Information Technology Reference
In-Depth Information
where
t
B
is the time of occurrence of phenomenon
Y
and
t
const
is a time window
after
t
B
given as the one of the parameters.
Here, these values derive for comparisons. In this case, the nuisance variable
B
,
N
(
)
time
,
t
B
and
t
c
onst
are fixed. Therefore, we can simplify above two formulas:
time
n
A
(
time
)
p
(
A
)=
time
N
A
(
)
∝
n
A
(
time
)
,
(8)
time
time
t
const
p
(
A
,
B
)
p
(
A
|
B
)=
∝
p
(
A
,
B
)
∝
n
A
(
time
)
.
(9)
p
(
B
)
t
B
on the time axis and the
number of times for query of a certain phenomenon
X
. A slash part is a value of
p
Fig. 3 shows the value of
p
(
A
)
and
p
(
A
|
B
)
∝
p
(
A
,
B
)
.
These simplifications have a fatal fault. The value of anteroposterior correlation
and co-occurrence are same. We cannot extract the actual anteroposterior correla-
tion. Therefore, we compare and extract anteroposterior correlations by calculating
the degree of dependence as shown in the section 4.2.
By calculating the degree of dependence, we extract appropriate anteroposterior
correlations.
We show a simple example, as shown in Fig. 4. We relatively compare the case
of
A
1
,
A
2
,
A
3
,and
A
4
.
We cannot calculate the degree of dependence each item
A
1
,
A
2
,
A
3
,and
A
4
,
because their values are not normalized. Without normalization, the item which is
not meant will become a big value. For example,
A
3
is a bigger value without nor-
(
A
)
. A colored part is a value of
p
(
A
|
B
)
∝
p
(
A
,
B
)
Phenomenon
A
: The number of
occurrence times of a query
X
p
(
A,B
)
p
(
A
)
time
t
B
(phenomenon
Y
occurs)
Fig. 3
The graph for time axis and the number of occurrence times of a query
X
. A slash part
is a value of
p
(
A
)
. A colored part is a value of
p
(
A
,
B
)
.
Search WWH ::
Custom Search