You are on page 1of 17

The Induced Fuzzy Integral

Aggregation Operator
Ronald R. Yager*
Machine Intelligence Institute, Iona College, New Rochelle, NY 10801

We discuss the fuzzy integral. The centrality of the ordering operation, based upon the arguments
to be aggregated, is pointed out. We then extend the fuzzy integral aggregation operator by
allowing the ordering operation to be based upon values other then those being aggregated. This
leads to the induced fuzzy integral aggregation operator. We look at this new operator and study
its properties. We show its relationship to a formulation called limited fuzzy integral aggregation.
It is shown how this new induced fuzzy integral operator provides a natural framework for the
implementation of nearest neighbor rules. Throughout this work, use is made only of the ordinal
aspects of the information used. © 2002 Wiley Periodicals, Inc.

1. INTRODUCTION
The Sugeno or Fuzzy Integral1,2 provides an aggregation operator which can
be seen as providing a generalization and extension of the logical anding and oring
operations. Some applications in which the Fuzzy Integral (FI) can be used are in
the aggregation of the truths of propositions, the construction of multicriteria
decision functions from individual criteria, and the determination of the satisfac-
tion of a hypothesis to a collection of evidence. It can be used to provide operators
for the aggregation of fuzzy subsets. In these applications the arguments being
aggregated can be seen as some kinds of truth or belief values. While applications
of this operator have typically used values drawn from the unit interval I ⫽ [0, 1],
an important feature of this operator is that it only requires that the information
used be drawn from an ordinal scale. This situation opens its usefulness to
applications in many environments in which we are trying to model aspects of
human intelligence. Particularly notable are the possibilities for aggregating infor-
mation expressed in linguistic terms as these terms can often only be ordered. This
situation makes this aggregation operator useful for the development of Zadeh’s
agenda of computing with words.3,4
Central to the implementation of the FI aggregation operator is an ordering
operation based upon the arguments to be aggregated. This ordering operation, in

*e-mail: yager@panix.com.

INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS, VOL. 17, 1049 –1065 (2002)


© 2002 Wiley Periodicals, Inc. Published online in Wiley InterScience
(www.interscience.wiley.com). • DOI 10.1002/int.10055
1050 YAGER

an interesting nonlinear way, plays a fundamental role in determining the param-


eters associated with the arguments in the aggregation process. In this work we
consider a generalization of the FI to what we call the Induced Fuzzy Integral
(I-FI). Under this generalization, we allow the ordering process to be based upon
some values other those being aggregated. This generalization allows for the
inclusion of context-dependent information, such as relevancy, in the aggregation
process.
While the operations involved in the FI and the I-FI can be implemented when
the information is drawn from a numeric scale, in the discussion to follow we
assume a weaker requirement where the values are drawn from a finite ordinal
scale. In the following we shall let T ⫽ {t 0 , t 1 , . . . , t m } be a set of elements
having a linear ordering t j ⬎ t i if j ⬎ i. Such a structure provides a finite ordinal
scale. This type of ordinal scale often arises when we are dealing with linguistic
values associated with a variable. A typical example of such a scale is {Perfect,
Very High, High, Medium, Low, Very Low, None}. We note that such a scale
allows the performance of the binary operations Max(∨) and Min(∧) and the unary
operation of complementation. We recall5 for any element t j 僆 T its complement,
denoted t̃ j , is the value t m⫺j . We note that t̃ 0 ⫽ t m , t̃ m ⫽ t 0 and if t a ⬎ t b then
t̃ b ⬎ t̃ a .

2. THE FUZZY MEASURE AND FUZZY INTEGRAL


Underlying and guiding the application of the Fuzzy Integral is a fuzzy
measure.1,2,6 Before turning to the FI aggregation operator, we briefly introduce the
fuzzy measure.

DEFINITION. A fuzzy measure defined on a finite set X using the scale T is a


mapping ␮: 2X 3 T such that (1) ␮(A) ⫽ t0, (2) ␮(X) ⫽ tm, and (3) ␮(A) ⱕ ␮(B)
if A 傺 B.
We call a fuzzy measure binary, if for all subsets E, ␮ (E) 僆 {t 0 , t m }. Two
examples of fuzzy measures which are binary are: ␮* where ␮ * (X) ⫽ t m and
␮ * (F) ⫽ t 0 for F ⫽ X and ␮*, where ␮ *(A) ⫽ t 0 and ␮ *(F) ⫽ t m for F ⫽ A.
One class of fuzzy measures are the cardinality based measures7; ␮ (F) just
depends on the cardinality of F. For these measures ␮ (F) ⫽ ␮ (E) if 兩F兩 ⫽ 兩E兩.
Another class of fuzzy measures are the possibility measures8,9 here ␮ (E 艛 F) ⫽
Max[ ␮ (E), ␮ (F)]. Another class useful in ordinal environments are necessity
measures9,10 here ␮ (E 艚 F) ⫽ Min[ ␮ (E), ␮ (F)]. In discussing necessity
measures it is useful to introduce subsets F j ⫽ X ⫺ { x j } since any subset E can
be expressed as E ⫽ 艚 x jⰻE F j and hence ␮ (E) ⫽ Minx j僆E៮ [ ␮ (F j )].
There exists a very special relationship between possibility measures and
necessity measures.9 If ␮ is a fuzzy measure, then the measure ␮ d defined such that
˜
␮d 共 A兲 ⫽ ␮共A៮ 兲 is called its dual (we recall that the tilde means the complement in
the space T and the overbar means the set complement in X). If ␮ is a necessity
measure, then its dual is a possibility measure. Specifically, if ␮ is a necessity
measure with ␮ (F ) ⫽ ␤ , then its dual possibility measure has ␮d 共兵 x 其兲 ⫽ ␤ ˜.
i i i i
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1051
Similarly, if ␮ is a possibility measure, then its dual measure is a necessity
measure, for this measure ␮ d (E) ⫽ Mini僆E៮ ␣˜ i and therefore ␮ d (F i ) ⫽ ␣˜ i .
We now define the Fuzzy Integral (FI) aggregation operator.1,11 Let Ꮽ ⫽
{A 1 , . . . , A n } be a collection of objects each of which has an associated value
drawn from the space T, we let a j be the value associated with A j . Furthermore, we
let ␮ be a fuzzy measure over Ꮽ with image space T. The Fuzzy Integral
Aggregation of the a j guided by ␮ is defined as
n

