Database Reference
In-Depth Information
Atomic condition. The basic building block of a query condition is the
atomic condition
. Basically, it
contains a name of an attribute and a constraint imposed on the value of this attribute. Such a constraint
may be a traditional, crisp one as, e.g., in
Price <= 200,000
It may also employ one of linguistic terms as, e.g.:
1.
Price =
low
(numeric fuzzy value)
2.
Land area =
very large
(numeric fuzzy value + modifier)
3.
Price
is not much greater than
250,000
(fuzzy relation)
4.
Location belongs to
favorite regions
(fuzzy set constant)
5.
Life quality indicators
are
compatible
with
high quality of life pattern
(multi-valued-attribute + fuzzy set constant)
Numeric fuzzy values
are to be used in connection with numeric fields as, e.g., with the field
Price
.
Meaning of such a linguistic term is intuitively obvious, although rather subjective. Thus, it should be
possible for each user to define his or her meaning of the linguistic term
low
. On the other hand, it would
be advantageous to make it possible to use an already once defined term, like
low
, in various fields.
Numeric fuzzy values may be accompanied by
modifiers
as, e.g.,
very
, that directly correspond to the
similar structures of the natural language.
Fuzzy relations
make it possible to soften rigidness of crisp relations. In the third example given
above, the atomic condition employs the
much greater than
fuzzy relation accompanied with the nega-
tion operator treated as a modifier. Thus, such a condition will accept the price of, e.g., 255,000, which
seems to be much more practical than treating 250,000 as a sharp limit.
The examples discussed so far employed linguistic terms to be used along with the numeric data. The
fourth example introduces a
fuzzy set constant
which is similar to numeric fuzzy values but meant to be
used with scalar data. In this example, the
favorite regions
constant represents the user's preferences as
to the location of the house sought. The concept of favorite regions will quite often turn out to be fuzzy,
i.e., some regions will be perceived by the user as the best location, some will be completely rejected,
and the rest will be acceptable to a degree. Obviously, such a concept is highly subjective.
Finally, the fifth example presents the concept of a
multi-valued attribute
but we will not discuss this
case here, for simplicity, referring the reader to (Zadrożny & Kacprzyk, 1996).
The
sequence of atomic conditions
is just a conjunction of atomic conditions. Due to the fact that
particular atomic conditions may be satisfied to a degree, we need to employ some generalization of
the classical AND logical connective, notably using a
t
-norm, in particular the min operator. In order
to achieve a flexibility of the aggregation postulated earlier, linguistic quantifiers are implemented in
FQUERY for Access.
Each sequence of atomic conditions may be additionally assigned an importance coefficient. That
way, the user may vary the degree to which given sequence contributes to the overall satisfaction degree
of the whole query.
Finally, the sequence of atomic conditions, possibly accompanied by a linguistic quantifier and an
importance coefficient, is called a
subcondition
.
The
sequence of subconditions
is the disjunction of subconditions. This structuring of various elements
of the condition adheres to the scheme assumed in Microsoft Access. As in case of the AND connec-
