國立交通大學]
On: 28 April 2014, At: 04:43
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer
Street, London W1T 3JH, UK
Cybernetics and
Systems: An
International Journal
Publication details, including
instructions for authors and subscription
information:
http://www.tandfonline.com/loi/
ucbs20
AGGREGATING FUZZY
OPINIONS IN THE GROUP
DECISION-MAKING
ENVIRONMENT
SHYI-MING CHEN
Published online: 29 Oct 2010.
To cite this article: SHYI-MING CHEN (1998) AGGREGATING FUZZY
OPINIONS IN THE GROUP DECISION-MAKING ENVIRONMENT, Cybernetics
and Systems: An International Journal, 29:4, 363-376, DOI:
10.1080/019697298125641
To link to this article:
http://dx.doi.org/10.1080/019697298125641
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of
all the information (the “Content”) contained in the publications
on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as
the Content. Any opinions and views expressed in this publication
are the opinions and views of the authors, and are not the views
of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified
with primary sources of information. Taylor and Francis shall not
be liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with,
in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private
study purposes. Any substantial or systematic reproduction,
redistribution, reselling, loan, sub-licensing, systematic supply,
or distribution in any form to anyone is expressly forbidden.
Terms & Conditions of access and use can be found at
http://
www.tandfonline.com/page/terms-and-conditions
GROUP DECISION-MAKING ENVIRONMENT
SHYI-MING CHEN
Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan, Republic of China
This pape r pre se nts a ne w m ethod for de aling with fuzzy opinion aggre ga-tion in group de cision-m aking proble ms. The propose d me thod has the
s .
following advantage s: 1 The e xpe rts’ estimate s do not ne ce ssarily have a s x s .
com mon inte rse ction at a -leve l cuts, whe re a g 0, 1 . 2 It can pe rform s .
fuzzy opinion aggre gation in a more efficient manne r. 3 It doe s not ne ed to use the De lphi m ethod to adjust trape zoidal fuzzy numbe rs give n by e xpe rts.
s
Some re searche rs B ardossy e t al., 1993; Chen & Lin, 1995; Chen et al., 1989; Hsu & Chen, 1996; Ishikawa et al., 1993; Kacprzyk & Fe drizzi, 1988; Kacprzyk e t al., 1992; Le e, 1996; Nurmi, 1981; Spillman et al.,
.
1980; Tanino, 1984, 1990; Xu & Z hai, 1992; Che n, 1997 have focuse d on the fuzzy opinion aggregation problem in the multicrite ria group
s .
de cision-making MCD M e nvironme nt based on fuzzy set the ory sZ ade h, 1965 to combine the individual opinions of experts, whe re e ach. e xpe rt usually has its own opinion or e stimate d rating unde r e ach crite rion for e ach alte rnative. Thus, to find a group conse nsus function
This work was supporte d in part by the National Scie nce Council, Republic of China, unde r grant NSC 86-2623-E-009-018.
The author would like to thank Profe ssor H. M. Hsu, De partm ent of Industrial Engine ering, National Chiao Tung Unive rsity, for he r e ncourage m ent in this work.
Addre ss corre sponde nce to Profe ssor Shyi-Ming Che n, De partme nt of Compute r and Inform ation Scie nce , National Chiao Tung Unive rsity H sinchu, Taiwan, Republic of China. E-mail: sym che n@ cis.nctu.e du.tw
Cybernetics and Systems: An International Journal, 29:363] 376, 1998
CopyrightQ1998 Taylor & Francis
0196-97 22 ¤98 $12.00 + .00 363
for aggre gating these e stimate s ratings to a common opinion is an important issue in handling multicrite ria group decision-making prob-lems. Be cause in multicriteria de cision making with group de cision-mak-ing problem s, the e stimate s of e xperts of a crite rion for an alte rnative may involve subjective ness, im pre cision, and vague ness, fuzzy set the ory can provide us with a useful way to deal with the fuzzine ss of human judgm ents.
s .
