Information Technology Reference
In-Depth Information
lowing the execution of WPCs with semantically
untyped event interfaces, that is, interfaces that have
no mapping within the PDO. Moreover, the new
approach exhibits higher viscosity with respect to
a specific subcategory of changes, namely WPC
additions that require the introduction of new
concepts in the PDO. Still, we expect that once
the PDO has gained some maturity, these types of
changes will be rare. All other types of changes
had lower viscosity in the new approach.
interface descriptions understandable by humans
and machine agents. This trend of semantically
enriching component interfaces in order to facili-
tate their composition is now also being applied
to the grid distributed computing environments
(Pastore, 2007) and service-oriented architectures
(Narayanan & Mcllraith, 2002).
While many of the approaches referenced
previously include algorithms for the component
selection, this problem is outside the scope of our
project, since we rely on the user's ability to select
appropriate WPCs for a particular situation. It is
important to note that the components we deal
with in our research are mainly for the purpose
of information content delivery and data input. In
other words, they can primarily be characterised
as data components, as opposed to computational
components. As such, they comprise data-oriented
interfaces involving very simple protocols only.
Several research and commercial projects lever-
age component composition concepts to promote
end-user-initiated evolution of software systems.
Won and Cremers (2002) describe the concept of
interactive integrity checking in support end-user
tailoring of component compositions. Integrity
checks are used for ensuring the correctness of
compositions and also to provide feedback infor-
mation to end users on how to correct malformed
component compositions. In our approach, a check-
feedback-correction interaction is performed by the
user who first uses a portal page.
Many workflow editors (e.g., Edge Diagrammed,
2006; VisiQuest, 2007) have been developed to allow
non-expert users to develop end-to-end workflows
by enabling user selection and composition of
shared components. Congruent with our approach,
Kim, Spraragen, and Gil (2004) seek to overcome
problems associated with manual or user-directed
component/workflow compositions by defining
components in terms of a domain-specific ontology.
The ontology is utilised as a means for evaluating the
correctness of a component (or, more specifically,
a workflow) composition and implicitly to support
end-user comprehension of the workflows designed.
related work
Our work combines two separate, but related
areas of software engineering research, namely
component-based software composition
and
end-
user programming
.
Two key problems in the area of component-
based software are the definition of interfaces
for reusable components and selection of an ap-
propriate set of components needed to achieve the
goal of the composition. Techniques various power
and complexity have been developed to describe
interfaces of components. These approaches range
from abstract interface definition languages like
the Balboa Interface Definition Language (BIDL)
(Doucet, Shukla, Gupta, & Otsuka, 2002), over
XML-based component description languages
(Pahl & Ward, 2002) and logic-based specifications
(Lammermann & Tyugu, 2001) to automata-based
languages (de Alfaro & Henzinger, 2001) that also
capture temporal aspects of component interfaces.
While the goal of these techniques is to facilitate
the composition of components, they generally
require programming expertise and the knowl-
edge of technical details and, as such, are not
suitable for the end users. This issue started to be
addressed with the emergence of the Semantic Web
(W3C Semantic Web, 2001) paradigm. The use of
ontologies to describe component interfaces allowed
approaches (Fujii & Suda, 2004; Gomez, Han,
Toma, Sapkota, & Garcia-Crespo, 2006; Medja-
hed, Bouguettaya, & Elmagarmid, 2003) to make
Search WWH ::
Custom Search