Linking Goal-Oriented Requirements and Model-Driven Development - Intentional Perspectives on Information Systems Engineering - page 254

2.2 The OO-Method MDD Approach Overview

OO-Method is an MDD approach that separates the application and business logic
from the platform technology, allowing the automatic code generation from the con-
ceptual representation of the software systems [ 34] . The OO-Method production
process (Fig. 1) is comprised of three models: the Conceptual Model ,the Execution
Model , and the Implementation Model . The OO-Method Conceptual Model cap-
tures the static and dynamic properties of the system in a Class Model ,a Dynamic
Model , and a Functional Model . The conceptual model also allows the specification
of the user interfaces in an abstract way through the Presentation Model . These four
models represent the different views of the whole conceptual model, which has all
the details needed for the generation of the corresponding software application. The
complete definition of the OO-Method Conceptual Model is presented in [ 35] .
The class model is the core of the OO-Method conceptual model; the rest of
the models involved are defined starting from elements of the class model. For this
reason, the OO-Method class model has been chosen to explain the linking approach
presented in this chapter.

Conceptual Model

Implementation Model

Model Compiler Class

Model

Functional
Model

Persistence Tier (SQL Server, ORACLE, DB2, MySQL)
Business Logic Tier (EJB, COM, .NET)
Graphical User Interface Tier (JSP, ASP .NET, VB, .NET)
Model to Code
Transformation

Execution
Model

Dynamic
Model

Presentation
Model

Model - to-Model
Transformation

Fig. 1 The OO-method software production process

3 Linking Goal-Oriented and MDD Approaches

Our proposal for linking GORE modeling and MDD starts from the idea that
there are two kinds of models that must be coordinated to represent specific
parts of the development process: GORE models, and MDD models, which rep-
resent the intended systems at the conceptual level. These models are represented
by using modeling languages whose abstract syntax is specified by means of
metamodels.
For the coordination of these two models, we assume that it is possible to partially
infer an initial MDD model from both the information that is represented in the
GORE model and from extra information that is added when necessary. This MDD
model generation is possible if constructs of the MDD modeling language can be
inferred from constructs of the GORE modeling language. The constructs involved
are represented by the metaclasses of the corresponding metamodels.
It is important to note that we are referring to an initial MDD model and not a
complete MDD model because there are aspects related to specific system function-
ality that cannot be obtained from requirement models. Therefore, these functional