Kacprzyk e t al. 1992 showe d how fuzzy logic with linguistic quanti-s .
fie rs can be use d in group decision making. Tanino 1984 discussed fuzzy prefe rence orde rings in group decision making. Bardossy e t al. s1993 re prese nte d expert opinions or imprecise e stimate s of a physical. variable by using fuzzy numbe rs and de velope d five technique s for combining the se fuzzy numbe rs into a single fuzzy number e stim ate. The guide lines for the choice of com bination te chnique are also
pro-s .
vide d in B ardossy et al. Ishikawa e t al. 1993 propose d the max-min Delphi me thod and fuzzy De lphi me thod via fuzzy inte gration. Xu and
s .
Z hai 1992 pre sented e xte nsions of the analytic hierarchy process in a fuzzy e nvironme nt, where e ach e xpe rt repre se nts it’s subje ctive judg-me nt by an inte rval value rating of each criterion for e ach alte rnative.
s .
Le e 1996 pre sented a method for group de cision making using fuzzy se t theory for e valuating the rate of aggre gative risk in software deve
l-s .
opme nt. Nurmi 1981 prese nte d some approache s to collective de cision s .
making with fuzzy pre fere nce re lations. Hsu and Che n 1996 pre sented s .
a similarity aggregation m ethod SAM for aggregating individual fuzzy opinions into a group fuzzy conse nsus opinion, whe re the e stimate s of e xpe rts are re prese nte d by positive trape zoidal fuzzy numbe rs.
Howe ve r, the re are some drawbacks of the me thod prese nte d in s .
Hsu and Che n 1996 , shown as follows:
1. It re quires that the expe rts’ estimates have a com mon interse ction s x
at some a -le vel cut, whe re a g 0, 1 . If the initial e stimates of the
ith expe rt and the jth e xpert have no interse ction, then it must use
s .
the De lphi method Satty, 1980 or get m ore information to adjust the trapezoidal fuzzy numbe r given by e ach e xpert to obtain a com mon interse ction at the a -le vel cut. Howe ve r, applying the De lphi m ethod to adjust the trape zoidal fuzzy numbe rs given by the expe rts will take a large amount of time to pe rform the operations. 2. It re quires a large amount of tim e to calculate the de gre e of agreem ent betwe en experts’ e stimate s be cause it use s a com plicated
similarity me asure function S to calculate the de gre e of agre em ent
Ä
of the subjective e stimate Ri of expe rt Ei and subje ctive estim ate
Ä
Rj of e xpe rt Ej, s . s .H
ut
m inw
fRÄi u ,fRÄj u5
/
d uÄ
Ä
S Rt
i,Rj/
s s . s .H
ut
maxw
fRÄi u ,fRÄj u5
/
d uÄ
Ä
whe re Ri and Rj are positive trapezoidal fuzzy num bers and the
Ä
Ä
membership functions of the trape zoidal fuzzy numbe rs Ri and Rj
are fRÄi and fRÄj, re spective ly.
Thus, it is necessary to de velop a ne w me thod for dealing with the fuzzy opinion aggre gation proble m in a more flexible and m ore efficient manner.
In this pape r, we prese nt a ne w m ethod for dealing with the fuzzy opinion aggregation proble m. The propose d me thod can ove rcome the
s .
drawbacks of the one pre se nte d in Hsu and Che n 1996 due to the fact that
1. The e xpe rts’ e stim ates do not necessarily have a common interse c-s x
tion at the a le vel, whe re a g 0, 1 . Thus, it is more flexible than the one pre sented in Hsu and Che n.
2. It doe s not ne e d to use the De lphi me thod to adjust trape zoidal fuzzy numbe rs given by e xpe rts.
3. It can calculate the de gre e of similarity be twe e n the subje ctive estimate s of e xpe rts in a more e fficie nt manne r. Thus, it can pe rform fuzzy opinion aggre gation in a more efficient manne r.
