Information Technology Reference
In-Depth Information
Undirected graph
Abs. workflows
Workflows
Q: workflow-service relation
wf-1
wf-1
wf-1
wf-2
A
A
W: workflow-workflow relation
S: service-service relation
A
CD
B
C
C
Abstract
Combine
B
B
wf-2
wf-2
A
C
C
C
S': operation-operation relation
…
…
B
D
D
D
Directed graph
WSDL Services
Other blocks
Figure 8.5
The network analysis of the myExperiment workflows.
removing non-Web (i.e., WSDL based) services, such as local bean-
shells and XML manipulating blocks, while maintaining the data flows
between services. Afterwards, these abstract workflows are combined
into two disjoint networks (graphs): an undirected
workflow-service
network
and a directed
operation network
. In the former, nodes are
either workflows or services and edges represent the inclusive relations
between them, that is, a workflow
i
is connected to a service
j
if
i
consists of
j
. In the latter, nodes are operations in services, and a directed
edge represents a data link between two operations in some workflow.
More details regarding the myExperiment workflow set, how networks
are built and analyzed can be found in [222].
From the workflow-service network, we derive two additional
networks: a
workflow-workflow network
in which two workflows
are connected if they comprise services in common and a
service-
service network
in which two services are connected if they appear in
some workflow together. We use Pajek [223], a widely used social
network analysis tool, to produce all the three graphs. Table 8.1
summarizes the metrics of the myExperiment dataset and we will
further explain them later.
8.2.2 Undirected Network
We parse the myExperiment workflows to create the workflow-service
relation
Q
, formalized as an
m
n
matrix, where
m
is the number of
