Introduction - User-Level Workflow Design: A Bioinformatics Perspective - page 12

In total, the proclaimed semantic framework constitutes a kind of search
engine specialized on services and workflows: Like web search engines such
as Google Search [9] help to find single resources in the web, the semantic
framework helps to discover individual resources in the domain model. The
distinguishing characteristic of the semantic framework is, however, that it is
designed for finding entire resource combinations based on the provided do-
main model. This is indeed a crucial feature with regard to user-level workflow
design, as it supports the automatic identification of service and data com-
binations that are useful in a particular workflow scenario. The search space
for Google is constituted by the vast amounts of content in the World Wide
Web, which the search engine manages by indexing and ranking [54]. In con-
trast, the domain models for workflow applications are as such comparatively
small, but the search space that is constituted by the possible resource com-
binations can easily grow extremely large, which makes the automatic search
for workflows inherently complex. And clearly, the obtained results can only
be as good as what is provided by the domain model. However, also Google's
search engine has not changed the world because it added content - which it
did not - but because it has made content easily available to everybody. Sim-
ilarly, the workflow synthesis method itself does not introduce new resources,
but aims at making the workflow potential that is inherent in existing ser-
vice libraries easily accessible to anyone who needs to design workflows, but
especially to non-technical users like biologists and other natural scientists,
business process experts and project planners.
Semantic Handling of Services: Discovery
Finding available services in a network is a technical challenge that has lead
to the development of a number of service discovery techniques, such as the
UDDI (Universal Description, Discovery and Integration) registry for web ser-
vices [20] or the WS-Discovery (Web Services Dynamic Discovery) specifica-
tion [240]. Finding appropriate services within the set of available services is
a challenge which goes beyond this standard discovery of services and service
interfaces and which has to take into account the semantics of the individual
services.
In fact, effective workflow modeling depends on the knowledge of appro-
priate services for the involved steps. However, it is impossible for a human
to overview the plethora of available services manually: The BioCatalogue
[112], for instance, which has become the central registry for life science web
services, contains about 2300 services from around 160 providers at the time
of this writing, and is continuously growing. Given such service collections,
mechanisms are needed that do not only detect the principally available ser-
vices, but also help to identify those that are useful in the particular set-
ting. Accordingly, workflow management systems with an intended user group
of semi-computer-savvy application experts should provide mechanisms that