Information Technology Reference
In-Depth Information
ontologies and service interface descriptions) and explicitly gives the user
the possibility to adapt predefined domain models according to his expert
knowledge or specific requirements. What is more, it enables a very flexible
way of expressing additional knowledge that is cleanly separated from the
implementation of the synthesis algorithm.
Semantics-Based Service Composition Support (Requirement 4)
If service composition support is provided, in most systems it is restricted to
single steps of the workflow, that is, service discovery: the systems support
the identification of services that can provide or use the data that are used or
provided by services that have already been inserted into the workflow. This
semantically supported step-by-step workflow construction is actually rather
service discovery than real automatic workflow composition, and comes with
the risk that users get stuck when trying to construct the globally intended
solution stepwisely.
Truly comprehensive abstract workflow descriptions are only supported by
Bio-jETI (SLTL specifications), SADI/SHARE (arbitrarily complex SPARQL
queries) and the SeaLife Argumentation Interface (higher-level planning
goals). The final target specification used for automatic workflow creation
with GNU make, the start-to-end specification used by jORCA and the work-
flow templates that are applied within Wings are less expressive, but can still
be regarded as abstract workflow descriptions.
Workflow Validation and Verification (Requirement 5)
Verification and validation of workflow models is mostly realized on a purely
syntactic level (e.g. syntax or type checking on the component level), whereas
semantics-aware verification methods that consider the entire workflow model
(e.g. model checking) are provided only by very few systems. In Bio-jETI, vali-
dation and verification methodology is available via the LocalChecker and the
GEAR model checking plugins, respectively. They enable constraint-driven
workflow design by continuous monitoring of the workflow development pro-
cess in terms of the constraints that are defined within the domain model.
To the best of the author's knowledge, no other workflow system in bioinfor-
matics provides similarly integrated verification and validation facilities.
Regarding the workflow systems considered here, the application of model
checking techniques has only been described for the analysis of Discovery Net
workflows [75]. However, there the model checking is not performed directly
on the workflow model within the workflow modeling framework, but as a
separate step on a specific, derived representation. This means in particular
that the results of the analysis can not be adequately used to support the
workflow design phase, as they only provide information about the (finally)
exported workflow model. Moreover, the examples provided in [75] merely
show how general properties of the workflow models (such as reachability of
Search WWH ::
Custom Search