actas del iii simposio sobre lógica fuzzy y soft computing...

Actas del III Simposio sobre Lógica Fuzzy y Soft Computing, LFSC2010 (EUSFLAT)

Editores: Luis Martínez, Edurne Barrenechea, Macarena Espinilla Jesús Alcalá, Victoria López, Manuel Mucientes, José Ángel Olivas, Rosa Mª Rodríguez ISBN: 978-84-92812-65-3 IBERGARCETA PUBLICACIONES, S.L., Madrid, 2010

Edición: 1ª Impresión: 1ª Nº de páginas: 494 Formato: 17 x 24 Materia CDU: 004 Ciencia y tecnología de los ordenadores. Informática

Reservados los derechos para todos los países de lengua española. De conformidad con lo dispuesto en el artículo 270 y siguientes del código penal vigente, podrán ser castigados con penas de multa y privaci6n de libertad quienes reprodujeren o plagiaren, en todo o en parte, una obra literaria, artística o científica fijada en cualquier tipo de soporte sin la preceptiva autorización. Ninguna parte de esta publicación, incluido el diseño de la cubierta, puede ser reproducida, almacenada o trasmitida de ninguna forma, ni por ningún medio, sea éste electrónico, químico, mecánico, e1ectro-óptico, grabación, fotocopia o cualquier otro, sin la previa autorización escrita por parte de la editorial.

Diríjase a CEDRO (Centro Español de Derechos Reprográficos), www.cedro.org, si necesita fotocopiar o escanear algún fragmento de esta obra.

COPYRIGHT © 2010 IBERGARCETA PUBLICACIONES, S.L. [email protected] Actas del III Simposio sobre Lógica Fuzzy y Soft Computing, LFSC2010 (EUSFLAT) Derechos reservados ©2010 respecto a la primera edición en español, por LOS AUTORES Derechos reservados ©2010 respecto a la primera edición en español, por IBERGARCETA PUBLICACIONES, S.L. 1ª Edición, 1ª Impresión

ISBN: 978-84-92812-65-3

Depósito legal: M- Maquetación: Los Editores

Coordinación del proyecto: @LIBROTEX

Portada: Estudio Dixi Impresión y encuadernación: OI: 28/2010 PRINT HOUSE, S.A. IMPRESO EN ESPAÑA -PRINTED IN SPAIN

Nota sobre enlaces a páginas web ajenas: Este libro puede incluir referencias a sitios web gestionados por terceros y ajenos a IBERGARCETA PUBLICACIONES, S.L., que se incluyen sólo con finalidad informativa. IBERGARCETA PUBLICACIONES, S.L., no asume ningún tipo de responsabilidad por los daños y perjuicios derivados del uso de los datos personales que pueda hacer un tercero encargado del mantenimiento de las páginas web ajenas a IBERGARCETA PUBLICACIONES, S.L., y del funcionamiento, accesibilidad y mantenimiento de los sitios web no gestionados por IBERGARCETA PUBLICACIONES, S.L., directamente. Las referencias se proporcionan en el estado en que se encuentran en el momento de publicación sin garantías expresas o implícitas, sobre la información que se proporcione en ellas.

Contenido

III Simposio sobre Lógica Fuzzy y Soft Computing, LFSC2010 (EUSFLAT)

SESIÓN ESPECIAL 1 Aplicaciones del Soft Computing Organizador: Manuel Mucientes

Cálculo del Grado de la Normalidad en Sistemas Expertos de Vídeo-Vigilancia Escalables........................................................................................................................ 3

Javier Albusac, David Vallejo, J. J. Castro-Schez, L. Jiménez-Linares

Localización probabilística del humanoide Nao en el campo de la Robocup............... 11

Francisco J. Rdez. Lera, Juan F. García, Camino Fdez. Llamas, Vicente Matellán

Reconstrucción volumétrica de personas utilizando la teoría de Dempster-Shafer....... 19

Luis Díaz, Rafael Muñoz, Enrique Yeguas, Rafael Medina

Expert Fuzzy System for Prediction of Pedestrian Accidents......................................... 27

J. L. Castro, M. Delgado, J. Medina, M. D. Ruiz-Lozano

Un Modelo Evolutivo para el Diseño de Materiales Compuestos Laminados Simétricos........................................................................................................................ 35

E. Tenorio, J. I. Peláez, J. M. Doña, J. Gómez Ruiz

Localización de Olivos en Imágenes de Alta Resolución mediante Técnicas de Razonamiento Aproximado............................................................................................. 41