BASIC CONCEPTS OF FUZZY SET THEORY
The theory of fuzzy sets was propose d by Z ade h in 1965. Le t U be the
Ä
v 4
unive rse of discourse, Us u1,u2, . . . ,un . A fuzzy se t A of U is a set v s s .. s s .. s s ..4
of ordere d pairs u1,fAÄ u1 , u2,fAÄ u2 , . . . , un,fAÄ un , where fAÄ
w x is the me mbe rship function of the fuzzy se t A, fAÄ: Uª 0, 1 , and
Ä
s .
fAÄ ui indicates the degree of me mbership of ui in the fuzzy se t A. A
Ä
fuzzy se t A of the universe of discourse U is called a norm al fuzzy set if s .
’ uig U, fAÄ ui s 1. If for all u1,u2 in U,
s s . . s s . s . . s .
fAÄ l u1q 1 y l u2 0 Min fAÄ u1 ,fAÄ u2 1
Ä
then the fuzzy set A is calle d a conve x fuzzy se t. A fuzzy numbe r is a fuzzy subset in the unive rse of discourse U that is both normal and
Ä
conve x. For e xample , Figure 1 shows a fuzzy numbe r A of the unive rse of discourse U. A standardize d fuzzy numbe r is a fuzzy numbe r define d
w x in the universe of discourse U, whe re Us 0, 1 .
Ä
A trape zoidal fuzzy numbe r M of the unive rse of discourse U can
s .
be characte rized by a quadruple a,b,c,d shown in Figure 2.
In the following, we brie fly revie w the de fuzzification technique of
s .
trape zoidal fuzzy numbe rs from Chen 1994, 1996 and Kauffman and
Ä
s .
Gupta 1988 . Conside r the trape zoidal fuzzy numbe r M shown in Figure 3, whe re e is the defuzzification value of the trapezoidal fuzzy
Ä
numbe r M. From Figure 3, we can see that
1s . s . s . s . 1s . s . s . s . by a 1 q ey b 1 s dy c 1 q cy e 1 2 2 1s . s . s . 1s . s . « 2 by a 1 q ey b s 2 dy c q cy e 1 1 s . s . s . s . « ey b y cy e s 2 dy c y 2 by a 1s . 1s . « 2es 2 dy c y 2 by a q bq c F ig u r e 1 . A fuzzy num be r.
F igu r e 2 .A trape zoidal fuzzy numbe r. dy cy bq aq 2bq 2c « 2es 2 aq bq cq d s 2 aq bq cq d s . « es 2 4
SIMILARITY MEASURES
s .Z wich e t al. 1987 have made a comparative analysis of 19 sim ilarity s .
me asure s among fuzzy se ts. In Che n and Lin 1995 we m ade a compari-son of sim ilarity of fuzzy se ts. In the following, we introduce a me thod for measuring the de gre e of similarity betwe en trapezoidal fuzzy num
-F igu r e 3 . De fuzzification of a trape zoidal fuzzy numbe r.
Ã
Ã
s .
be rs Che n & Lin, 1995 . Le t A and B be two standardize d trape zoidal fuzzy numbe rs,
Ã
s .As a1,b1,c1,d1
Ã
s .Bs a2,b2,c2,d2
whe re 0 ( a1( b1( c1 ( d1( 1 and 0 ( a2( b2( c2 ( d2 ( 1. Then the de gre e of sim ilarity be twe e n the standardize d trapezoidal fuzzy
Ã
Ã
numbe rs A and B can be me asure d by the similarity function S, <a1y a2<q <b1y b2<q <c1y c2<q <d1y d2<
Ã
Ã
s . s . S A,B s 1 y 3 4Ã Ã
Ã
Ã
s . w x s .whe re S A, B g 0, 1 . The larger the value of S A,B , the greate r the
Ã
similarity betwe en the standardized trapezoidal fuzzy numbe rs A and
Ã
B.
