Elsevier

Information Sciences

Volume 248, 1 November 2013, Pages 68-88
Information Sciences

Interval-valued intuitionistic fuzzy implications – Construction, properties and representability

https://doi.org/10.1016/j.ins.2013.06.020Get rights and content

Abstract

Firstly, this work studies the class of representable (co)implications obtained by idempotent aggregations and pair of dual interval functions, namely fuzzy implications and coimplications. Following the same construction, as the main contribution in the context of the interval-valued intuitionistic fuzzy logic, which is conceived by Atanassov, the class of representable Atanassov’s intuitionistic fuzzy implications is obtained by composition of idempotent interval aggregations and dual pairs of representable fuzzy implications and coimplications. Additionally, the conditions under which relevant properties of fuzzy implications and Atanassov’s intuitionistic fuzzy implications are preserved by such constructions are investigated. Furthermore, taking into account the projection functions and related (interval-valued) Atanassov’s intuitionistic fuzzy implications, it also shows that representable (interval-valued) Atanassov’s intuitionistic fuzzy implications preserve (degenerate) diagonal elements.

Introduction

Since [33], [54], [63], [64], such and many other extensions of Fuzzy Logic (FL) have been studied (see also [30], [44]) by providing logical interpretation of information obtained by fuzzy connectives and by reducing the uncertainty and the imprecision from the membership grades.

This work focuses on the integration of two mathematically equivalent but semantically different approaches: (i) Interval-Valued Fuzzy Logic (IVFL), dealing with representations in terms of fuzzy connectives on U = [0, 1] and (ii) Intuitionistic Fuzzy Logic (IFL), here based on Atanassov’s intuitionistic fuzzy sets introduced in [2], [6], which were defined as a pair of membership functions by relaxing the complementary operation stated over fuzzy sets in FL.

In the former, we obtained the narrowest interval containing the range of such fuzzy connectives together with the correctness of interval computations based on such interval-valued fuzzy connectives, as pointed out by Hickey et al. in [34]. Moreover, the measure of vagueness and uncertainty is reflected in the diameter of an interval membership degree. On the other hand, the latter considers not only to what degree an element belongs to a particular set defined by an Atanassov’s intuitionistic fuzzy connective but also to what degree this element does not belong to the corresponding set. Thus, the hesitation related to the dual construction is modelled by an inequality, i.e., the difference between an interval membership degree and its interval non-membership degree is less or equal to the degenerate interval [1, 1].

As proposed by Atanassov and Gargov [7], in 1989, interval-valued Atanassov’s intuitionistic fuzzy sets (IVIFSs) extend the Atanassov’s intuitionistic fuzzy sets (IFSs) by considering a membership function and a non-membership function whose values, which are not necessarily complementary, can be given as intervals rather than real numbers. Based on the Atanassov’s IVIFSs, we consider the Interval-valued Atanassov’s Intuitionistic Fuzzy Logic (IVIFL) integrating concepts from IVFL and IFL as representability of aggregation and dual Atanassov’s intuitionistic fuzzy connectives.

In such logical context, following previous works, see [13], [52], this paper considers the canonical representation [55] and idempotent aggregation class, in order to map Atanassov’s intuitionistic fuzzy connective class into interval-valued Atanassov’s intuitionistic fuzzy connective class. Thus, as the main contribution, the paper introduces the class of interval-valued Atanassov’s intuitionistic fuzzy implications (IVI-implications) obtained by representable idempotent aggregation functions, interval-valued implications (IV-implications) and their dual constructions, the interval-valued fuzzy coimplications (IV-coimplications). By applying projection functions and considering the sets of diagonal elements and degenerate intervals the main properties of fuzzy implications are preserved by IVI-implications. Additionally, the properties from the Atanassov’s intuitionistic approach related to Atanassov’s intuitionistic index are also investigated.

Our motivation is to contribute with the study of IVIFIs as foundations to inference systems in the context of IVIFL [5]. In [3], [5], [49], different operators are defined over the interval valued Atanassov’s intuitionistic fuzzy sets and their basic algebraic properties are studied. Such research area has received much attention from theoretical [4], [38], [39], [47] and applied approach in the fields of multi-criteria decision [20], [35], [61], [67], group decision [48], [60], [65], computational game theory [38] and aggregation functions [47], [46], [62], modelling of fuzzy-valued information systems [36], [62], [65], [66].

In order to consider different models to the membership degree and to the non-membership degree in systems based on IVFL, theoretical and applied research are frequently based on aggregation functions, which play an important role in statistics, regression analysis, pattern recognition, decision sciences and image processing, mainly reported in [58]. So, several aggregation function classes have been studied in detail in the fuzzy literature [19].