J. Moreno-Garcia, L. Jiménez, L. Rodriguez-Benítez, C. Solana-Cipres

SESIÓN ESPECIAL 2 Toma de Decisiones en Ambientes de Incertidumbre Organizadoras: Macarena Espinilla y Rocío de Andrés

A new mobile decision support system to manage dynamic decisión situations............ 51

I. J. Pérez, F. J. Cabrerizo, S. Alonso, E. Herrera-Viedma

Modelo Lingüístico de QoS para Servicios de Red........................................................ 59

Sergio Gramajo, Luis Martínez

Aplicación Web para Evaluación de Desempeño 360 Grados Basada en un Modelo Lingüístico Multigranular................................................................................................ 67

A new mobile decision support system to manage dynamic decisionsituations

I.J. Pérez F. J. Cabrerizo S. Alonso E. Herrera-ViedmaComputer Science and A.I. Software Engineering Software Engineering Computer Science and A.I.

Univ. of Granada UNED Univ. of Granada Univ. of Granada

[email protected] [email protected] [email protected] [email protected]

Abstract

The aim of this contribution is to presenta new prototype of decision support systembased on mobile technologies and dynamic in-formation. There are a large number of scenar-ios in which the deployment of decision sup-port systems on mobile devices is desirable.So, users can run the system on their own mo-bile devices in order to provide their prefer-ences at anytime and anywhere. The systemincorporates a mechanism that allows to man-age dynamic decision situations in which someinformation about the problem is not constantthrough the time, it gives more realism to de-cision processes with high or dynamic set ofalternatives. In this prototype we allow theexperts to use linguistic preference relationsto express their preferences.

1 Introduction

Group Decision Making (GDM) arises frommany real world situations. Thus, the studyof decision making is necessary and importantnot only in Decision Theory but also in areassuch as Management Science, Operations Re-search, Politics, Social Psychology, ArtificialIntelligence, Soft Computing, and so on. Inthese situations, there is a problem that canbe solved in different ways and a group of ex-perts trying to achieve a consensual solution.To do this, experts have to express their pref-erences by means of a set of assessments overa set of alternatives.

In the last years, the interaction human-

technology has had several significant ad-vances. The spread of mobile devices has in-creased accessibility to data and, in turn, in-fluenced the time and the way in which usersmake decisions. Users can make real-time de-cisions based on the most up-to-date data ac-cessed via wireless devices, such as portablecomputers, mobile phones, and personal digi-tal assistants (PDAs). So, the application ofthe latest technologies extends opportunitiesand allows to carry out consensual processeswhere previously could not be correctly ad-dressed. We assume that if the communica-tions are improved the decisions will be up-graded, because the discussion could be fo-cussed on the problem with less time wastedon unimportant issues [12, 14].

Several authors have provided interestingresults on GDM with the help of fuzzy the-ory [5, 10, 11]. There are decision situa-tions in which the experts’ preferences cannotbe assessed precisely in a quantitative formbut may be in a qualitative one, and thus,the use of a linguistic approach is necessary[3, 10, 11]. The linguistic approach is an ap-proximate technique which represents qualita-tive aspects as linguistic values by means oflinguistic variables, that is, variables whosevalues are not numbers but words or sentencesin a natural or artificial language.

In this contribution we present a prototypeof mobile decision support system (DSS) todeal automatically with dynamic GDM prob-lems based on mobile technologies. At everystage of the decision process, users, in order toreach a common solution, receive recommen-

dations to help them to change their prefer-ences and they are able to send their updatedpreferences at any moment. Additionally, tobetter simulate real decision making processes,the mobile DSS includes a tool to manage dy-namic sets of alternatives [13].

In order to do this, the paper is set outas follows. Some preliminary aspects aboutGDM models, linguistic approach and mobiletechnologies usage in GDM problems are pre-sented in Section 2. Section 3 defines the pro-totype of a mobile DSS. Finally, in Section 4we point out our conclusions.

2 Preliminaries

In this section we present some considerationsabout GDM problems, the fuzzy linguistic ap-proach and the use of mobile technologies inconsensual processes.

2.1 GDM problems

In a GDM problem we have a finite setof feasible alternatives. X = {x1, x2, . . ., xn}, (n ≥ 2) and the best alternatives fromX have to be identified according to the in-formation given by a set of experts, E ={e1, e2, . . . , em}, (m ≥ 2).