Ã
Ã
Le t A and B be two standardized trapezoidal fuzzy numbers,
Ã
s .As a1,b1,c1,d1
Ã
s .Bs a2,b2,c2,d2
whe re 0 ( a1( b1( c1 ( d1( 1 and 0 ( a2( b2( c2 ( d2 ( 1. Then
Ã
Ã
Ã
Ã
s . s .
it is obvious that S A,B s S B,A .
w x w x
Le t xand y be two real value s, where xg 0, 1 and yg 0, 1 . It is obvious that the real value s xand ycan be repre se nte d by standardize d
s .
trape zoidal fuzzy numbe rs x and y, re spectively, where xs x,x,x,x
s . s .
and ys y,y,y,y . B y applying formula 3 , the degre e of sim ilarity be twe e n the re al value s xand y can be evaluate d as follows:
<xy y<q <xy y<q <xy y<q <xy y< s . S x,y s 1 y 4 < < s . s 1 y xy y 4
It is obvious that this re sult coincide s with the one shown in Che n et al. s1989 ..
Ä
Le t A be a positive trapezoidal fuzzy numbe r in the unive rse of discourse U, where
w x
Us 0 ,m
Ä
s .As a,b,c,d
and 0 ( a ( b ( c ( d ( m. The n, the positive trapezoidal fuzzy num
-Ä
be r A can be translate d into the standardized trapezoidal fuzzy numbe r
Ã
A shown as follows: a b c dÃ
Ast
, , ,/
m m m mwhe re 0 ( a
r
m ( br
m ( cr
m ( dr
m ( 1 and the mem bershipfunc-Ã
tion curve of the standardize d trapezoidal fuzzy numbe r A is as shown
Ã
in Figure 4. In this case, the standardize d trapezoidal fuzzy number A is w x
de fined in the unive rse of discourse U, whe re Us 0, 1 .
A NEW METHOD FOR HANDLING FUZZY OPINION
AGGREGATION PROBLEMS
In the following, we prese nt a ne w m ethod for handling fuzzy opinion aggregation proble ms. The algorithm e ssentially is a modification of the
s .
one pre sented in Hsu and Che n 1996 . Le t U be the unive rse of
w x s .
discourse , Us 0,m . Assume that e ach e xpert Ei is 1, 2, . . . ,n
Ä
s .constructs a positive trape zoidal fuzzy num ber Ris ai,bi,ci,di to re prese nt the subjective e stimate of the rating to a given crite rion and
F ig u r e 4 . A standardiz ed trape zoidal fuzzy numbe r A.
alternative , where 0 ( ai ( bi ( ci ( di ( m. Furtherm ore , assume that
s .
the degre e of importance of export Ei is 1, 2, . . . ,n is wi, whe re
w x n
wig 0, 1 and p is 1wis 1. The algorithm is prese nte d as follows:
Ä
s .Step 1: Translate e ach trape zoidal fuzzy numbe r Ris ai,bi,ci,di
Ã
given by e xpe rt Ei into standardize d trapezoidal fuzzy number Ri s is 1, 2, . . . ,n., whe re ai bi ci di
Ã
Rist
, , ,/
m m m m s U U U U. s ai ,bi ,ci ,di and 0 ( aUi ( bUi ( cUi ( dUi ( 1. s .Step 2: Based on formula 3 , calculate the de gre e of agree ment
Ã
Ã
s .
S Ri,Rj of the opinions betwe en each pair of e xpe rts Ei and Ej,
Ã
Ã
s . w x
whe re S Ri,Rj g 0, 1 , 1( i ( n, 1 ( j ( n, and i/ j. s .
Step 3: Calculate the average de gre e of agre eme nt A Ei of e xpert Ei s is 1, 2, . . . ,n., whe re n 1
Ã
Ã
s . s . A Ei sp
S Rt
i,Rj/
6 ny 1 js 1 j/ i s .Step 4: Calculate the re lative degre e of agree me nt RA Ei of e xpe rt Ei s is 1, 2, . . . ,n., whe re s . A Ei s . s . RA Ei s n s . 7 p is 1 A Ei
Step 5: Assum e that the we ight of the de gre es of importance of the e xperts and the we ight of the relative de gree of agre em ent of the
w x w x
e xperts are y1and y2, re spective ly, where y1g 0, 1 and y2g 0, 1 . s .