S ␮共a 1, a 2, . . . , a n兲 ⫽ Max 关␮共Hj 兲∧aa-index共j兲 兴


j⫽1

where a-index( j) is the index of the jth largest of the a i and H j ⫽ {A a-index(r) 兩r ⫽
1 to j}. Letting b j ⫽ a a-index( j) , the jth largest of the a i , and letting m j ⫽ ␮ (H j ),
we can express S ␮ (a 1 , a 2 , . . . , a n )) ⫽ Maxj [m j ∧ b j ].
For a particular aggregation, S ␮ (a 1 , . . . , a n ), we let B denote a vector having
b j as its jth element, we call this the ordered argument vector. We also let M denote
a vector, called the weighting vector, whose jth component m j is ␮ (H j ). Even
though the weighting vector M is dependent upon the arguments, it must always
have m j ⱖ m i if j ⬎ i and m n ⫽ t m . We shall say a vector having these two
properties is proper. It is well known that the FI aggregation operator it is bounded,
Mini [a 1 ] ⱕ S ␮ (a 1 , . . . , a n ) ⱕ Maxi [a i ] and monotonic, if â i ⱖ a i for all i then
S ␮ (â 1 , . . . , â n ) ⱖ S ␮ (a 1 , . . . , a n ).
A prototypical application of the FI aggregation operator is in the problem of
hypothesis verification. Let the Ꮽ ⫽ {A 1 , . . . , A n } be a collection of evidence
and let a i be the degree of compatibility of a proposed hypothesis with the evidence
A i . Here we let the fuzzy measure ␮ be defined such that ␮ (E) indicates the
credibility associated with a hypothesis that is compatible with all the pieces of
evidence in the set E. In this case the FI aggregation can be used to provide a
measure of the validation of the hypothesis.
By appropriate choice of the underlying fuzzy measure the FI operator leads
to a large class of different aggregation operators. We refer the interested reader to
Refs. 1, 12, and 13 for more details.

3. THE INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR


As we previously noted, the FI aggregation operator is S ␮ (a 1 , . . . , a n ) ⫽
Maxj [m j ∧ b j ] where b j ⫽ a a-index( j) and m j ⫽ ␮ (H j ), where H j ⫽ {A a-
index(k) 兩k ⫽ 1 to j}. Central to the application is the step of ordering the arguments.
This step produces the ordering index function a-index( j), which indicates the
index of the argument having the jth largest value.† Inspired by Refs. 14 and 15,
we shall consider a more general policy towards the ordering process used in this
aggregation. This will lead us to the idea of Induced Fuzzy Integral (I-FI) aggre-
gation operator.


Essentially a-index is a mapping a-index: N 3 N, where N ⫽ {1, 2, . . . , n}.
1052 YAGER

We again assume we have n objects participating in the aggregation, Ꮽ ⫽


{A 1 , . . . , A n }, and the existence of a fuzzy measure ␮ on Ꮽ taking values in the
ordinal space T. In this more general framework we assume that each of the objects
A i participating in the aggregation has an associated two-tuple 具v i , a i 典. In this tuple
we shall refer to v i as the order inducing variable and a i as the argument
variable. We shall assume the a i are drawn from the same space T as ␮ while V i
can be drawn from any ordinal space R.
In this environment which we shall call Induced Fuzzy Integral (I-FI) Ag-
gregation we are still interested in aggregating the a i but the ordering is induced by
the variable v i instead of a i . In particular, we define an indexing function v-index
such that v-index( j) is the index of the object having the jth largest of the v i values.
Thus v-index( j) is obtained by ordering the v i in descending value. Using this, we
define the I-FI aggregation operator as:
S ␮共具v 1, a 1典, 具v 2, a 2典, . . . , 具v n, a n典兲 ⫽ Max关wj ∧dj 兴
j

where d j ⫽ ∧ j
a v-index(k) and w j ⫽ ␮ (G j ), where G j ⫽ {A v-index(k) 兩k ⫽ 1 to
k⫽1
j}, the subset of elements with the j highest values for the inducing variable. We
further note that d j is the minimum of the arguments of the objects with the j
highest-order inducing values. At times we shall find it convenient to denote a
vector W consisting of the w j as the weighting vector W and a vector D consisting
of the d j as the ordered argument vector.
A simple example will illustrate the I-FI aggregation.

Example. Let Ꮽ ⫽ {A 1 , A 2 , A 3 , A 4 } and let T ⫽ {t 1 , t 2 , . . . , t 10 }. Let R ⫽


{r 1 , r 2 , . . . , r 7 } be another ordered set where r j ⬎ r i if i ⬎ j. Let ␮ be a fuzzy
measure on Ꮽ with image T for which:

␮ 兵兵A 1其其 ⫽ t 1, ␮ 兵兵A 2其其 ⫽ t 1, ␮ 兵A 3其 ⫽ t 2, ␮ 兵A 4其 ⫽ t 2


␮ 兵A 1, A 2其 ⫽ t 2, ␮ 兵A 1, A 3其) ⫽ t 2, ␮ 兵A 1, A 4其 ⫽ t 3, ␮ 兵A 2, A 3其 ⫽ t 3,
␮兵A2 , A4 其 ⫽ t3
␮ 兵A 3, A 4其 ⫽ t 4
␮ 共兵A 1, A 2, A 3其兲 ⫽ t 6, ␮ 兵A 1, A 2, A 4其 ⫽ t 7, ␮ 共兵A 1, A 3, A 4其兲 ⫽ t 8,
␮ 共兵A 2, A 3, A 4 ⫽ t 9
␮ 共兵A 1, A 2, A 3, A 4兲 ⫽ t 10.
Consider the I-FI aggregation where our tuples are:
A 1 ⫽ 具r 3, t 2典, A 2 ⫽ 具r 4, t 6典, A 3 ⫽ 具r 6, t 5典, A 4 ⫽ 具r 2, t 4典
The first step is to order the objects with respect to their order inducing value:
A 3 ⫽ 具r 6, t 5典
A 2 ⫽ 具r 4, t 6典
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1053
A 1 ⫽ 具r 3, t 2典
A 4 ⫽ 具r 2, t 4典
Using this, we get v-index(1) ⫽ 3, v-index(2) ⫽ 2, v-index(3) ⫽ 1, v-index(4) ⫽ 4
From this we can obtain G j ⫽ {A v-index(k) 兩k ⫽ 1 to j}:
G 1 ⫽ 兵A 3其, G 2 ⫽ 兵A 3, A 2其, G 3 ⫽ 兵A 3, A 2, A 1其, G 4 ⫽ 兵A 1, A 2, A 3, A 4其
Using the fuzzy measure defined above, we have ␮ (G 1 ) ⫽ t 2 , ␮ (G 2 ) ⫽ t 3 ,
␮ (G 3 ) ⫽ t 6 , ␮ (G 4 ) ⫽ t 10 . In addition with d j ⫽ Mink⫽1–j [a v-index(k) ], we get:
d 1 ⫽ Min关a3 兴 ⫽ t5
d 2 ⫽ Min关a3 , a2 兴 ⫽ Min关t5 , t6 兴 ⫽ t5
d 3 ⫽ Min关a3 , a2 , a1 兴 ⫽ Min关t6 , t5 , t2 兴 ⫽ t2
d 4 ⫽ Min关a4 , a3 , a2 , a1 兴 ⫽ Min关t5 , t6 , t2 , t4 兴 ⫽ t2
Using this, we get:
S ␮共具v i, a i典兲 ⫽ Maxj 关␮共Gj 兲∧dj 兴 ⫽ 共t2 ∧t5 兲∨共t3 ∧t5 兲∨共t6 ∧t2 兲∨共t2 ∧t10 兲 ⫽ t3
We shall now take a look at this I-FI aggregation operator. The first obser-
vation we should make is that this I-FI operator is an extension of the ordinary
fuzzy integral. In particular, if the order inducing variable is the same as the
argument variable we end up with the ordinary FI operator. Consider S ␮ (具v i ,
a i 典) ⫽ Maxj [ ␮ (G j ) ∧ Mink⫽1–j [a v-index(k) ]] and assume v i ⫽ a i . In this case
a v-index(k) ⫽ a a-index(k) and G j ⫽ {A v-index(k) 兩k ⫽ 1 to j) ⫽ H j ; thus ␮ (G j ) ⫽
␮ (H j ) ⫽ m j ; hence S ␮ (具v i , a i 典) ⫽ Maxj [m j ∧ Mink⫽1–j [a a-index(k) ]]. Since
a a-index(k 1 ) ⱖ a a-index(k 2) for k 1 ⬍ k 2 , it follows that Mink⫽1–j [a a-index(k) ] ⫽
a a-index( j) ⫽ b j ; thus in this case S ␮ (具v i , a i 典) ⫽ Maxj [m j ∧ b j ], it is the fuzzy
integral aggregation.
We note that if ␮ is a binary measure, then S ␮ (具v i , a i 典) always takes as its
value one of the a i , S ␮ (具v i , a i 典) 僆 {a 1 , . . . , a n }.
Let us now investigate some properties of this I-FI operator. With w j ⫽ ␮ (G j )
where G j ⫽ {A v-index(k) 兩k ⫽ 1 to j} we see that it has the two basic properties
described before: w j ⱖ w i if j ⬎ i and w n ⫽ t m . Thus the vector W is proper. From
this it follows that the I-FI operator acts like a mean type aggregation operator with
respect to the argument values the a i ’s:
Min 关ai 兴 ⱕ S␮ 共具vi , ai 典兲 ⱕ Max 关ai 兴 Boundedness
i i

if ai ⫽ a* for all i, then S␮ 共具vi , ai 典兲 ⫽ a* Idempotency


if ai ⱖ âi for all i, then S␮ 共具vi , ai 典兲 ⱖ S共vi ; âi 典) Monotonicity with respect
to the arguments
It can also be seen as being symmetric with respect to the arguments.
One important property of this operator can be seen from the following
1054 YAGER

observations. With G j ⫽ {A v-index(k) 兩k ⫽ 1 to j} we see ∧ k⫽1 j


a v-index(k) is the
minimal of argument values for the objects in G j . We have denoted this as d j . We
see that ␮ (G j ) is increasing in j, if j⬘ ⬎ j, then G j 傺 G j⬘ , and hence ␮ (G j ) ⱕ
␮ (G j⬘ ). On the other hand, d j is decreasing in j; if j⬘ ⬎ j, then d j ⱖ d j⬘ . With
A a-index(n) indicating the object with the smallest argument value, we see that if
d j ⫽ a a-index(n) for some j, then d k ⫽ a a-index(n) for all k ⱖ j. Furthermore, we
see that if the object with the smallest argument value is in G 1 , then d j is always
that value; thus, if d 1 ⫽ a a-index(n) , then d j ⫽ a a-index(n) for all j. In this special
case when d 1 ⫽ a a-index(n) , we have:
S ␮共具v i, a i典兲 ⫽ Max 关␮共Gj 兲∧dj 兴 ⫽ Max 关␮共Gj 兲∧aa-index共n兲 兴 ⫽ aa-index共n兲
j j

This leads us to the observation that if the object with the largest order inducing
value has the smallest argument value, then its argument value is the valuation of
S ␮ (具v i , a i 典).
We can also show that there exists an even tighter boundary on the I-FI
aggregation operator then the Max and Min of the argument values. We first see
that d 1 ⫽ a v-index(1), the argument value of the object with the largest order
inducing value. Since d j ⱕ d 1 for all j, then:
Max 关␮共Gj 兲∧dj 兴 ⱕ d1 ⱕ av-index共1兲
j

Furthermore, since d n ⫽ Mini [a i ] and ␮ (G n ) ⫽ t m , then Maxj [ ␮ (G j ) ∧ d j ] ⱖ


Mini (a i ). Thus we see that the effective boundary I-FI aggregation is:
a v-index共1兲 ⱖ S␮ 共具vi , ai 典兲 ⱖ Min 关ai 兴
i

Specifically we observe that the upper boundary of the I-FI operator is the
argument value of the object with the largest order inducing value.

4. EXAMPLES OF I-FI
Let us look at this induced FI operator for some special cases of fuzzy
measures. First we shall consider the case when ␮ is a cardinality based measure.
For these measures the weighting vector W, whose components are the ␮ (G j ), just
depends on the cardinality. Thus w j does not depend on the indexing function,
v-index, it is constant in this situation. In this case S ␮ (具v j , a j 典) ⫽ Maxj [w j ∧ d j ]
with w j independent of the argument and d j ⫽ Mink⫽1 to j [a v-index(k) ].
Consider first the special case of ␮ * , w n ⫽ t m and w j ⫽ t 0 for all j ⫽ n, here:
S ␮共具v j, a j典兲 ⫽ Max 关wj ∧dj 兴 ⫽ wn ∧dn ⫽ dn ⫽ Min 关ai 兴
j i

It is the minimal argument value.