Usual resolution methods for GDM prob-lems are composed by two different processes[4, 10] (see Fig 1):

1. Consensus process: Clearly, in any deci-sion process, it is preferable that the ex-perts reach a high degree of consensus onthe solution set of alternatives. Thus, thisprocess refers to how to obtain the max-imum degree of consensus or agreementamong the experts on the solution alter-natives.

2. Selection process: This process consists inhow to obtain the solution set of alterna-tives from the opinions on the alternativesgiven by the experts. Furthermore, theselection process is composed of two dif-ferent phases:

(a) Aggregation phase: This phase usesan aggregation operator in order to

transform the individual preferenceson the alternatives into a collectivepreference.

(b) Exploitation phase: This phasetransforms the collective preferenceinto a partial ranking of alternativesthat helps to make the final decision.

Figure 1: Resolution process of a GDM prob-lem

2.2 Fuzzy linguistic approach

There are situations in which the informationcannot be assessed precisely in a quantitativeform but may be in a qualitative one. For ex-ample, when attempting to qualify phenomenarelated to human perception, we are often ledto use words in natural language instead ofnumerical values, e.g. when evaluating qual-ity of a restaurant, terms like good, mediumor bad can be used. In other cases, precisequantitative information cannot be stated be-cause either it is unavailable or the cost for itscomputation is too high and an “approximatevalue” can be applicable, eg. when evaluatingthe speed of a car, linguistic terms like fast,very fast or slow can be used instead of nu-meric values [3]. The use of Fuzzy Sets The-ory has given very good results for modellingqualitative information [15].

Fuzzy linguistic modelling is a tool basedon the concept of linguistic variable to dealwith qualitative assessments. It has proven

52 Actas del III Simposio sobre Lógica Fuzzy y Soft Computing

its usefulness in many problems, e.g., in de-cision making, quality evaluation, informa-tion retrieval models, etc. Ordinal fuzzy lin-guistic modelling [8] is a very useful kind offuzzy linguistic approach proposed as an al-ternative tool to the traditional fuzzy linguis-tic modelling which simplifies the computingwith words process as well as linguistic as-pects of problems. It is defined by consid-ering a finite and totally ordered label setS = {si}, i ∈ {0, ..., g} in the usual sense, i.e.,si ≥ sj if i ≥ j, and with odd cardinality(usually 7 or 9 labels). The mid term rep-resents an assessment of “approximately 0.5”,and the rest of the terms are placed symmetri-cally around it. The semantics of the label setis established from the ordered structure of thelabel set by considering that each label for thepair (si, sg−i) is equally informative [3]. Forexample, we can use the following set of sevenlabels to represent the linguistic information:S = { N=Null, VL=Very Low, L=Low,

M=Medium, H=High, VH=Very High,P=Perfect}.

In any linguistic model we also need somemanagement operators for linguistic informa-tion. An advantage of the ordinal fuzzy lin-guistic modeling is the simplicity and speed ofits computational model. It is based on thesymbolic computational model [8] and acts bydirect computation on labels by taking into ac-count the order of such linguistic assessmentsin the ordered structure of labels. Usually,the ordinal fuzzy linguistic model for comput-ing with words is defined by establishing i) anegation operator and ii) comparison opera-tors based on the ordered structure of linguis-tic terms. Eventually, using these operatorsit is possible to define automatic and sym-bolic aggregation operators of linguistic infor-mation, as for example the LOWA operator[9].

In a GDM problem the experts can presenttheir opinions using different elements of pref-erence representation (preference orderings,utility functions or preference relations) [5],but in this contribution, we assume that theexperts give their preferences using fuzzy lin-guistic preference relations.

A Fuzzy linguistic Preference Relation(FLPR) P i given by an expert ei is a fuzzyset defined on the product set X × X, that ischaracterized by a linguistic membership func-tion

µP i : X ×X −→ S

where the value µP i(xl, xk) = pilk is inter-

preted as the linguistic preference degree ofthe alternative xl over xk for the expert ei.

2.3 Mobile technologies usage in GDMproblems

During the last decade, organizations havemoved from face-to-face group environmentsto virtual group environments using commu-nication technology. More and more workersuse mobile devices to coordinate and share in-formation with other people. The main objec-tive is that the members of the group couldwork in an ideal way where they are, havingall the necessary information to take the rightdecisions [12, 14].

To support the new generation of decisionmakers and to add real-time process in theGDM problem field, we propose to incorpo-rate mobile technologies in a DSS obtaining aMobile DSS (MDSS). Using such a technologyshould enable a user to maximize the advan-tages and minimize the drawbacks of DSSs.

