Information Technology Reference
In-Depth Information
useful. Generalization is divided into two categories: constructive and selective.
If in generating concept description
D
, all descriptions used have been appeared
in initial concept description
i
∈
I), then it is called selective generalization
rule, otherwise constructive generalization rule.
1. Selective generalization rules
Assume CTX, CTX
1
, CTX
2
represent arbitrary expression.
D
i
(
(1) Condition elimination rule
CTX
∧
K
|
K
(7.9)
S
CTX
::
>
::
>
where S is a predicate or a logic expression.
(2) Add selective item rule
CTX
1
K
|
::
K
(7.10)
generalize concept description through adding selective item rule, such as:
CTX
∧
[color = red]
CTX
1
∨
CTX
2
::
>
>
|
CTX
∧
[
L
=
R
2
]
K
K
::
>
::
>
(3) Enlarge quote range rule
|
CTX
∧
[
L
=
R
1
]
CTX
∧
[color: red
∧
blue]
K
(7.11)
K
::
>
::
>
where R
1
¥
R
2
¥
DOM(
L
), DOM(L) is domain of
L
,
L
is a item,
R
i
is a set of
values
can have.
(4) Closed interval rule
L
L
∧
=
a
: :
>
K
CTX [
]
<
CTX
∧
[
L
=
a
ׂ
b
]::>
K
(7.12)
∧
L
=
b
: :
>
K
CTX [
]
where
L
is linear descriptor,
a
and
b
is specific value of
L.
(5) Climbing generalization tree rule
∧
L
=
a
: :
>
K
CTX [
]
∧
L
=
b: :
>
K
CTX [
]
<
CTX
∧
[
L
=
S
] ::>
K
(7.13)
@
∧
L
=
: :
i
>
K
CTX [
]
where
L
is structural descriptor, in the generalization tree domain of
L
,
S
represents the lowest parent node whose successors are
a,b,···,i
.
(6) Rule transforming constant into variable
F a
F b
[ ]
[ ]
< ∀
xF x
[ ]
(7.14)
@
F i
[]
