Sie sind auf Seite 1von 4

International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 4 Issue: 5 238 - 241


_______________________________________________________________________________________________
Generalized evaluations in sociological researches
Abdirimov Kudrat Raximberganovich Turimov Dilmurod Mustapoyevich
Department of software of information technologies Department of software engineering
Tashkent university of information technologies(TUIT) Samarkand branch of TUIT
Tashkent, Uzbekistan Samarkand, Uzbekistan
dasturchi@mail.ru turimov86@gmail.com

AbstractConsidered the calculation of generalized evaluations in heterogeneous characteristic space. Methods of data mining are used for
calculating generalized evaluations.

Keywords-experts,generalized evaluations,attributes,gradation, classess


__________________________________________________*****_________________________________________________

I. INTRODUCTION y f x1 ,..., xn by the value of cause indicators.


The use of generalized evaluations in sociological The problem of identification in standard setting is
researches has been significant in the field. It is possible to
determine the opinion of people about particular events,
consedered. The multiplicity of objects E0 S1 ,...,S m ,
actions or happenings, their intelligence quotient and how containing representatives of 2 uncrossing classes K1 , K 2
different social and demographic groups receive information has been stated.The description of objects is fullfilled with the
via generalized indicators [1]. Up to now, generalized
evaluations were carried out only by expert judgment (expert help of sets X n ( x1 ,..., xn ) fromdiverse features n , where
querying).
of those (a subset X ( I ) ( x1 ,..., x ) ) are measured
It is generally accepted, experts who have expertise in a
particular field are able to check generalized evaluations and in interval scales, n (subset X ( N ) ( x 1 ,..., xn ) )
they are the following [2]:
in nominal X n X (I ) X ( N ) .
Points of experts;
Organizing the level of exposure of analyzed It is possible to change any k (k>2) class problem into
features of investigated objects by experts; the form of 2 class problems.
Expert estimations in even comparison matrix view. The necessity to review the objectives of two class
The drawback of above mentioned expert assessment problems is based on the following:
method is revealed in the following: - firstly, any generalized mark (indicator) has a
Determination of informative collection of factors relative feature. Any class objects are set contrary to other
conveyed in different measurement units has not been proved class objects. For instance, the opinion of the class members
by theoretical-methodological relation.The reliance of experts on choosing a spouse who have higher education degree and
on subjective ideas when evaluating generalized evaluation. who does not have;
By the method suggested in the paper, the calculation - Secondly, there are no analytical functions to restore
of the generalized evaluations is done via the intellectual connections in space of features of different types classes.
analysis method. On the basis of the calculated generalized III. THE CALCULATION OF GENERALIZED EVALUATIONS [1]
evaluation, latent (hidden) knowledge is developed.
The objective of the expert is to analyze information on The weight of value features is found by separating
the basis of "object-property" table via this method. When objects into K1 , K 2 classes.
calculating the generalized evaluation, the unrelieved
information is found on the basis of intellectual analysis of The feature value of x j which is set in increasing
artificial neural network technology on the table [3].
The method is used to calculate the generalized
progression is separated into two intervals c , c , c , c
1 2 2 3
evaluation and to explain how they are identified. In the and each of them is considered as nominal gradation feature.
following database, there is a method to find hidden rules. The border setting of c2 is based on the hypothesis which
Experts have an opportunity to check their hypothesis
states that each period includes value features of only one
of considered matters on the basis of the method.
class.
Experts may give an account of acquired knowledge as
We shall consider the number of symbol value of
a result of investigation based on linguistic rules, if necessary
they may state it by a clear formula. xj , j I asui1 , ui2 which belongs to class K i , i 1,2
and in the interval c1 , c2 , c2 , c3 .
II. THE SOLUTION OF THE PROBLEM p - is the determination of elements sequence
The objective is to determine the function number which is arranged in increasing progression of the

238
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 238 - 241
_______________________________________________________________________________________________
feature values of xj from rj1 , rj 2 ,...,rjm and in the border The value of r of the equality level of K1 and K2 classes

interval of c1 rj , c2 rj , c3 rj . in r - attribute is found by the following formula:


1 p m

The following criterion creates an opportunity to


K d ldr 1 K d ldr , p 2
calculate optimal values of borders between intervals
Ddr
c , c , c , c and to express their quantitative features in
1 2 2 3 K d K d 1, p 2
nominal measurement.
g1r g1r 1 g 2 r g 2 r 1
p


ui K 3d u3i
u 1 u 1 1 u 2 u 2 1
2 2 2 t t t t

i i
d d
i i t 1
.
i 1 2 d 1 i 1 max r D1r D2 r
K i K i 1 2 K1 K 2 (3)
i 1 r J weight of the nominal attribute is identified as
(1) stated below on the basis of the above-given (2) and (3)
In the above given formula, the expression in the left bracket formulas:
expresses the resemblance in the class and in the right bracket,
the difference among classes.
vr r r . (4)

i I is the optimal value of i - attribute therein,


It is easy to check if the nominal attribute weight is
between interval [0, 1].
i i i Evidently, it is possible to express gradations of
according to the criterion wi (1). c1 , c2 , c3 is the separating
interval border which is equivalent to the value as well as nominal attribute in the form of 1,..., p set. When
corresponding number of values of K1 and K 2 class in calculating the generalized evaluation for object
S x1 ,...,xn , xi j attribute fraction
c , c interval of u
i i 1
, ui12
is determined
as in the following( i J , j 1,..., p )
1 2 i1
criterion.
The determination of generalized evaluation of objects
by quantitative attributes is as following. In order to calculate
ij1 ij2
the value of the object S x1 ,...,xn , the following i j vi ,
K K
function is used: 1 2

RS wi ti xi c2i / c3i c1i , ij1 , ij2 is the number of j gradation in i


n
attribute in
i 1
K1 and K 2 classes.
The value of T t1 ,...,t n set is found by the following
condition:
vi is the weight of i attribute calculated according to the

min RS p max RS p max .


formula (4).
S p K1 S p K 2 The generalized evaluation of each S a E0 ,
ti 1,1therein. Sa xa1 ,...,xan object which consists of nominal
It is necessary to identify the weights of nominal attributes is found by the following formula:
attributes and their gradation for the objects expressed by
various types of attributes. RS a wi ti xai c2i / c3i c1i i xai
The weight of nominal attributes and determination of iI iJ
their gradation: (5)
r - number of attribute gradation is determined via p. Generalized evaluation (5) is closely associated with the value
r J therein. g drt is the number of gradation t belonging of the weight of quantitative (1) and nominal attributes (4).A
conclusive logic is used to express the generalized evaluation
to the class K d of r attribute. ( 1 t p ) of E0 objects in interval [0, 1].
ldr is the number of gradation of r - attribute in K d For instance, test value according to generalized
class. evaluation of E0 objects shows mixture level of objects of
The distinction between K1 and K 2 classes is K1 , K 2 classes. The higher this value is, the lower its
determined according to r - attribute as following: mixture level. Accurate (error-free) separation of classes (1)
p

g t t occurs when the value equals to one.


1r g 2r In this case (5), a discriminant analysis of information
r 1 t 1
. (2) as a critical function may be suggested.
K1 K 2

239
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 238 - 241
_______________________________________________________________________________________________
IV. COMPUTING EXPERIMENT AND CONCLUSIONS 14 Entertaining 0,261 0,512 0,509
248 subjects have been used in the research. 1-class programs.
includes literate people of Pakistan, 2-class, 124 literate people 15 Do you discuss 0,256 0,502 0,511
of Afghanistan. In the contest 33 nominal attributes (question) with neighbors?
have been used. The informational level of each attribute
16 Do you 0,254 0,414 0,615
(question) has been identified by formula (4) and arranged in discuses with
digressive order in Table 1. friends?
TABLE I. THE LEVEL OF INFORMATION TRANSFER OF ATTRIBUTES 17 What is your 0,253 0,634 0,399
field of interest
in obtaining
Attribute Weight Difference among Similarit information?
classes ies
within 18 Why do you 0,252 0,682 0,369
class listen to the
radio?
1 Has there been 0,481 0,645 0,746
built a local 19 Region. (city or 0,249 0,491 0,507
radio station in countryside)
your location?
20 What is your 0,235 0,795 0,296
favorite radio
program?

2 Place of 0,362 1 0,362 21 What is the 0,231 0,31 0,745


residence source of
information in
3 If a local radio 0,362 0,851 0,425 your area?
station has been
built in your 22 What radio do 0,23 0,78 0,295
area, does it you listen the
meet most?
requirements
23 Do you discuss 0,217 0,309 0,702
set by you?
with your
4 National news. 0,348 0,591 0,589 spouse?