The need of a face-to-face meeting disap-pears with the use of this model, being theown computer system who acts as moderator.Experts can communicate with the system di-rectly using their mobile device from any placein the world and at any time. Hereby, a contin-uous information flow among the system andeach member of the group is produced, whichcan help to reach the consensus between theexperts on a faster way and to obtain betterdecisions.

In addition, MDSS can help to reduce thetime constraint in the decision process. Thus,the time saved by using the MDSS can be usedto do an exhaustive analysis of the problemand obtain a better problem definition. Thistime also could be used to identify more fea-sible alternative solutions to the problem, andthus, the evaluation of a large set of alterna-

Toma de Decisiones en Ambientes de Incertidumbre 53

tives would increase the possibility of findinga better solution. The MDSS helps to the res-olution of GDM problems providing a propi-tious environment for the communication, in-creasing the satisfaction of the user and, in thisway, improving the final decisions [13].

3 A new mobile decision supportsystem

In this section, we present the implementedprototype, explaining the architecture and thework flow that summarizes the functions ofthis system.

A DSS can be built in several ways, andthe used technology determines how a DSS hasto be developed. The most used architecturefor mobile devices is the “Client/Server” archi-tecture, where the client is a mobile device.The client/server paradigm is founded on theconcept that clients (such as personal com-puters, or mobile devices) and servers (com-puters) are both connected by a network en-abling servers to provide different services forthe clients. When a client sends a request toa server, this server processes the request andsends a response back to client.

We have chosen a thick-client model for ourimplementation. This allows us to use the soft-ware in all the mobile devices without tak-ing into account the kind of browser. Fur-thermore, the technologies that we have usedto implement the prototype comprise Javaand Java Midlets for the client software, PHPfor the server functions and MySQL for thedatabase management.

So, the prototype allows user to send his/herpreferences by means of a mobile device, andthe system returns to the experts the final so-lution or recommendations to increase the con-sensus levels, depending of the status of thedecision process. An important aspect is thatthe user-system interaction can be done any-time and anywhere which facilitates expert’sparticipation and the resolution of the deci-sion process. In what follows, we describe theclient and server of the prototype in detail.

3.1 Client side

The client software shows the next seven in-terfaces to the experts:

• Authentication: The device asks a userand a password to access the system.

• Connection: The device must be con-nected to the network to send/receive in-formation to the server.

• Problem description: When a decisionprocess is started, the device shows tothe experts a brief description about theproblem and the set of alternatives.

• Insertion of preferences: The device willhave a specific interface to insert the lin-guistic preferences using a set of labels.To introduce or change the preferences onthe interface, the user has to use the keysof the device (see Fig 2).

Figure 2: Insertion of preferences

• Swap of Alternatives: When a new alter-native appears in the environment of theproblem because some dynamic externalfactors have changed and this alternativedeserves to be a member of the discus-sion subset or when an alternative have alow dominance degree to the current tem-porary solution of consensus, the systemasks the experts if they want to modifythe discussion subset by swapping thesealternatives. The experts can assess if


they agree to swap the alternatives send-ing their answer to the question received(see Fig 3). If most of the experts agree,the server updates the set of alternativesand each expert has to send his prefer-ences about the new one.

Figure 3: Swap of alternatives

• Feedback: When opinions should be mod-ified, the device shows experts the rec-ommendations and allows experts to sendtheir new preferences (see Fig 4).

Figure 4: Recommendations

• Output: At the end of the decision pro-cess, the device will show the set of solu-tion alternatives as an ordered set of al-ternatives (see Fig 5).

Figure 5: Final solution

On the technical side of the developmentof the client part, it is worth noting that theclient application complies with the MIDP 2.0specifications [1] and that the J2ME Wire-less Toolkit 2.2 [2] provided by SUN wasused in the development phase. This wire-less toolkit is a set of tools that provideJ2ME developers with some emulation envi-ronments, documentation, and examples todevelop MIDP-compliant applications. Theapplication was later tested using a JAVA-enabled mobile phone on a GSM network us-ing a GPRS-enabled SIM card. The MIDPapplication is packaged inside a JAVA archive(JAR) file, which contains the application’sclasses and resource files. This JAR file isthe one that actually is downloaded to thephysical device (mobile phone) along with theJAVA application descriptor file when an ex-pert wants to use our this prototype.