In [16], an Atanassov’s intuitionistic fuzzy implication (IFI) operator recovers J. Fodor’s definition [31] of a fuzzy implication operator when the sets are fuzzy, generalizing the Atanassov’s operator [4]. Additionally, by [21], [22], [23], the definition of fuzzy connectives are undertaken by such theories. In [27], a new implication class obtained by binary aggregation operators based on t-norms on the unit interval is introduced. In [29], aggregation functions integrating concepts of interval-valued fuzzy and Atanassov’s intuitionistic fuzzy set theory were discussed. In [19], generalized Atanassov’s operators are studied, including the OWA aggregation operators. Recently, in [15], the media operator is defined by aggregation functions whose values depend on central inputs.

Following such approaches, we make use of interval-valued idempotent aggregation functions in order to define representable (co)implications and representable Atanassov’s intuitionistic fuzzy implications, extending the work proposed in [16]. Such extension may contribute to interpret the truth and non-truth of conditional rules in inference systems based on IVFL and IVIFL, respectively.

The paper presents an N-dual structure of an (I,IN,N,MA)-representable implication generated from a set M of aggregation functions and mutual dual fuzzy implications I and IN, based on the isomorphism between U2 and L by mapping the notions presented in [16, Definition 3] to the canonical representation [14]. We discuss under which conditions the (I,IN,N,MA)-representable implications preserve the main properties of interval implications, showing the applicability of N-dual interval approach for representable implications.

Another contribution provides two equivalent characterizations of an IVI-implication II. On one hand, (I,IN,N,MA)-representable Atanassov’s intuitionistic implication which is generated by a set MA of representable idempotent IV-aggregation functions and a pair of an IV-implication and its corresponding dual IV-coimplication. It also discusses some properties of (I,IN,N,M)-representable IVI implications. On the other hand, IVI-implications generated by (I,IN,N,MA)-representable IV-implications. Thus, the conditions under which pairs of degenerate intervals are preserved by (I,IN,N,MA)-representable IVI-implications are also studied.

The paper is organized as follows. Section 2 contains preliminary definitions, discussing the notions of interval representations of real functions and presenting definitions and several results concerned with fuzzy negations, duality relationship and aggregation operators. The basic concepts of fuzzy (co)implications are studied in Section 3, including main properties and some examples.

Section 4 reports definitions of Atanassov’s intuitionistic fuzzy connectives, focusing on the study of Atanassov’s intuitionistic fuzzy implications generated by aggregation functions and a pair of an implication and its corresponding dual coimplication, obtained by a strong fuzzy negation, as conceived in [16]. Interval-valued (co)implications are studied in Section 5, where firstly, some results related to the interval extensions of fuzzy (co)implications complement previous works, see e.g. [12], [13], [14], extending their main properties from FL approach.

Thus, in Section 6, this paper presents a study of IV-(co)implications generated by IV-aggregation functions and by mutual dual fuzzy (co)implications, proving that several analogous properties of fuzzy (co)implications are also held for such representable (co)implications.

In Section 7, interval-valued Atanassov’s intuitionistic fuzzy implications generated by IV-aggregation functions and a pair of an IV-implication and corresponding IV-coimplication are discussed. In addition, such construction can also be obtained by representable implications.

We conclude by pointing out the main results of this paper, the ongoing work and some final remarks.

Section snippets

Interval representations

Consider the real unit interval U=[0,1]R and let U be the set of subintervals of U, that is, U={X=[a,b]:0ab1}. The interval set has two projections l,r:UU, defined by l(X) = l([a, b]) = a and r(X) = r([a, b]) = b and denoted by X and X¯, respectively.

In the following, we report the component-wise Kulisch–Miranker partial order (also called product order), defined by:X,YU:XUYX̲Y̲X¯Y¯.

In addition, we also consider UU2 as the binary relation given by:X,YU:XUYX¯Y̲,which was based in the

Fuzzy (co)implications

Coimplication functions are conceived as N-dual structures of implication functions extending the classical (co)implications [9], [18], [31], [42]:

Definition 3.1

56, Section 3.1

The binary operator (J)I: [0, 1]2  [0, 1] is a fuzzy (co)implication if it satisfies the boundary conditions:I1:I(1,1)=I(0,1)=I(0,0)=1,I(1,0)=0;J1:J(1,1)=J(1,0)=J(0,0)=0,J(0,1)=1.