Consider now ␮ *, w j ⫽ t m for all j, here S ␮ (具v j , a j 典) ⫽ Maxj [w j ∧ d j ] ⫽
t m ∧ Maxj [d j ]. Since d 1 ⱖ d j for all j, then S ␮ (具v j , a j 典) ⫽ t m ∧ d 1 ⫽ a v-index(1).
It is equal to the argument value of the object with the largest order inducing value.
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1055
We consider now the cardinality based measure where ␮ (A) ⫽ t 0 , ␮ (Ꮽ) ⫽
t m , and ␮ (E) ⫽ ␣ for all others. In this case:
S ␮共具v j, a j典兲 ⫽ Max 关wj ∧dj 兴 ⫽ dn ∨共␣∧d1 兲 ⫽ Min 关ai 兴∨共␣∧av-index共1兲 兲
j i

We note that both ␮* and ␮ are special cases of this. If ␣ ⫽ t m , we get ␮* and,
*
if ␣ ⫽ t 0 , we get ␮ .
*
Another special case of cardinality based measures are what we shall call the
Kth component measure. Here ␮ (E) ⫽ t 0 if Card(E) ⬍ K and ␮ (E) ⫽ t m if
Card(E) ⱖ K. From the nature of ␮ , w j ⫽ t 0 if j ⬍ K and w j ⫽ t m if j ⱖ K. Thus
in this case:
K
S ␮共具v j, a j典兲 ⫽ Max 关wj ∧dj 兴 ⫽ ∧ av-index共i兲
j i⫽1

It is the minimal argument value among the K objects with the largest order
inducing values. A special case of this is a kind of median. Here K is equal to n⫹1 2
if the number of arguments is odd or K ⫽ n/ 2 if the number of arguments is even.
Here we essentially take half the objects with the largest inducing value and then
use as our aggregation the minimal argument value of these. We note that if the
argument and order inducing variable are the same this reduces the median, “we
take the smallest argument value among the half the objects with the largest
argument value.”
Let us now consider the possibility measure in this induced environment. In
this case ␮ (E) ⫽ Maxi僆E [ ␣ i ] where ␣ i ⫽ ␮ ({A i }) and at least one ␣ i ⫽ t m . With
G j ⫽ {A v-index(k) 兩k ⫽ 1 to j} we see that ␮ (G j ) ⫽ ∨ k⫽1 j
␣ v-index(k) , it is the
highest possibility among the j elements with largest inducing values. In this
situation S ␮ (具v i , a i 典) ⫽ Maxj [∨ k⫽1 j
␣ v-index(k) ∧ ∧ k⫽1
j
a v-index(k) ].
We note that if v-index(1) has the largest possibility, ␣v-index(1) ⫽ t m , then
S ␮ (具v i , a i 典) ⫽ a v-index(1). Thus, if the object with the largest order inducing value
has the highest possibility, then its argument value is the aggregated value. More
generally, we see that if the object with the qth largest order inducing value has
possibility t m , ␣ v-index(q) ⫽ t m , then S ␮ (具v i , a i 典) ⱖ ∧ k⫽1
q
a v-index(k) . It is at least
as big as the smallest argument among the elements with the q largest order
inducing values.
Let us consider the case of the necessity measure. We recall for this measure
if we denote ␤ i ⫽ ␮ (Ꮽ ⫺ {A i }), then ␮ (E) ⫽ ∧ iⰻE ␤ i . It is also assumed that
at least one ␤ i ⫽ t 0 . In this case with G j ⫽ {A v-index(k) 兩k ⫽ 1–j}, we have
␮ (G j ) ⫽ ∧ k⫽j⫹1n
␤ v-index(k) . Using this, we have:

S ␮共具v i, a i典兲 ⫽ Max 关␮共Gj 兲∧dj 兴 ⫽ Max


j j
冋 n j

∧ ␤v-index共k兲 ∧ ∧ av-index共k兲 .
k⫽j⫹1 k⫽1

One issue that must be addressed when using this induced type aggregation is
related to the situation when there exists ties among the arguments with regard to
the order inducing variable.
1056 YAGER

Example. Consider the case of aggregating two objects Ꮽ1 and Ꮽ2 in which we


have the measure:

␮ 共兵A 1其兲 ⫽ t m, ␮ 共兵A 2其兲 ⫽ t m, ␮ 共兵A 1, A 2其兲 ⫽ t m


Let the two pairs being aggregated be:
具v 1, a 1典 ⫽ 具r 3, t 5典
具v 2, a 2典 ⫽ 具r 3, t 4典
We see that there exists a tie between the order inducing variables. This means that
there exists two possible definitions for v-index.
v-index1 共j兲 v-index2 共j兲
1 2
2 1
Thus the different possible adjudications of the tied values lead to different
approaches.
If we use v-index1, then:
S ␮共具a i, v i典兲 ⫽ 共 ␮ 共兵A 1其兲∧a 1兲∨共 ␮ 共兵A 1, A 2其兲∧共a 1∧a 2兲兲 ⫽ 共t m∧t 5兲∨共t m∧t 4兲 ⫽ t 5
If we use v-index2, then:
S ␮共具a 1, v 2典兲 ⫽ 共 ␮ 共A 2兲∧a 2兲)∨共 ␮ 共A 1, A 2兲∧共a 1∧a 2兲兲 ⫽ 共t m∧t 4兲∨共t m∧t 4兲 ⫽ t 4
Thus the different possible adjudications of the tied values lead to different
answers.
When faced with arguments that are tied with respect to the order inducing
variables, we shall adjudicate the tied elements by using their argument variables.
In particular, we shall lexicographically order the tied elements in descending order
of their argument value.

