Database Reference
In-Depth Information
at the global site, and results are received from the global site. The global site maintains information
about each local fuzzy database structure, such as fuzzy schema definitions and which kind of fuzzy
relations are stored in which local fuzzy database. This allows the global site to efficiently schedule a
global fuzzy query processing plan. Each local fuzzy database can accommodate any number of fuzzy
relations and optimally process each local fuzzy query given by the global site.
The Fuzzy Query Syntax
Well defined semantics of FTS-relational operations may not be directly suitable for fuzzy query formu-
lation; hence we have designed FTS-SQL (the fuzzy version of TS-SQL (Lim, Chiang and Cao, 1999)).
Demonstration of syntax of a simple FTS-SQL fuzzy query may be given as follows:
SELECT < target attributes >[anyDB]/[sameDB]
WITH Source Context {optional}
FROM < FTS relation >
WHERE < selection/join conditions > [anyDB]/[sameDB]
HAVING < value of α,β >
It may be noted that the keyword
anyDB
or
sameDB
are optional for SELECT clause and WHERE
clause and at the same time, a simple FTS-SQL fuzzy query has an optional WITH clause and HAVING
clause which makes it different from normal SQL. For both clauses, default keyword used is anyDB. If
WITH clause is specified in a FTS-SQL fuzzy query, the fuzzy query result will also include the source
context allowing the user to interpret the fuzzy query tuples using source context. The HAVING clause
specifies the value of α,β
∧
[0,1]where α is used for α-resemblance (Rundensteiner, Hawkes and Bandler,
1989) of fuzzy attribute values and β is used as a threshold of the membership grade of fuzzy tuple to
the fuzzy relation to satisfy the selection criteria. A FTS-SQL fuzzy query can be written as:
SELECT R.A, S.B [sameDB]
FROM R,S
WHERE R.X EQ S.X and R.U EQ `abc' and S.V EQ HIGH [anyDB]
HAVING 0.8, 0.6
This fuzzy query can be translated into an equivalent FTS-relational expression as shown below:
(
)
fs
fs
fs
fs
sameDB
anyDB
(
R
)
(
S
)
DD
σ
π
σ
R A S B
.
,
.
,. .
8 6
RU EQ abc
.
'
' . ,.
8 6
RU EQ S X
.
.
,. ,
8
.
6
S V EQ High
.
,. .
8 6
This translation is allowed because both
D
fs
sameDB
α β
and
D
fs
anyDB
α β
,
are commutative and associative,
however, source predicate must be evaluated before any FTS join is carried out, since attributes of op-
erand relations are merged during the join. A FTS-SQL fuzzy query involving union, intersection or
subtraction of two or more FTS-relation, can be written as given below:
,
SELECT < target attributes >[anyDB]/[sameDB]
WITH SOURCE CONTEXT {optional}