24 Do you discuss 0,216 0,317 0,681


5 Local news. 0,347 0,568 0,612 with children?

25 Why do you 0,187 0,801 0,234


6 International 0,34 0,584 0,582 specifically
news. favour this
program?
7 Religious 0,325 0,56 0,58
manuals. 26 Why do you 0,185 0,81 0,2274
specifically
8 What is your 0,315 0,922 0,342 favour this
opinion about radio?
how different
local radio 27 Native 0,164 0,192 0,853
from national language.
and
international 28 What programs 0,163 0,849 0,1911
ones? do you like
discussing?
9 What is your 0,312 0,56 0,557
sex? 29 What is your 0,156 0,832 0,1874
age?
10 Cultural events. 0,288 0,5 0,576
30 What is your 0,138 0,865 0,1591
job?
11 Type of school. 0,284 0,5 0,569
31 Do you discuss 0,125 0,145 0,862
with other
12 Do you discuss 0,274 0,43 0,638 people?
with your
parents? 32 I do not discuss 0,098 0,107 0,914
with anybody.
13 Formal and 0,262 0,503 0,522
unofficial 33 And other 0,079 0,086 0,916
education. programs.

240
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 238 - 241
_______________________________________________________________________________________________
The rate of questions, their difference among classes examples of it and their value belonging to state class is
and likeness in the class is between interval [0, 1]. The rate of nearly 0.
questions occurs from multiplication of differences among 51,13 object state class of the selection is considered
classes and likeness in the class. If the value of likeness in the as boundary elements. Their belonging to the class is revealed
class is 1, it means that the members of the class answered relatively slowly.
unanimously. If the value draws to 0, it is assumed that class Belonging to their own class is revealed much slowly in
members have different opinions on the attributes (questions) the objects 10,11 of the state class in the selection.
given to them. If the value equals to 1 among classes, both 228 object of non-state class has some belonging to
class members have answered to questions differently from state class.
each other. If the value is draws closer to 0, both class
members provided similar opinions. ACKNOWLEDGMENT
The difference among classes is 1 in the 2nd question A database of sociology laboratory of Uzbekistan National
Place of residence in Table 1. It means each member of University is used as a calculation experiment.
class has answered differently from members of class 2. That
REFERENCES
is, they reside in different countries. The similarity in the class
(Pakistan 0.173; Afghanistan 0.189) means that respondents [1] Tolstova Y.N. Analysis of sociological data. Descriptive
statistics, the study of relations among nominal attributes.
are from different regions. Scientific world, Moscow, 2000.
[2] Ayvazyan S.A., Bukhshtaber V.M., Yenyukov I.S.,
Meshalkin L.D. Applied statistics. Finances and statistics,
TABLE II. THE GENERALIZED EVALUATION OF OBJECTS BELONGING TO Moscow, 1989.
K1 CLASS. Article in a journal:
[3] Ignatev N.A. Computation of generalized indicators
Object (class) R(S) Object (class) R(S) (evaluations) and data mining. Automation and
telemechanics. 2011. 5. Pp.183-190.

1 70th object 1,0 129 199- object 0,40


(state ) (non-state )

2 69- object 0,99 130 184- object 0,40


(state ) (non-state )

3 74- object 0,99 131 172- object 0,39


(state ) (non-state )

4 77- object 0,95 ..


(state )

5 72- object 0,93 160 10- object 0,28


(state ) (state )

... ... ... 161 11- object 0,28


(state )

58 228- object 0,69 ..


(non-state)

59 31- object 0,69 245 234- object 0,02


(state ) (non-state )

.. 246 221- object 0,02


(non-state )

127 51- object 0,41 247 214- object 0,01


(state ) (non-state )

128 13- object 0,40 248 220- object 0,00


(state ) (non-state )

Definitely, it is possible to draw the following


conclusions as a form of knowledge based on the obtained
results in the case that the qualitative explanation of the
generalized evaluation of the objects is in the authority of field
experts:
The value of objects 70, 69, 74, 77, 72 belonging to
state class of the selection almost equals to 1 and therefore
they are explicit examples of the class. The same goes for
objects 220, 214, 221, 234 of non-state class which are

241
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________

Das könnte Ihnen auch gefallen