Information Technology Reference
In-Depth Information
Motivation for Relation-Aware and Cross-Workflow
Search
We propose to use network
S
0
and explore a relation-aware and cross-
workflow search technique. Before diving into details, let us take an
example to explain how
S
0
supports such a search. A typical question in
bioinformatics is that “Given a DNA sequence, can I first find similar
ones from
WU-BLAST
and then compare them with those from
ClustalW2
?” Note that
WU-BLAST
and
ClustalW2
are two popular
sequence alignment services. To answer this question, we can leverage
the relation-aware search to find candidate workflows that start
from some operation in
WU-BLAST
and end with some operation in
ClustalW2
. The question can be thus rewritten into the following query:
search workf lowwhereWU
-
BLAST
!
ClustalW
2
It is quite likely that in the repository there is not a single workflow
that contains an operation chain starting from
WU-BLAST
and ends at
ClustalW2
. In this case, the result should be a new workflow concatenat-
ing snippets from several existing ones. We use the largest cluster in
S
0
to
demonstrate this idea. This cluster, which is in the top-left of Figure 8.10,
is enlarged and shown in Figure 8.14 with the name of each node (i.e.,
operation) in it. Operations 116, 117, 168, and 169 belong to service
WU-
BLAST
; 128 and 130 belong to service
ClustalW2
. Two paths
<
169, 116,
117, 119, 128, 130
>
and
<
168, 116, 117, 119, 128, 130
>
are two
Figure 8.14
Obtain a service chain between two operations.