Example. Consider the following:

A 1 ⫽ 具r 4, t 4典
A 2 ⫽ 具r 6, t 3典
A 3 ⫽ 具r 3, t 2典
A 4 ⫽ 具r 4, t 6典
A 5 ⫽ 具r 4, t 3典
Ordering these by their induced values, we get:
A2
A1 ⫺ A4 ⫺ A5
A3
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1057
We see that A 1 , A 4 , and A 5 are tied with respect to their order inducing value.
Adjudicating the tied elements in descending order of their argument value, we get:
A2
A4
A1
A5
A3
This adjudication gives us an order inducing function v-index( j) such that
v-index共1兲 ⫽ 2, v-index共2兲 ⫽ 4, v-index共3兲 ⫽ 1, v-index共4兲 ⫽ 5, v-index共5兲 ⫽ 3
In the following section we shall provide some additional ideas which enable
us to justify this method of dealing with ties.

5. ALTERNATIVE REPRESENTATIONS OF THE FI


AGGREGATION OPERATOR
In the subsequent section we shall provide a general view of the relationship
between the FI aggregation operator and the I-FI aggregation operator; however,
before proceeding to this, we shall introduce some alternative interpretations and
representations of the FI aggregation operator.16,17
Let Ꮽ ⫽ {A 1 , . . . , A n } be a collection of objects, and let ␮ be a fuzzy
measure over Ꮽ taking values in the ordinal space T. Let a j 僆 T be a value
associated with A j indicating the degree to which A j satisfies some condition.
Consider now the aggregation:
Agg␮ 共a1 , . . . , an 兲 ⫽ Max 关␮共E兲∧MinA i僆E 关ai 兴兴
E債Ꮽ

Let us denote MinA i僆E [a i ] as Q(E). Q(E) indicates the degree to which all the
objects in E satisfy our condition.
As an example that brings some useful intuition to this aggregation we
consider a multicriteria aggregation problem, here we are trying to determine the
appropriateness of an alternative solution to our requirements. Here Ꮽ are a set of
criteria of interest, and ␮ (E) indicates the degree that we are satisfied with a
solution that satisfies all the criteria in E. We shall more simply denote ␮ (E) as the
degree to which E provides a credible collection of criteria. In this framework a i
indicates the degree that an alternative satisfies criteria A i . Under this semantics
Q(E) indicates the degree to which the alternative being considered satisfies all the
criteria in E. Here, then, our evaluation of an alternative is obtained as
Agg␮ ( A 1 , . . . , a n ).
The semantics of this aggregation is the following: We note that the term
␮ (E) ∧ Q(E) indicates the degree to which the subset E provides an acceptable
collection of criteria and the degree to which E is satisfied by the alternative being
evaluated. With this in mind, our operator Agg␮ can be viewed as determining the
satisfaction using the following imperative:
1058 YAGER

An alternative is acceptable if there exists any credible


collection of criteria which it satisfies
We note the Max operation over the subsets results from our imperative of only
requiring satisfaction to any credible subset of criteria.
We note that if we view the fuzzy measure ␮ as a fuzzy subset over the space
2Ꮽ and Q also as a fuzzy subset of 2Ꮽ, then letting R ⫽ ␮ 艚 Q we take
Agg␮ (a 1 , . . . , a n ) ⫽ MaxE {R(E)].
Before proceeding, we shall introduce an alternative expression for
Agg␮ (a 1 , . . . , a n ). Let a-index( j) be the index of the jth largest of the a i . With
n being the cardinality of Ꮽ, let N ⫽ {1, 2, 3, . . . , n}. If G is a subset of N, we
shall define E G as a subset of Ꮽ such that E G ⫽ {A a-index(k) 兩k 僆 G}. Using this
notation, we can express Agg␮ (a 1 ):
Agg␮ 共a1 , . . . , an 兲 ⫽ Max 关␮共E兲∧Min 关ai 兴
E債Ꮽ A i 僆E

Agg␮ 共a1 , . . . , an 兲 ⫽ Max 关␮共EG 兲∧Min 关aa-index共i兲兴兴


E債Ꮽ i僆G

11
What is of interest to us is that it can be shown that Agg␮ (a 1 , . . . , a n ) is
the FI aggregation operator, Agg␮ (a 1 , . . . , a n ) ⫽ S ␮ (a 1 , . . . , a n ).

6. LIMITED FUZZY INTEGRAL AGGREGATION


In the following we shall use the representation of the FI in terms of Agg␮ to
introduce the concept of a limited fuzzy integral aggregation operation. This
concept will allow us to get a deeper understanding between the FI and I-FI. First
recall that the FI aggregation is essentially a weighted aggregation over all subsets
of the space Ꮽ:
S ␮共a 1, . . . , a n兲 ⫽ Agg␮ 共a1 , . . . , an 兲 ⫽ Max 关␮共E兲∧Min 关ai兴兴
E債Ꮽ A i 僆E

⫽ Max关␮共EG 兲∧Min 关aa-index共i兲兴兴


G債N i僆G

In the following we shall introduce a mapping ␭ : 2 Ꮽ 3 T called a selecting


or limiting function. We shall say it is a binary selecting function if ␭ (E) 僆 {t 0 ,
t m }. We now define a Limited Fuzzy Integral as:
Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ Max关␮共E兲∧␭共E兲∧Min 关ai兴兴
E債Ꮽ A i 僆E

We easily see that Agg␮ 兩 ␭ (a 1 , . . . , a n ) ⱕ Agg␮ (a 1 , . . . , a n ).


In the case of the binary selecting function the limited fuzzy integral is
selecting a subset of subsets of Ꮽ to include in the aggregation. In the case when
␭ is not binary it is selecting a fuzzy subset of subsets of Ꮽ to include.
We can view the limiting function as a fuzzy subset ␭ of Ꮽ. In this case
denoting R̂ ⫽ ␭ 艚 ␮ ∧ Q and then take as our aggregation any subset that satisfies
this:
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1059

Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ Max 关R̂共E兲兴


E債Ꮽ

While the limited FI aggregation operator provides a formal mechanism for


modifying the aggregation process, the choice of the limiting function ␭ provides
a vehicle for including various types of addition requirements and information into
the aggregation process and as such can be used to aid in the construction of
intelligent systems. As we shall see, this can allow information about contextual
and relevance into the aggregation process.
In the following we shall show that the induced FI aggregation operator is a
special case of the limited FI. Assume we have some linear ordering S on the
elements of Ꮽ. Let s-index be the index function of this ordering, that is, s-index( j)
is the index of the object in Ꮽ, which is jth in the ordering. In the following we
shall describe the construction of a selection function ␭ based on this ordering;
however, before proceeding we shall introduce some terminology. We shall say a
subset of Ꮽ corresponds to a rooted subordering of our ordering S if it contains
only and all elements up to some j, that is the subset contains only the j first
elements in the ordering.
We now define ␭ such that ␭ (E) ⫽ t m if E is a rooted sub-ordering of and
␭ (E) ⫽ t 0 if E is not. That is, if E ⫽ {A s-index(k) 兩k ⫽ 1 to j} for some j, then
␭ (E) ⫽ t m ; otherwise ␭ (E) ⫽ t 0 . Using this choice of ␭, we would like to
calculate:
Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ Max 关␮共E兲∧␭共E兲∧Min 关ai兴兴
E債Ꮽ A i 僆E

