Information Technology Reference
In-Depth Information
Table 9.1
Meta-actions
influence matrix
Meta-actions
a
b
c
M
1
-
b
1
c
2
ₒ
c
1
M
2
a
2
ₒ
a
1
b
2
-
M
3
a
1
ₒ
a
2
-
c
2
ₒ
c
1
M
4
-
b
1
c
1
ₒ
c
2
M
5
-
-
c
1
ₒ
c
2
M
6
a
1
ₒ
a
2
-
c
1
ₒ
c
2
Let us define
M
(
S
)
as a set of meta-actions associated with a decision system
S
.
Let
f
∈
F
,
x
∈
X
, and
M
ↂ
M
(
S
)
, then, applying the meta-actions in the set
M
on an object
x
will result in
M
, where object
x
is converted to object
y
by applying all meta-actions in
M
to
x
. Similarly,
M
(
f
(
x
))
=
f
(
y
)
, where
x
is
converted to the state of
y
by applying all meta-actions in
M
to
x
for all
f
(
F
(
x
))
=
F
(
y
)
∈
F
.
Also, by
F
.
It should be mentioned here that an expert knowledge concerning meta-actions
involves only classification features. Now, if some of these features are correlated
with the decision feature, then the change of their values will cascade to the decision
through the correlation. The goal of action rule discovery is to identify possibly all
such correlations. At the same time, the goal of meta-actions discovery is to identify
all the effects triggered by the application of such meta-actions.
(
Y
)
, where
Y
ↆ
X
, we mean
{
F
(
y
)
:
y
∈
Y
}
9.4.2 Mining Meta-actions Effects
The application of meta-actions on a set of objects triggers a set of effects within the
closed information system (newobjects can not be added). Those effects translate into
changes in some flexible features values. Meta-actions are applied by practitioners
without certainty on the effects for all objects. Each object might react differently to
each meta-action, thus practitioners shall explore the personalized effects of applying
meta-actions. In this section we present the methodology used to extract meta-actions
effects on a closed information system.
Commonly, an information system is represented in a tabular format, where fea-
tures represent the columns as attributes and rows represent the individual transac-
tions. Each transaction is uniquely identified by a transaction ID,
TID
. In the context
of meta-actions, objects might have one or more transactions and each object is iden-
tified by an Object ID,
OID
. In addition, mining meta-actions effects translates into
mining the objects state transitions. In other words, mining the objects' transitions
from an initial transaction, where meta-actions were applied, to another transaction
belonging to the same object. Meta-actions transitions are mined after the applica-
tion of meta-actions, and thus, it is important to keep track of the transactions' order.
Therefore, an additional feature representing the order of the sequence of transactions
is necessary. In addition, in a decision system, a particular feature called decision
Search WWH ::
Custom Search