Database Reference
In-Depth Information
Defi nition 5. (Activity information cohesion).
For a valid activity
t
on an operation
structure (
D
structure ( ,
O
), its information cohesion µ(
t
) is defi ned as follows:
structure (
D
The activity information cohesion focuses on all information elements that are used
either as input or output by any operation within the respective activity. It determines how
many information elements are used more than once in proportion to all the information
elements used, which is a relative measure between 0 and 1.
The total cohesion of an activity is now given as the product of both the relation and
information cohesion. An activity has to score high on both cohesion metrics to say it is
cohesive in total. Clearly, an extreme score on one coeffi cient may outweigh a mediocre
result on the other, which is seen as satisfactory.
Defi nition 6. (Activity cohesion).
For a valid activity
t
on an operation structure (
on an operation structure (
D
on an operation structure ( ,
D
t
on an operation structure (
O
), its (general) cohesion
c
(
t
) is defi ned as follows:
c
(
t
) = λ(
t
)⋅µ(
t
)
In the following subsection, we will incorporate the cohesion metric in an overall
strategy for its application and explore two examples where we apply the introduced cohe-
sion metrics.
Application of the Cohesion Metric
Let us assume that there is an activity X, which is relatively incohesive in an overall
workfl ow design on the basis of the presented cohesion metric. It is subsequently considered
to be split up into validly ordered activities A and B. An evaluation that could take place on
the basis of the same activity cohesion is then as follows:
1.
Determine the cohesion of A and B (the cohesion of X is already known).
2.
If both cohesion coeffi cients of A and B are higher than that of X, then the division
into A and B is preferable.
3.
If the cohesion coeffi cient of X is higher than both cohesion coeffi cients of A and B,
then the larger activity X is preferable.
4.
In all other cases, the heuristic is indecisive.
Note that our heuristic
does not describe how the candidates A and B can be determined
.
does not describe how the candidates A and B can be determined
Obviously, in small enough cases it is feasible to generate a great number of partitions, but
it grows exponentially in the number of operations that are considered. Also note that a
similar approach could be taken when activity X is considered to be integrated with another
activity Y, resulting in activity C.
Consider the example in Figure 1. We totally abstract at this place from the content
of the information elements and operations within the operations structure. The example is
