Information Technology Reference
In-Depth Information
4.7 Examples of Practical Application of the Developed
Methods
Example 4.1. Building of the generalized result of expert evaluations of
educational literature quality.
Let us consider conditions of the example 3.3 and
the related results obtained. Based on the analysis described in that example, the
conclusion is drawn that the result of an evaluation of the educational literature
quality made by the third expert essentially differs from results of evaluation of
the educational literature quality made by other experts. This conclusion is
validated with the fact that the result system excluding the results provided by the
third expert has an additive index of the general consistency 0.266, and
multiplicative index of the general consistency 0.252. Considering that results of
experts are fractionalized onto two groups of results similar to each other, let us
determine consistency indexes of two subsystems {2, 4, 5}, {1, 3, 6}. The first
subsystem has additive and multiplicative indexes 0.522 and 0.5204, accordingly,
and the second subsystem - 0.503 and 0, accordingly.
According to the research carried out, the system of results of the second,
fourth and fifth experts is considered, and within the limits of their results the
generalized result of the evaluation is determined. By definition of the experts,
weight coefficients are equal to following values:
=ω
.
The generalized result of an evaluation of quality of the educational literature is
defined as a linear combination of results of the second, fourth and fifth experts.
This result is summarized in Table 4.1 in the form of parameters of membership
functions of fuzzy evaluations of each object, the accurate evaluations obtained as
a result of defuzzification of fuzzy evaluations by a method of a gravity, and
levels of a verbal scale assigned to each object.
1
/
6
ω
=
1
/
2
,
,
ω
=
1
/
3
2
5
4
Example 4.2. Building of the generalized expert approach to an evaluation of
a condition of “large-leaved linden” plants.
Let us consider conditions of an
example 3.4 and define weight coefficients of expert results within the limits of
similarity relation
R
. With the confidence level equal to 0.684, expert results
enter the same cluster, therefore
2
=
i
. With the confidence level
equal to 0.685, results of the second and third experts enter one cluster, therefore
their weight coefficients are equal:
ω
=
1
/
3
,
1
1
ω
= ω
=
5
/
12
. The weight coefficient of
2
3
ω
result of the first expert
is equal to 1/6. With the confidence level equal to
1
ρ
=
2
32
unity,
ρ
=
2
319
ρ
=
2
369
ω
=
1
/
6
,
,
, therefore
,
ω
=
1
/
2
,
1
3
2
1
2
=ω
( Table 4.2).
As said above, weight coefficients of expert results within the limits of
similarity relation
1
/
3
3
R
coincide with weight coefficients of similarity relation
R
.
2
With that, confidence levels are equal 0.857; 0.863; 1, accordingly.