Calculate the conse nsus de gre e coe fficie nt C Ei of expert Ei s is 1, 2, . . . ,n., whe re
y1 y2
s . s . s .
C Ei s ) wiq ) RA Ei . 8
y1q y2 y1q y2
Ä
Step 6: The aggregation result of the fuzzy opinions is R, where
Ä
s . s . s . s .Rs C E1 mR1
Å
C E2 mR2Å
? ? ?Å
C En mRn 9 ope rators m andÅ
are the fuzzy multiplication ope rator and the fuzzy addition ope rator, re spe ctively.In the following, we use an e xample to illustrate the fuzzy opinion aggregation proce ss.
Example: Assume that expe rts E1,E2, and E3construct positive trape
-Ä
Ä
Ä
zoidal fuzzy numbers R1, R2, and R3 to re prese nt the subje ctive e stimate of the rating to a given crite rion and alte rnative, re spectively, whe re
Ä
s . R1s 1 , 2 , 3 , 4Ä
s . R2s 4 , 5 , 6 , 7Ä
s . R3s 7 , 8 , 9 , 10Ä
Ä
Ä
Assume that the trape zoidal fuzzy num bers R1, R2, and R3are define d
w x s .
on the universe of discourse U, whe re Us 0, 10 i.e., m s 10 , and assume that the we ights of the expe rts E1, E2, and E3 are 0.4, 0.4, and
s .
0.2, respe ctively i.e., w1s 0.4, w2s 0.4, and w3s 0.2 . Furthe rmore , assume that the we ight of the degree s of importance of the e xpe rts and the we ight of the relative de gree s of agre eme nt of the expe rts are 0.9
s .
and 0.6, re spectively i.e ., y1s 0.9 and y2s 0.6 . Based on the pro-posed algorithm, we can ge t the following re sult:
wSte p 1 Be causex ms 10, we can translate the trapezoidal fuzzy
Ä
Ä
Ä
numbe rs R1, R2, and R3 into the standardize d trape zoidal fuzzy
Ã
Ã
Ã
numbe rs R1, R2, and R3, re spe ctively, whe re
1 2 3 4
Ã
R1st
, , ,/
10 10 10 10 s . s 0 .1 , 0 .2 , 0 .3 , 0 .44 5 6 7
Ã
R2st
, , ,/
10 10 10 10 s . s 0 .4 , 0 .5 , 0 .6 , 0 .7 7 8 9 10Ã
R3st
, , ,/
10 10 10 10 s . s 0 .7 , 0 .8 , 0 .9 , 1 .0wSte p 2 B ase d on formula 3 , we can ge t the following re sults:x s . <0 .1y 0 .4 q 0 .2 y 0 .5 q 0 .3 y 0 .6 q 0 .4 y 0 .7< < < < < < <
Ã
Ã
S Rt
1,R2/
s 1 y 4 s 0 .7Ã
Ã
S Rt
2,R1/
s 0 .7 <0 .1y 0 .7 q 0 .2 y 0 .8 q 0 .3 y 0 .9 q 0 .4 y 1 .0< < < < < < <Ã
Ã
S Rt
1,R3/
s 1 y 4 s 0 .4Ã
Ã
S Rt
3,R1/
s 0 .4 <0 .4y 0 .7 q 0 .5 y 0 .8 q 0 .6 y 0 .9 q 0 .7 y 1 .0< < < < < < <Ã
Ã
S Rt
2,R3/
s 1 y 4 s 0 .7Ã
Ã
S Rt
3,R2/
s 0 .7wSte p 3 Based on formula 6 , the degre e s of agree ment of expe rtsx s . E1, s . s .
E2, and E3 can be evaluate d and are e qual to A E1 , A E2 , and
s .
