions, that accept as input a graph representation of the user interface to be reengineered, apply the transformation, and obtain a result that can be further exploited to re-create a new user interface.