In model-driven software engineering, model
transformation plays a key role for automatically generating
and updating models. Transformation rules define how
source model elements are to be transformed into target
model elements. However, defining transformation rules is
a complex task, especially in situations where semantic differences
or incompleteness allow for alternative interpretations
or where models change continuously before and
after transformation. This paper proposes constraint-driven
modeling where transformation is used to generate constraints
on the target model rather than the target model
itself. We evaluated the approach on three case studies
that address the above difficulties and other common transformation
issues. We also developed a proof-of-concept
implementation that demonstrates its feasibility. The implementation
suggests that constraint-driven transformation is
an efficient and scalable alternative and/or complement to
traditional transformation.
Forschungs-