Information Technology Reference
In-Depth Information
283 operations in 241 services. Nodes are operations in services, and a
directed edge represents a data link between two operations in some
workflow.
Based on
S
and
S
0
, we can define two levels of collaboration among
services. The collaboration relationship among services invoked in the
same workflow in
S
is
weak
. Compared to
S
,
S
0
contains the operation-
level information of both intra- and inter-workflow invocation
sequences. Thus, a
strong
collaboration relationship between two
services implies a direct operation invocation between them in some
workflow.
8.2.4 Summary of Findings
The network analysis of scientific workflows in the myExperiment
repository reveals the current usage pattern of services in scientific
workflows. It shows that the use of life science services is low in
myExperiment workflows, and only a couple of utility services are
frequently used in many workflows. Meanwhile, many services used in
myExperiment workflows function individually without collaborating
with each other. In summary, current service reuse in scientific work-
flows is unsatisfactory.
How can this knowledge be extracted to facilitate workflow reuse?
Our work demonstrates the effectiveness of constructing a workflow-
service network and its derived networks. The usage pattern embedded
in these networks provides quantitative answers to the following four
relations:
Workflow-Service
. How workflows use services.
Service-Workflow
. How services are used in different workflows.
Workflow-Workflow
. How multiple workflows use common
services.
Service-Service
. How services collaborate with one another.
Such knowledge embeds the best practice of using services in
workflows, and therefore is well suited to feed into a recommendation
system to facilitate services-oriented workflow reuse. In the next
section, we will discuss how to use the information, derived in this
section, to provide map and GPS-like assistance to service composition.
