Global Positioning System Reference
In-Depth Information
breaking the Model Web principle on lowering entry barriers. For example,
when putting services together we need easy-to-use tools to allow us to
assemble these WPS-based services together and to understand the new
context led by the service composition (see Section “Not One but Multiple
Infrastructures”).
Not One but Multiple Service Interfaces
It seems unrealistic to assume that every single environmental model fi ts into
WSDL and WPS service description. Models by nature are heterogeneous and
are often tightly tied to the singularities of each discipline (e.g., biodiversity
models, watershed models, ecological models), which by large exceed the
capabilities of these service description services. For example, several authors
have found several limitations in matching data structure between models
and service description specifi cations because of the way inputs and output
parameters are declared. Inputs parameters in models are much richer than
those supported by WSDL and WPS specifi cations. Although the modeling
community requires the development of standards for describing and
publishing environmental models in suitable form for automated access and
integration (Laniak et al. 2013b), in practical terms, however, no silver bullet
exists and the description of models are being proposed to vertebrate sharing
and integration of models within each discipline. These proposals range
from describing biomedical (de Bono et al. 2011) and biodiversity models
(Endresen and Knüpffe 2012; Constable et al. 2010) to ecology (Schmolke
et al. 2010) and climate modeling (Parsons 2011).
The wide diversity of descriptions of models raises the question
whether such models can interact across disciplines. Some recent works
have addressed this limitation. One approach relies on the brokering
pattern (Buschmann et al. 1996). A brokering component would implement
the required business logic to allow clients for discovering and accessing
heterogonous models from distinct disciplines. It extends the traditional
SOA architecture (Papazoglou and van den Heuvel 2007) by introducing an
“intelligent” component to consume heterogeneous models in a transparent
way and interact with them using a single and well-known endpoint, i.e.,
the brokering component (Nativi et al. 2011). In addition, if brokering
components are well designed to turn into reusable building blocks, various
brokers can be aggregated and work cooperatively to deliver meaningfully
functionality and connect increasingly more models and sources across
disciplines (Vaccari et al. 2012; Díaz et al. 2012).
An alternative to the brokering approach is to design a common
interface that abstracts from service description specifi cations and other
specifi c APIs to access and interact with backend models in a uniform
manner. To this respect, Granell et al. (2013a) recently discussed on design
Search WWH ::
Custom Search