A E3 , respe ctively, whe re 0 .7q 0 .4 s . A E1 s s 0 .35 2 0 .7q 0 .7 s . A E2 s s 0 .7 2 0 .4q 0 .7 s . A E3 s s 0 .35 2
wSte p 4 Based on formula 7 , the relative de gree s of agre e ment ofx s . s . e xperts E1, E2, and E3 can be e valuated and are equal to RA E1 ,
s . s .
RA E2 , and RA E3 , re spe ctively, whe re 0 .35 s . RA E1 s s 0 .25 0 .35q 0 .7 q 0 .35 0 .7 s . RA E2 s s 0 .5 0 .35q 0 .7 q 0 .35 0 .35 s . RA E3 s s 0 .25 0 .35q 0 .7 q 0 .35
wSte p 5 B ecause the we ights of the expertsx E1, E2, and E3 are 0.4, 0.4,
s .
and 0.2, respectively i.e ., w1s 0.4, w2s 0.4, w3s 0.2 , and be -cause the we ight of the de gree s of importance of the expe rts and the we ight of the re lative degree s of agree me nt of the e xpe rts are
s .
0.9 and 0.6, re spectively i.e ., y1s 0.9 and y2s 0.6 , then base d on s .
formula 8 we can ge t the following re sults:
0 .9 0 .6 s . C E1 s ) 0 .4q )0 .25 0 .9q 0 .6 0 .9q 0 .6 s 0 .34 0 .9 0 .6 s . C E2 s ) 0 .4q )0 .5 0 .9q 0 .6 0 .9q 0 .6 s 0 .44
0 .9 0 .6 s .
C E3 s ) 0 .2q )0 .25
0 .9q 0 .6 0 .9q 0 .6 s 0 .22
wSte p 6 Based on formula 9 , we can see that the aggregation result ofx s .
Ä
the fuzzy opinions is the trape zoidal fuzzy numbe r R, where
Ä
s .Ä
s .Ä
s .Ä
Rs C E1 mR1Å
C E2 mR2Å
C E3 mR3 s . s . s 0 .34 m 1 , 2 , 3 , 4Å
0 .44m 4 , 5 , 6 , 7 s .Å
0 .22m 7 , 8 , 9 , 10 s . s . s 0 .34, 0 .68 , 1 .02, 1 .36Å
1 .76 , 2 .2 , 2 .64 , 3 .08 s .Å
1 .54 , 1 .76 , 1 .98 , 2 .2 s . s 3 .64, 4 .64 , 5 .64, 6 .64The mem bership function curve s of the trapezoidal fuzzy numbe rs
Ä
Ä
Ä
Ä
R1,R2,R3 and the aggre gation re sult R are shown in Figure 5.
CONCLUSIONS
s . In this pape r, we have e xtended the work of Hsu and Chen 1996 to propose a new me thod for de aling with fuzzy opinion aggregation with group decisionmaking proble ms. From the illustrative e xample pre
-F igu r e 5 .Me m be rship functions of R , R , R , and R.1 2 3
se nte d pre viously, we can se e that the propose d me thod can ove rcome s .
the drawbacks of the one pre sented in Hsu and Che n 1996 because
1. The e xpe rts’ e stim ates do not necessarily have a common interse c-s x
tion at a -leve l cuts, whe re a g 0, 1 . Thus, it is more fle xible than the one pre sented in Hsu and Che n.
2. It can calculate the de gre e of similarity be twe e n the subje ctive estimate s of e xpe rts in a more e fficie nt manne r. Thus, it can pe rform fuzzy opinion aggre gation in a more efficient manne r. 3. It doe s not ne e d to use the De lphi me thod to adjust trape zoidal
fuzzy numbe rs given by e xpe rts.
REFERENCES
Bardossy, A., L. Duckste in, and I. Bogardi. 1993. Combination of fuzzy num be rs
repre se nting e xpert opinions. Fu zzy Sets Syst. 57:173] 181.
Che n, S.-M. 1994. Using fuzzy re asoning technique s for fault diagnosis of the
J-85 je t engines. Proc. o f th e Th ird N atio n al Co n feren ce o n Scien ce an d
Tech n o lo gy o f N atio n al D efen se, Taoyuan, Taiwan, Re public of China, pp. 29] 34.
