Network Analysis and Reuse of Scientific Workflows - Business and Scientific Workflows - page 204

In the social network analysis, a component is a set of network nodes
connected by some relationship such that they are strongly connected.
This largest component in relation S (which consists of 241 services)
has 113 services and therefore covers 47% of all services. This indicates
that many services in the myExperiment function individually rather
than work together to form a linked research community.
In S , it is straightforward to see that a service with high degree
centrality collaborates with more peer services in all workflows. The
average degree centrality of a service in S is 4.4, which indicates that on
average a service collaborates with 4.4 peers in myExperiment. Mean-
while, a service with high betweenness centrality as highlighted in
Figure 8.9 indicates that service collaborations are more likely to go
through it—that is, it has more control over the information flow
between services.
8.2.3 Directed Graph
To study finer-grained collaboration among services in our dataset, we
zoom in on S to the operation level. We derive a directed relation S 0 ,as
shown in Figure 8.10, by examining the invocation relations among
Figure 8.10 Visualization of operations invocation network S 0 . Operations are grouped
into weakly connected components called clusters.