Since Definition 3.1 imposes a very weak condition for a binary function to be a fuzzy (co)implication, several other extra properties are considered in the literature, in

Atanassov’s intuitionistic fuzzy (co)implications

A formal treatment of Atanassov’s intuitionistic fuzzy sets are studied in [6], [22], [23], [26], [31]. In this section, firstly, some basic concepts are introduced in order to study Atanassov’s intuitionistic fuzzy implications generated by aggregation functions, following the work of Bustince at. al. in [16].

Let U={x̃=(x1,x2):(x1,x2)U2,x1+x21} be the set of all Atanassov’s intuitionistic fuzzy membership degrees. For all x̃U, it holds that:

  • (i)

    0̃=(0,1)Ux̃ and 1̃=(1,0)Ux̃; and

  • (ii)

    x̃Uỹx1y1

Interval-valued (co)implications

Since real numbers may be identified with degenerate intervals in the context of interval mathematics, the boundary conditions which must be satisfied by the classical fuzzy implications can be naturally extended to interval fuzzy degrees, whenever degenerate intervals are considered, see e.g. [32]. The IV-(co)implications are considered, grounded by the canonical representation.

Definition 5.1

The binary function I(J):U2U is called an IV-(co)implication iff it satisfies the boundary conditions given by

  • I1:

    I(1,1)=

Representable IV-(co)implications generated from aggregations and pairs of dual operators

An expression for IV-(co)implication generated from aggregation functions and mutual dual fuzzy (co)implications is studied in this section, including the presentation of constraints assuring that such interval functions are well defined and may extend the canonical representation in the sense of [14]. Such study extends the previous results presented in [50].

Definition 6.1

Let I be an implication, J be a coimplication, N be a fuzzy negation and MA be a set of binary idempotent aggregation functions. Then, I(I

Interval-valued Atanassov’s intuitionistic fuzzy implications

Based on the work introduced in [6] and later in [22], this section briefly studies the interval-valued Atanassov’s intuitionistic fuzzy sets, their main concepts and properties. Thus, IVI-implications generated by (I,IN,N,MA)-representable IV-implications are defined. In the following, IVI-implications generated by aggregation functions and pairs of mutual dual IV-implications are discussed, showing the conditions under which such constructions are equivalent.

Let U={X=(X1,X2):(X1,X2)U2andX1+

Conclusion and final remarks

Interval-valued Atanassov’s intuitionistic fuzzy sets have been studied as a natural generalization of Atanassov’s intuitionistic fuzzy sets modelling the uncertainty due to the lack of information and the hesitation in the definition of a membership and non-membership function. Dealing with representable fuzzy implication generated from aggregation and dual interval operators, the paper introduces a representable Atanassov’s intuitionistic fuzzy implication extending the definition introduced

Acknowledgments

This work is supported by CNPq (under the Process Numbers 480832/2011-0 and 307681/2012-2) and FAPERGS (under the Process Number 11/1520-1 of Edital PqG 02/2011).

References (67)

  • G. Deschrijver

    A representation of t-norms in interval-valued L-fuzzy set theory

    Fuzzy Sets and Systems

    (2008)
  • G. Deschrijver et al.

    Implicators based on binary aggregation operators in interval-valued fuzzy set theory

    Fuzzy Sets and Systems

    (2005)
  • D. Dubois et al.

    Random sets and fuzzy interval analysis

    Fuzzy Sets and Systems

    (1991)
  • S. Gottwald

    Set theory for fuzzy sets of higher level

    Fuzzy Sets and Systems

    (1979)
  • D.H. Hong

    A note on correlation of interval-valued intuitionistic fuzzy sets

    Fuzzy Sets and Systems

    (1998)
  • B. Huang et al.

    Using a rough set model to extract rules in dominance-based interval-valued intuitionistic fuzzy information systems

    Information Sciences

    (2013)
  • D. Paternain et al.

    An alternative to fuzzy methods in decision-making problems

    Expert Systems with Applications

    (2012)
  • B. Pekala

    Properties of Atanassov’s intuitionistic fuzzy relations and Atanassov’s operators

    Information Sciences

    (2012)
  • Y. Shi et al.

    On the characterizations of fuzzy implications satisfying I(x, y) = I(x, I(x, y))

    Information Sciences

    (2007)
  • C. Wei et al.

    Entropy, similarity measure of interval-valued intuitionistic fuzzy sets and their applications

    Information Sciences

    (2011)
  • J. Wu et al.

    Intuitionistic fuzzy-valued Choquet integral and its application in multicriteria decision making

    Information Sciences

    (2013)
  • Z. Xu

    A method based on distance measure for interval-valued intuitionistic fuzzy group decision making

    Information Sciences

    (2010)
  • L.A. Zadeh

    The concept of a linguistic variable and its application to approximate reasoning-I

    Information Sciences

    (1975)
  • Q. Zhang et al.