3.2 Server side

The server is the main side of the proto-type. It implements the main modules and thedatabase that stores the problem data (expertsand alternatives) as well as problem parame-ters, that are introduced at the beginning bythe moderator, and the information generatedduring the decision process. The communi-cation with the client to receive/send infor-mation from/to the experts is supported by


mobile Internet (M-Internet) technologies (seeFig. 6). Concretely, the three modules of theserver are:

Figure 6: Server modules

3.2.1 Decision module

In this contribution, we assume that theexperts give their preferences using FLPRs.Once experts have sent their preferences, theserver starts the decision module to obtain atemporary solution of the problem. In thismodule the consensus measures are also cal-culated. This module has two different pro-cesses: 1) selection process and 2) consensusprocess.

1. Selection Process: This process has twodifferent phases:

(a) Aggregation phase:This phase defines a collective pref-erence relation, P c = (pc

lk), obtainedby means of the aggregation of allindividual linguistic preference re-lations

{P 1, P 2, . . . , Pm}

. It indi-cates the global preference betweenevery pair of alternatives accordingto the majority of experts’ opin-ions. The aggregation is carried outby means of a LOWA operator φQ

guided by a fuzzy linguistic non-decreasing quantifier Q [9]:

pclk = φQ(p1

lk, . . . , pmlk)

(b) Exploitation phase:This phase transforms the globalinformation about the alternativesinto a global ranking of them, fromwhich the set of solution alternatives

is obtained. The global ranking isobtained applying these two choicedegrees of alternatives on the collec-tive preference relation [8]:

• QGDDl: This quantifier guideddominance degree quantifies thedominance that one alternativexl has over all the others in afuzzy majority sense.• QGNDDl: This quantifier

guided non-dominance degreegives the degree in which eachalternative xl is not dominatedby a fuzzy majority of theremaining alternatives.

2. Consensus Process:

We assume that the consensus is a mea-surable parameter whose highest valuecorresponds to unanimity and lowest oneto complete disagreement. We use someconsensus degrees to measure the currentlevel of consensus in the decision process.They are given at three different levels[4, 10]: pairs of alternatives, alternativesand relations. The computation of theconsensus degrees is carried out as follows:

(a) For each pair of experts, ei, ej (i <j), a similarity matrix, SM ij =(smij

lk), is defined where smijlk =

1−|I(pi

lk)− I(pjlk)|

g.

(b) A consensus matrix, CM , is calcu-lated by aggregating all the simi-larity matrices using the arithmeticmean as the aggregation functionx̄. cmlk = x̄(smij

lk) where (i =1, ...,m− 1, j = i+ 1, ...,m).

(c) Once the consensus matrix, CM , iscomputed, we proceed to calculatethe consensus degrees:

i. Consensus degree on pairs of al-ternatives, cplk. It measures theagreement on the pair of alter-natives (xl, xk) amongst all theexperts: cplk = cmlk.


ii. Consensus degree on alterna-tives, cal. It measures theagreement on an alternative xl

amongst all the experts: cal =∑nk=1 cplk

n.

iii. Consensus degree on therelation, cr. It measuresthe global consensus degreeamongst the experts’ opinions:cr =

∑nl=1 cal

n.

Initially, in this consensus model we con-sider that in any nontrivial GDM prob-lem the experts disagree in their opinionsso that decision has to be viewed as aniterative process. This means that agree-ment is obtained only after some roundsof consultation. In each round, we calcu-late the consensus measures and check thecurrent agreement existing among expertsusing cr.

3.2.2 Management of dynamic infor-mation module

Classical GDM models are defined in staticframeworks. In order to make the decisionmaking process more realistic, this module isable to deal with dynamic parameters in deci-sion making. The main parameter that couldvary through the decision making process isthe set of alternatives of the problem becauseit could depend on dynamical external factorslike the traffic [7], or the meteorological con-ditions [6], and so on. In such a way, we cansolve dynamic decision problems in which, atevery stage of the process, the discussion iscentered on different alternatives.

This tool allows to introduce new alterna-tives in the discussion subset, but this changehas to be approved by the experts. To do so,the mechanism has two phases. At the firstone, the system identifies the new alternativeto include in the set of discussion alternatives(discussion subset) and the worst alternativeof the current discussion subset. The secondone is to ask experts about if they agree the re-placement and updating the discussion subset[13].

3.2.3 Feedback module

To guide the change of the experts’ opinions,the DSS simulates a group discussion sessionin which a feedback mechanism is applied toquickly obtain a high consensus level. Thismechanism is able to substitute the modera-tor’s actions in the consensus reaching process.The main problem for the feedback mechanismis how to find a way of making individual posi-tions converge and, therefore, how to supportthe experts in obtaining and agreeing with aparticular solution. To do that, we computeothers consensus measures, called proximitymeasures [4, 10].