First we note that ␭ (E) ⫽ t m only for sets G j ⫽ {A s-index(k) 兩k ⫽ 1 to j}; otherwise
it is t 0 . Hence, in this case, Agg␮ 兩 ␭ (a 1 , . . . , a n ) ⫽ ∧ j⫽1
n
[ ␮ (G j ) ∧ MaxA i僆G j[a i ]].
However, from the definition of G j we see that MaxA i僆G j[a i ] ⫽ ∧ k⫽1 j
a s-index(k) .
Thus Agg␮ 兩 ␭ (a 1 , . . . , a n ) ⫽ ∧ j⫽1 [ ␮ (G j ) ∧ ∧ k⫽1 a s-index(k) ], which is essentially
n j

an I-FI aggregation. We see that the I-FI is an example of a limited FI aggregation.


In the case of the I-FI the ordering is induced by the order inducing variable. Since
we have shown that the limited FI is never greater than the associated FI, then we
see the I-FI never aggregates to a value greater than the FI operator.
The view of the induced aggregation as a type of limited FI aggregation
allows us to provide an understanding and justification of procedure suggested
earlier for handling the I-FI in the case of induced orderings in which there
are ties.
In the case of the limited FI operator just studied we defined a limiting
function ␭, based on our given linear ordering S such that ␭ (E) ⫽ t m if E is a
rooted subordering of S and ␭ (E) ⫽ t 0 ; otherwise, we recall E is a rooted
subordering of S if E contains all and only elements up to some position in the
ordering. We now extend the ideas to the case in which we have ties. We recall an
ordering which has ties is called a weak ordering. We first generalize our idea of
a rooted subordering to weak orders in a way that maintains its definition for linear
orderings.
1060 YAGER

DEFINITION. Let P be a weak ordering on the set Ꮽ. We say that a subset E of Ꮽ


corresponds to a rooted subordering with respect to P if E contains all the elements
ahead of the lowest element in it.‡
Thus, if E contains the jth element of the ordering as its lowest element, then
it must contain all the elements higher than j in the ordering. What is implicit in
this definition is that while all elements ahead of the lowest element in a rooted
subordering E must be in E any elements tied with the lowest element in E may
or may not be in E.

Example. As an example illustrating the concept of rooted subordering consider


the following weak ordering:
A1
A2 A3 A4
A5
A6 A7
The following are all the rooted suborderings associated with this weak ordering;
兵A 1其
兵A 1, A 2其
兵A 1, A 3其
兵A 1, A 4其
兵A 1, A 2, A 3其
兵A 1, A 3, A 4其
兵A 1, A 2, A 4其
兵A 1, A 2, A 3, A 4其
兵A 1, A 2, A 3, A 4, A 5其
兵A 1, A 2, A 3, A 4, A 5, A 6其
兵A 1, A 2, A 3, A 4, A 5, A 6其
兵A 1, A 2, A 3, A 4, A 5, A 6, A 7其
We note that for a linear ordering with n elements we have n subsets that are
rooted suborderings. More generally, for a weak ordering with N with levels in
which n i elements are tied at that ith level, the number of subsets which are rooted
suborderings is ¥ i⫽1 N
2 n i⫺1 . This is because any level with n i elements introduces
n i ⫺1
an additional 2 subsets. We note for the special case where n i ⫽ 1, then 21 ⫺
1 ⫽ 1.
In the earlier part of the paper we suggested the handling of ties in inducing
variables by ordering the tied elements by their argument value. We shall now
justify this approach. Specifically we shall show that the result obtained using a
limited FI aggregation in which the limiting function ␭ is induced by a weak


A more formal expression of this concept can be made if we represent P as a binary
relation R p such that R p ( A i , A j ) ⫽ 1 if A i ⱖ P A j and denote TOT( A i ) ⫽ ¥ j R p ( A i , A j ). A
subset E is called a rooted subordering of P if E contains all A j such that TOT( A j ) ⬎ TOT( A i* ),
where TOT( A i* ) ⫽ MinA k僆E TOT( A k ).
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1061
ordering is the same as the result obtained using a limited FI aggregation in which
the limited function is induced by a linear ordering, which is the same except the
tied elements are lexicographically ordered by their argument value.
Before proceeding we shall find the following terminology useful. Let P be a
weak ordering on a space Ꮽ. We shall say that P̂ is a linear ordering adjudicated
from P if the ordering between the untied elements are the same and the tied
elements are linear ordered lexicographically by some adjudication procedure. We
shall say that ␭ is a limiting function induced from a weak ordering.§ P if ␭ (E) ⫽
t m for all subsets that are rooted suborderings of P and ␭ (E) ⫽ t 0 if E is not a
rooted subordering.

THEOREM. Let P be a weak ordering on Ꮽ and let P̂ be a linear ordering


adjudicated from P. Let ␭ and ␭ˆ be their respective limiting functions. Then
Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ Agg␮ 兩 ␭ˆ 共a1 , . . . , an 兲
if P̂ is adjudicated by ordering the tied elements in P in descending order of their
argument value.