    Some information measures for interval-valued intuitionistic fuzzy sets

    Information Sciences

    (2010)
  • B.M. Acióly et al.

    A quasi-metric topology compatible with inclusion monotonicity on interval space

    Reliable Computing

    (1997)
  • K.H. Atanassov

    Intuitionistic Fuzzy Sets

    (1999)
  • K.H. Atanassov

    Intuitionistic fuzzy implications and modus ponens

    Academy of Sciences

    (2005)
  • K.H. Atanassov et al.

    Elements of intuitionistic fuzzy logic – Part I

    Fuzzy Sets and Systems

    (1989)
  • M. Baczyński et al.

    Fuzzy Implications

    (2008)
  • B. De Baets

    Coimplications, the forgotten connectives

    Tatra Mountains Mathematical Publications

    (1997)
  • E. Barrenechea, H. Bustince, M. Pagola, J. Fernández, J. Sanz, Generalized Atanassov’s intuitionistic fuzzy index....
  • G. Beliakov et al.

    Aggregation for Atanassov’s intuitionistic and interval valued fuzzy sets: the median operator

    IEEE Transactions on Fuzzy Systems

    (2012)
  • H. Bustince et al.

    Intuitionistic fuzzy implication operators – an expression and main properties

    Intl. Journal of Uncertainty Fuzziness and Knowledge-Based Systems

    (2004)
  • Cited by (24)

    • Bi-aggregated contrapositivisation: A new contrapositivisation technique for fuzzy implications

      2023, Fuzzy Sets and Systems
      Citation Excerpt :

      In addition, will intend to investigate the notion of contrapositivisation for the main extensions of fuzzy logic [9,10,27,34,37,43,47,48] and study the contrapositivisation of implications that satisfy the distributivity of Combs and Andrews [16] and other kinds of distributivity involving implication such as in Cruz et al. [17] and Balasubramaniam and Rao [7].

    • N Approach Driven Ranking System for Risky Gaits

      2022, Expert Systems with Applications
      Citation Excerpt :

      An extension of intuitionistic fuzzy sets in the interval domain was introduced by Atanassov (1999a) where the interval-valued intuitionistic fuzzy sets (IVIFS) express both the membership and non-membership degrees in the interval forms, but not in the exact values. These are effective tools for researchers when inexact intervals are required to deal with uncertainty (Reiser & Bedregal, 2013). In the literature, multi-criteria decision making techniques and models were also presented by some authors using IVIFS (Hajiagha et al., 2015; Wan & Dong, 2015; Wu & Chiclana, 2012).

    • A projection-based compromising method for multiple criteria decision analysis with interval-valued intuitionistic fuzzy information

      2016, Applied Soft Computing Journal
      Citation Excerpt :

      IVIF sets are characterized by the degree of membership, the degree of non-membership, and the degree of non-determinacy whose values are intervals rather than exact numbers [3,4]. Accordingly, IVIF sets are better than intuitionistic fuzzy sets for their great ability to handle imprecise and ambiguous information in practical applications [8,11,12,38]. Real-world decisions always require the use of more precise and accurate data [20]; thus, IVIF sets are appropriate for addressing MCDA problems within a complex and uncertain environment.

    • An interval-valued intuitionistic fuzzy permutation method with likelihood-based preference functions and its application to multiple criteria decision analysis

      2016, Applied Soft Computing Journal
      Citation Excerpt :

      The concept of IVIF sets is characterized by a membership function and a non-membership function whose values are intervals rather than exact numbers. The IVIF sets are better than IF sets due to their great ability to handle imprecise and ambiguous information in practical applications [5,7,9,10]. Real-world decisions often require the use of more precise and accurate data [70].

    • Multicriteria Pythagorean fuzzy decision analysis: A hierarchical QUALIFLEX approach with the closeness index-based ranking methods

      2016, Information Sciences
      Citation Excerpt :

      For this purpose, we further extend the PFS to propose the concept of interval-valued PFS (IVPFS) which is parallel to interval-valued intuitionistic fuzzy set (IVIFS) [4]. To our knowledge, IVIFSs have been successfully applied in various fields [10–13,15,21–22,34,38,40–42,45–46,50]. As an extension of IVIFSs from the perspective of the constraint condition, IVPFSs also have wide application fields, especially in MCDM fields such as the selection of suppliers in supplier chain management, evaluation of goods, etc.

    View all citing articles on Scopus
    View full text