Dedicated Model Measurement
Ma et al. [Ma et al., 2004] compare different versions of the UML metamodel using OO metrics defined in [Bansiya and Davis, 2002]. Ma et al. [Ma et al., 2005] define patterns linked to the lifecycle of metaclasses, and study them on different versions of the UML metamodel. This work is similar in spirit to those made at the OO level [Mattsson and Bosch, 1999, Bansiya, 2000, Gî rba et al., 2005].
MDE Processes Measurement
Berenbach et al. [Berenbach and Borotto, 2006] list a number of metrics for model driven requirements development and enounce some good practices. The Modelware project delivered three documents [Modelware Project, 2006a, Modelware Project, 2006b, Modelware Project, 2006c] in which several metrics about MDE processes are defined.
UML Models Measurement
Previous works about the measurement of UML models follows the same decomposition as the UML artifacts themselves. Some authors address the measurement of class diagrams (see [Genero et al., 2005] for a survey), others the measurement of dynamic models [Genero et al., 2002, Baroni, 2005], component models (e.g., [Mahmood and Lai, 2005]), and OCL expressions [Cabot and Teniente, 2006, Reynoso et al., 2003].
These works are dedicated to particular MDE artifacts, i.e., metamodels, processes, UML models. They do not note that all this artifacts are models too, w.r.t. a metametamodel, a process metamodel or the UML metamodel. These contributions do not leverage this idea for defining a generic metric usable at any moment of product life-cycle, from requirements to implementation.