Information Technology Reference
In-Depth Information
State of the art. As mentioned above, the main research line related to this
challenge corresponds to the transformation of i models into UML-like artefacts
[ 50] , generally in the context of model-driven development (MDD) [ 60] . Within the
OO-Method MDD methodology [ 49] , Alencar at al. have shown that it is possi-
ble to partially infer data conceptual models from i models [ 2] . Actors and their
relationships, and resources (both dependencies and internal SR elements), play a
fundamental role in this translation. Following the MDD foundations, transforma-
tion rules are defined to obtain an initial class diagram that is completed manually
(e.g., adding information about multiplicity, not present at i models) for obtain-
ing a complete OO-Method class model which can be used in the rest of the MDD
process.
Concerning use case generation, Estrada et al. [ 17] propose a method that cov-
ers identification of use cases and actors, and writing of scenarios. Use cases are
determined from both the task and resource dependencies that involve the actor that
represents the system. The actors at the other end of such dependencies are rep-
resented as use case actors. Finally, SR diagrams are used to fill some predefined
templates in order to generate the text of scenarios. A similar approach is followed
by Santander and Castro [ 59] .
Apart from these kind of models above, i has been used in other contexts.
Remarkably, Ncube et al. [ 47] report an extension to the RESCUE process [ 38]
in which a collection of 30 patterns were applied over an i model to generate tex-
tual candidate requirement statements using the VOLERE template, generating up
to almost 600 requirements. As a result of this work, the authors argued that require-
ments generated from i models resulted in a more complete overall requirements
specification.
Lucena et al. [42] have gone one step beyond in the development process and
they address the generation of architectural models. They combine two levels of
refinement, first by modularizing the departing model using the rules described in
[ 43] and then transforming the resulting i model into a software architecture model
described with the ACME architectural description language [ 25] .
Research directions. As shown above, the transformation of i models into other
models has been subject of much investigation. However, being a very complex
topic, it requires still much work to do. When considering i models as the starting
point of an MDD process, research is needed with respect to several topics [ 8] :
Automating as much as possible the model transformation. It seems clear from
previous work that full automation is not feasible since the underlying ontologies
cannot be completely aligned. However, the work undertook so far (see above)
looks promising and it may be expected that more results will be achieved soon.
An important result of these approaches should be the clear statement of the lim-
itations of the proposed methods regarding to automation. Also the possibility
of enriching the i framework with information that in fact belongs to the target
ontology (e.g., order of task in task decompositions [23] ) is a point to explore.
Validating the adequacy of the i model before applying the transformations.
Since the i model is not originally conceived for later transformation, it is
Search WWH ::




Custom Search