Che n, S.-M. 1996. Evaluating weapon syste ms using fuzzy arithme tic ope rations.
F u zzy Sets Syst. 77:265] 276.
Che n, S.-M. 1997. A new me thod for aggregating fuzzy opinions unde r the
group de cision making environm e nt. Pro c. of th e 1997 5t h N atio n al Co n fer
-en ce o n F u zzy Th eo ry an d Applica tio n s, Tainan, Taiwan, Republic of China,
pp. 475] 480.
Che n, S.-M., and S. Y. Lin. 1995. A new me thod for fuzzy risk analysis. Pro c.of
1995 Artificia l In telligen ce Wo rksh op, Taipe i, Taiwan, Re public of China, pp.
245] 250.
Che n, S.M., J. S. Ke, and J. F. Chang. 1989. Te chnique s for handling multicrite
-ria fuzzy de cision-making proble ms. Proc. o f th e 4th In tern atio n al Sym po
-siu m on C om pu ter an d In fo rm atio n Scien ces, Ce sme, Turke y, 2, pp. 919] 925.
Hsu, H. M., and C. T. Che n. 1996. Aggre gation of fuzzy opinions unde r group
de cision making. F u zzy Sets Syst. 79:279] 285.
Ishikawa, A., M. Amagasa, T. Shiga, G. Tom izawa, R. Tatsuta, and H. Mie no. 1993. The max-min Delphi me thod and fuzzy De lphi method via fuzzy
integration. F u zzy Sets Syst. 55:241] 253.
Kacprzyk, J., and M. Fe drizzi. 1988. A soft measure of conse nsus in the se tting
s .
of partial fuzzy prefere nce s. E u r. J.O per. Res. 34:315] 325.
Kacprzyk, J., M. Fedrizzi, and H. Nurmi. 1992. Group de cision making and
conse nsus unde r fuzzy pre ference s and fuzzy majority. F u zzy Sets Syst.
49:21] 31.
Kauffman, A., and M. M. Gupta. 1988. F u zzy m ath em atic al m od els in en gin eerin g
an d m a n agem en t scien ce. Amste rdam: North-Holland.
Le e, H. M. 1996. Group decision making using fuzzy set theory for e valuating
the rate of aggregative risk in software deve lopment. Fu zzy Sets Syst.
80:261] 271.
Nurmi, H. 1981. Approache s to colle ctive de cision making with fuzzy prefe re nce
relations. Fu zzy Sets Syst. 6:249] 259.
Satty, T. L. 1980.Th e an alytic h ie rarch y process. New York: McGraw-Hill.
Spillman, B., R. Spillman, and J. Be zdek. 1980. Fuzzy analysis of conse nsus in
small groups. In F u zzy sets: Th eory an d applica tio n s to po lic y an alysis an d
in fo rm atio n system s, e d. P. P. Wang and S. K. Chang, 291] 308. New York:
Ple num .
Tanino, T. 1984. Fuzzy pre fe re nce orde rings in group de cision making. F u zzy
Sets Syst. 12:117] 131.
Tanino, T. 1990. O n group decision making unde r fuzzy pre fe re nces. In Mu lti
-person d ecisio n m akin g u sin g fu zzy sets an d possibility th eo ry, ed. J. Kacprzyk
and M. Fedrizzi, 172] 185. Dordrecht: Kluwer.
Xu, R. N., and X. Y. Z hai. 1992. Exte nsions of the analytic hie rarchy proce ss in
fuzzy environme nt. Fu zzy Sets Syst. 52:251] 257.
Z adeh, L. A. 1965. Fuzzy sets. In fo rm.C on trol 8:338] 353.
Z wich, R., E. Carlstein, and D. V . Bude scu. 1987. Me asure s of similarity among
fuzzy conce pts: A comparative analysis. In t. J. Ap pro x. Reaso n. 1:221] 242.