These measures evaluate the agreement be-tween the individual experts’ opinions and thegroup opinion. To compute them for eachexpert, we need to use the collective FLPR,P c = (pc

lk), calculated previously.

1. For each expert, ei, a proximity matrix,PM i = (pmi

lk), is obtained where

pmilk = 1− |I(pi

lk)− I(pclk)|

g.

2. Computation of proximity measures atthree different levels:

(a) Proximity measure on pairs of al-ternatives, ppi

lk. It measures theproximity between the preferenceson each pair of alternatives of the ex-pert ei and the group: ppi

lk = pmilk.

(b) Proximity measure on alternatives,pai

l. It measures the proximity be-tween the preferences on each alter-native xl of the expert ei and thegroup: pai

l =∑n

k=1 ppilk

n.

(c) Proximity measure on the relation,pri. It measures the global proxim-ity between the preferences of eachexpert ei and the group: pri =∑n

l=1 pail

n.

These measures allow us to build a feedbackmechanism so that experts change their opin-ions and narrow their positions.


4 Conclusions

We have presented a prototype of Mobile DSSfor dynamic GDM problems based on dynamicsets of alternatives and tools of computingwith words, which uses the advantages of M-Internet technologies to improve the user satis-faction with the consensual process. With thisprototype, we shall develop decision processesin a flexible way, that is, we can manage de-cision situations in dynamic environments atanytime and anywhere.

Acknowledgments

This paper has been developed with the fi-nancing of FEDER funds in FUZZY-LINGproject (TIN2007-61079), PETRI project(PET2007-0460), project of Ministry of Pub-lic Works (90/07) and Excellence AndalusianProject (TIC5299).

References

[1] http://java.sun.com/products/midp/.

[2] http://java.sun.com/products/sjwtoolkit/.

[3] S. Alonso, E. Herrera-Viedma, F. Chi-clana, and F.Herrera. Individual andsocial strategies to deal with ignorancesituations in multi-person decision mak-ing. International Journal of InformationTechnology & Decision Making, 8(2):313–333, 2009.

[4] F.J. Cabrerizo, J.M. Moreno, I.J. Pérez,and E. Herrera-Viedma. Analyzing con-sensus approaches in fuzzy group decisionmaking: advantages and drawbacks. SoftComputing, 14(5):451–463, 2010.

[5] F. Chiclana, F. Herrera, and E. Herrera-Viedma. Integrating three representationmodels in fuzzy multipurpose decisionmaking based on fuzzy preference rela-tions. Fuzzy Sets and Systems, 97(1):33–48, 1998.

[6] H. Clarke. Classical decision rules andadaptation to climate change. The Aus-tralian Journal of Agricultural and Re-source Economics, 52:487–504, 2008.

[7] H. Dia. An agent-based approach to mod-elling driver route choice behaviour un-der the influence of real-time information.Transportation Research Part C, 10:331–349, 2002.

[8] F. Herrera and E. Herrera-Viedma. Lin-guistic decision analisis: steps for solvingdecision problems under linguistic infor-mation. Fuzzy Set and Systems, 115:67–82, 2000.

[9] F. Herrera, E. Herrera-Viedma, and J.L.Verdegay. Direct approach processes ingroup decision making using linguisticowa operators. Fuzzy Sets and Systems,79:175–190, 1996.

[10] F. Herrera, E. Herrera-Viedma, and J.L.Verdegay. A model of consensus in groupdecision making under linguistic assess-ments. Fuzzy Sets and Systems, 78(1):73–87, 1996.

[11] J. Kacprzyk. Group decision making witha fuzzy linguistic majority. Fuzzy Sets andSystems, 18:105–118, 1986.

[12] J. Katz. Handbook of Mobile Communi-cation Studies. The MIT Press, 2008.

[13] I.J. Perez, F.J. Cabrerizo, and E. Herrera-Viedma. A mobile decision support sys-tem for dynamic group decision makingproblems. IEEE Transactions on Sys-tems, Man and Cybernetics - Part A: Sys-tems and Humans, in press.

[14] J. Schiller. Mobile Communications (2ndEdition). Addison Wesley, 2003.

[15] L.A. Zadeh. The concept of a linguisticvariable and its applications to approx-imate reasoning. Information Sciences,Part I, II, III, 8,8,9:199–249,301–357,43–80, 1975.


actas del iii simposio sobre lógica fuzzy y soft computing...

Documents