Proof. We recall Agg␮ 兩 ␭ (a 1 , a 2 , . . . , a n ) ⫽ MaxE債Ꮽ [( ␮ (E) ∧ ␭ (E) ∧ Q(E))],


where Q(E) ⫽ MinA i僆E [a i ]. The collection of rooted subsets, those that have
␭ (E) ⫽ t m , can be partitioned into three classes**: Those with maximal element
are higher P than the tied elements, C1; those with maximal element are one of the
tied values, C2: those with maximal element lower than the tied value, C3. From
this we get Agg␮ 兩 ␭ (a 1 , . . . , a n ) ⫽ ∨ i⫽1 3
[∨ E僆C i ( ␮ (E) ∧ Q(E))]
Consider now Agg␮ 兩 ␭ˆ (a 1 , . . . , a n ) ⫽ MaxE債Ꮽ [( ␮ (E) ∧ ␭ˆ (E) ∧ Q(E))]
The collection of rooted subsets associated with P̂, those with ␭ˆ (E) ⫽ t m , can also
be partitioned into three classes: those with maximal elements higher than the tied
elements, Ĉ1; those with maximal elements equal to the tied elements, Ĉ2; and
those with maximal element is lower than the ties, Ĉ3. Using this, we get
Agg␮ 兩 ␭ˆ (a 1 , . . . , a n ) ⫽ ∨i⫽1
3
[∨ E僆Ĉ i ( ␮ (E) ∧ Q(E))].
First we note that Ĉ1 ⫽ C1 and Ĉ3 ⫽ C3; hence the only difference between
Agg␮兩␭ˆ and Agg␮兩␭ is in regard to the of the classes C2 and Ĉ2. Thus all we need
compare are ∨ E僆C 2 ( ␮ (E) ∧ Q(E)) and ∨ E僆Ĉ 2 ( ␮ (E) ∧ Q(E)).
We shall let G* indicate the subset of Ꮽ consisting of objects higher in the
ordering P then the tied elements and let g* ⫽ Q(G*) ⫽ MinA i僆G* [a i ].
In the following we shall denote the m tied objects as T t j, j ⫽ 1–m, where
a t j ⱖ a t j if t j ⬍ t j ; thus the t j are ordered in descending order of argument value.
Since the tied elements have been adjudicated in ␭ˆ by descending value of
argument value, we see that for any E 僆 Ĉ2 we have E ⫽ G* 艛 {A t k兩k ⫽ 1 to
j}. Let us denote this as G* 艛 T j . Furthermore, MinA ti僆T j[a t i] ⫽ a t j. Thus we get
∨ E僆Ĉ 2 ( ␮ (E) ∧ Q(E)) ⫽ ∨ j⫽1 m
( ␮ (G* 艛 T j ) ∧ ( g* ∧ a t j). Consider now the term

§
Here we note that a linear ordering is a special case of weak ordering and hence the same
can be done with linear ordering.
**For simplicity we assume only one group of tied elements the following analysis can
easily be extended to multiple groups of tied elements.
1062 YAGER

∨ E僆C 2 ( ␮ (E) ∧ Q(E)). We note that any element in C2 is of the form E ⫽ G*


艛 M, where M is some not empty subset of the tied elements; if m elements are
tied, then they are 2 m ⫺ 1 subsets. We see that Q(E) ⫽ Q(G*) ∧ Q(M) ⫽ g*
∧ Q(M). We further note that Q(M) is going to be equal to a t j, where t j is the
largest index in M. Using this, we can partition ∨ E僆C 2 ( ␮ (E) ∧ Q(E)) as

冉 冊
m

∨ 共␮共E兲∧Q共E兲兲 ⫽ ∨ ∨ 共␮共G* 艛 M兲∧共g*∧at j兲兲


E僆C 2 j⫽1 M僆Ᏺ j

where Ᏺ j is the collection of subsets of the tied elements that has A t j as its object
with the highest index. What is clear is that any M 僆 Ᏺ j also has M 債 T j . From
this it follows that:
∨ 共␮共E兲∧Q共E兲兲 ⫽ ∨ 共␮共E兲∧Q共E兲兲
ˆ2
E僆C E僆C 2

7. INDUCED FI OPERATOR AND THE


NEAREST NEIGHBOR METHOD
In the following we provide an application of the induced fuzzy integral to the
nearest neighbor aggregation. Consider a situation in which we have some object
of interest and we want to determine the degree to which this object has some
property, denoted the notable property. In order to determine this degree we shall
use a collection Ꮽ of other objects which we call the prototype objects. For each
prototype we have a value a i indicating the degree to which it has the notable
property. Furthermore, we shall assume the existence of a fuzzy measure ␮ on Ꮽ
such that, for each subset E of Ꮽ, ␮ (E) indicates the degree of credibility
associated with using this subset to determine the notable feature. This measure can
be seen as reflecting some indication of the quality of the information about the
prototype objects. The information described so far, the a i ’s and the fuzzy measure,
provides no information characterizing the object of interest. In order to provide
this information, we assume we have, for each prototype, an assessment v i of how
similar it is to the object of interest. Clearly not all prototype should be used
equivalently; the similarity between the prototype objects and the object of interest
should be used to determine the relevancy of the individual prototypes to the
calculation. In order to decide which prototypes should be used to determine the
missing value we shall use the nearest (nearer) neighbor principle. The nearest
neighbor principle says
Don’t use a prototype to determine the value unless
you also use all the prototypes that are closer
Using the preceding information and this principle we can formulate the
calculation of the degree to which the object of interest has the notable feature as
Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ Max关␭共E兲∧␮共E兲∧Q共E兲兴.
E債Ꮽ
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1063
Here the limiting function ␭ is generated by the nearest neighbor principle and the
proximity ordering of the prototypes. Using the similarity of the object to the
prototypes v i , we can induce an ordering over Ꮽ such that v-index( j) is the index
the jth nearest neighbor. This ordering in turn induces a limiting function ␭ (E)
such that ␭ (E) ⫽ t m iff E ⫽ G j ⫽ {A v-index(K) 兩k ⫽ 1 to j} otherwise ␭ (E) ⫽ t 0 .
Using this limiting function, we get:

冉 冊
n n j

Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ ∨ 关␮共Gj 兲∧Q共Gj 兲兴 ⫽ ∨ ␮共Gj 兲∧ ∧ av-index共k兲


j⫽1 j⫽1 k⫽1

We see this is the induced FI aggregation operator. Thus the nearest neighbor
method can be seen as a type of I-FI aggregation.
We now look at some specific cases of nearest neighbor rule. These special
cases are determined by the structure of ␮. First we shall consider the class of
cardinality based measures. For this class we let ␮ (G j ) ⫽ w j , where w j just
depends on j and w j ⱖ w i if j ⬎ i. If we consider the specific cardinality based
measure where w j ⫽ t m for all j, then:

冉 冊
n j

Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ ∨ ∧ av-index共k兲 ⫽ av-index共1兲


j⫽1 k⫽1

It is the degree to which the nearest neighboring prototype has the notable property.
We see this as corresponding to the simple strict nearest neighbor rule.
If we consider the case where w j ⫽ t 0 for j ⫽ 1 to p ⫺ 1 and w j ⫽ t m for
j ⫽ p to n, then:

冉 冊
n j p

Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ ∨ ∧ av-index共k兲 ⫽ ∧ av-index共k兲


j⫽p k⫽1 k⫽1

It is the minimal degree to which any of the p nearest neighbor prototypes have the
property. This can be seen as a kind of pth nearest neighbor rule.
Another class of fuzzy measures occurs when we associate with each proto-
type A i a degree of credibility ␣ i and assume the ␮ (E) ⫽ MaxA i僆E [ ␣ i ], ␮ is a
possibility measure. In this case:

冉冉 冊冉 冊冊
n n n

Agg␮ 兩 ␭ 共a1 , . . . , an 兲 ⫽ ∨ ∨ ␣v-index共k兲 ∧ ∧ av-index共k兲


j⫽1 k⫽1 k⫽1

Other examples of nearest neighbor rules can be obtained by different specifica-


tions of ␮.
A generalization of the methodology just used can be made to provide what
we shall call Prototype Based Reasoning (PBR). This technology can be used to
determine the degree to which some Object of Interest (OI) has some Notable
Feature (NF). Let Ꮽ ⫽ {A 1 , . . . , A n } be a collection of objects which we shall
call standards or prototypes. Assume we have associated with each prototype a
value a i indicating the degree to which it has the notable feature (NF). In addition,
we assume the existence of a fuzzy measure ␮ over the set of prototypes indicating
1064 YAGER

the degree of credibility associated with using a particular subset of prototypes to


determine the notable feature; this indicates a kind of weight we assign to the
“opinion” of this group. In order to determine the relevance of the prototypes to the
current object of interest, we assume the existence of some function ␭: 2 Ꮽ 3 T
such that ␭ (E) indicates the relevance of the subset E of prototypes to the current
object of interest. Under the PBR paradigm we say that the degree to which the OI
has the notable feature (satisfies a condition) is the degree to which
there exists a subset of relevant credible prototypes
having the notable feature.
Under this paradigm, we are interested in finding a subset of prototypes that are
Relevant and Credible and Have the Feature
We can formally express this degree as MaxE債Ꮽ [ ␭ (E) ∧ ␮ (E) ∧ Q(E)], where
Q(E) ⫽ MinA i僆E [a i ]. Different examples of PBR systems are manifested by the
structure of the fuzzy measure ␮ and the methodology used to generate the
relevancy function from the similarity of the prototypes and the object of interest.

8. CONCLUSION
We discussed the fuzzy measure and the related idea of fuzzy integral. We
pointed out the centrality within the implementation of fuzzy integral aggregation
of the ordering operation based upon the arguments to be aggregated. We then
extended the fuzzy integral aggregation operator by allowing the ordering opera-
tion to be based upon values other then those being aggregated. This lead us to the
induced fuzzy integral aggregation operator. We looked at this operator and studied
its properties. We showed its relationship to a formulation called limited fuzzy
integral aggregation. It was shown how this new induced fuzzy integral operator
provides a natural framework for the implementation of nearest neighbor rules. We
introduced the idea of prototype based reasoning. Throughout this work the ordinal
nature of fuzzy integral was exploited to enable us only to use data drawn from a
ordinal scale; this situation makes the techniques developed here particularly
suitable for application within the scope of Zadeh’s approach to computing
with words.

References
1. Sugeno M. Fuzzy measures and fuzzy integrals: A survey. In: Gupta MM, Saridis GN,
Gaines BR, editors. Fuzzy automata and decision process. Amsterdam: North-Holland;
1977. pp 89 –102.
2. Murofushi T, Sugeno M. Fuzzy measures and fuzzy integrals. In: Grabisch M, Murofushi
T, Sugeno M, editors. Fuzzy measures and integrals. Heidelberg: Physica-Verlag; 1999.
pp. 3– 41.
3. Zadeh LA. Fuzzy logic ⫽ computing with words. IEEE Trans Fuzzy Syst 1996;4:103–111.
4. Zadeh LA. From computing with numbers to computing with words—from manipu-
INDUCED FUZZY INTEGRAL AGGREGATION OPERATOR 1065
lation of measurements to manipulations of perceptions. IEEE Trans Circuits Syst
1999;45:105–119.
5. Yager RR. A new methodology for ordinal multiple aspect decisions based on fuzzy sets.
Decision Sci 1981;12:589 – 600.
6. Grabisch M, Murofushi T, Sugeno M. Fuzzy measures and integrals. Heidelberg: Physica-
Verlag; 1999.
7. Yager RR. A general approach to criteria aggregation using fuzzy measures. Int J Man-
Machine Stud 1993;38:187–213.
8. Zadeh LA. Fuzzy sets as a basis for a theory of possibility. Fuzzy Sets Syst 1978;1:3–28.
9. Dubois D, Prade H. Possibility theory: An approach to computerized processing of
uncertainty. New York: Plenum Press; 1988.
10. Zadeh LA. Fuzzy sets and information granularity. In: Gupta MM, Ragade RK, Yager RR,
editors. Advances in fuzzy set theory and applications. Amsterdam: North-Holland; 1979.
pp 3–18.
11. Sugeno M. Theory of fuzzy integrals and its application. Doctoral Thesis, Tokyo Institute
of Technology, 1974.
12. Grabisch M. Fuzzy integral in multicriteria decision making. Fuzzy Sets Syst 1995;69:
279 –298.
13. Marichal JL. On Sugeno integral as an aggregation function. Fuzzy Sets Syst 2000;114:
347–365.
14. Yager RR, Filev DP. Induced ordered weighted averaging operators. IEEE Trans Syst Man
Cybernet 1999;29:141–150.
15. Yager RR. Choquet aggregation using order inducing variables. Technical Report# MII-
2109 Machine Intelligence Institute, Iona College, New Rochelle, NY, 2001.
16. Yager RR. Quantifiers in the formation of multiple objective decision functions. Inf Sci
1983;31:107–139.
17. Yager RR. General multiple objective decision making and linguistically quantified state-
ments. Int J Man-Mach Stud 1984;21:389 – 400.
18. Sugeno M. Theory of fuzzy integrals and its application. Doctoral Thesis, Tokyo Institute
of Technology, 1974.

You might also like