Available online at www.sciencedirect.com

www.elsevier.com/locate/eswa

Expert Systems with Applications 35 (2008) 2017–2031

Expert Systemswith Applications

Intelligent impact assessment of HRM to the shareholder value

George Xirogiannis a,*, Panagiotis Chytas b,1, Michael Glykas c,2, George Valiris b,1

a University of Piraeus, Department of Informatics, 80, Karaoli and Dimitriou Street, 185 34 Piraeus, Greeceb University of Aegean, Department of Business Administration, 8, Michalon Street, Chios 82 100, Greece

c University of Aegean, Department of Financial and Management Engineering, 31, Fostini Street, Chios 82 100, Greece

Abstract

Despite the extensive research in human capital management and performance measurement, intelligent treasoning mechanisms,which integrate human resource (HR) practices into strategic-level shareholder decisions, are still emerging. This paper discusses a novelapproach of designing a decision-modeling tool, which assesses the impact of contemporary human resource management (HRM) prac-tices to the shareholder value and satisfaction. The underlying research addresses the problem of establishing HRM interrelationships inorder to drive the overall business performance from the shareholder value perspective. The proposed methodology tool utilizes the fuzzycausal characteristics of fuzzy cognitive maps (FCMs) to generate a hierarchical and dynamic network of interconnected HR perfor-mance drivers. The intelligent computing characteristics of FCMs are also employed to establish a dynamic feedback and bi-directionalalignment of HRM practices and strategic improvement. Finally, this research provides a practical insight on the applicability of softapproaches in capturing and illustrating the effect of HRM practices.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Fuzzy cognitive maps; HR performance; Decision support; Strategic planning

1. Introduction

Enterprises are evolving in turbulent and equivocal envi-ronments (e.g. Drucker, 1993; Grove, 1999; Kellys, 1998).This requires enterprises to be alert and watchful for thedetection of weak signals (e.g. Ansoff, 1975) or discontinu-ities of emerging threats and to initiate further probingbased on such detection (Walls et al., 1992). Enterprisestoday face critical business challenges (Ulrich, 1998) likeglobalization, profitability through growth, technologyintegration, intellectual capital management, continuouschange, etc. Such challenges require organizations to buildnew capabilities, but it is not always apparent who shouldbe responsible for developing those capabilities. Perhaps,

0957-4174/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.eswa.2007.08.103

* Corresponding author. Tel.: +30 210 4142000; fax: +30 210 4142328.E-mail addresses: georgex@unipi.gr (G. Xirogiannis), p.chytas@chios.

aegean.gr (P. Chytas), mglikas@aegean.gr (M. Glykas), gval@aegean.gr(G. Valiris).

1 Tel.: +30 22710 35000; fax: +30 22710 35099.2 Tel.: +30 22710 35400; fax: +30 22710 35499.

everyone and no one, but in any case this is a uniqueHR’s opportunity to play a leadership role in enablingorganizations to meet such competitive challenges. Ensur-ing that human resource (HR) strategies are in place to dealwith these challenges is increasingly recognized as criticalto success (Leopold et al., 1999).

Human resource management (HRM) in the literaturehas been considered a second- or third-order strategy, lar-gely related to implementation rather than shareholderlevel decision-making. The process of HR strategy formu-lation and evaluation had not been widely conceptualizeduntil recently. Moreover, the impact of HRM practices tothe shareholder strategic value is not modeled adequately,despite the utilization of sophisticated performance evalua-tion mechanisms at the employee level. The evidence thatHR issues are fundamental to business is compelling atthe level of unit labor costs, but whether they are funda-mental to the strategy process remained highly question-able until recent years (Ritson, 1999). This can beattributed to the fact that contemporary performance eval-uation mechanisms focus on analyzing the operational

2018 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

effectiveness of the human capital rather than addressingthe issue of strategic alignment.

This paper addresses the problem of designing a novelmodeling methodology tool to act as an intelligent decisionsupport mechanism for evaluating the impact of HRMpractices to the shareholder value and satisfaction. Thisattempt focuses on bridging the gap between the opera-tional characteristics of HRM practices and the strategicdecisions at the shareholder level. The proposed methodol-ogy tool utilizes the fuzzy causal characteristics of fuzzycognitive maps (FCMs) to generate a hierarchical anddynamic network of interconnected HR performance driv-ers. By using FCMs, the proposed novel mechanism simu-lates the operational efficiency of HR models withimprecise relationships and quantifies the impact of HRMactivities to the overall shareholder satisfaction model.

This research builds on contemporary HRM experienceto establish a ‘‘soft computing’’ approach on how to inter-relate HRM activities and shareholder value. The FCMapproach does not pose as a substitute of traditionalHRM operations nor does it offer an alternative to HRperformance evaluation. It presents an intelligent deci-sion-making framework for strategic-level HRM basedon scenario building and ex ante impact assessment.

Primarily, the proposed model targets the principledirectors and stakeholders of HRM projects (e.g. HRdepartment, change management leaders, business strategyleaders, etc) assisting them to reason effectively about thestatus of strategic-level performance metrics, given the(actual or hypothetical) implementation of a set of HRpractice changes. However, the holistic nature of the pro-posed model may couple effectively with other strategicperformance evaluation systems. Given the demand foreffective shareholder positioning, such a succinct mecha-nism of conveying the essential dynamics of HR practicesis believed to be useful for anyone contemplating or under-taking a strategy formulation exercise. Nevertheless, theexplanatory nature of the mechanism can prove to be use-ful in a wider educational setting.

As far as the contribution to decision-making is con-cerned, the application of FCMs as an intelligent modelerof HR knowledge is believed to be novel. Moreover, thispaper extends typical FCM algorithms in order to adaptto the distributed nature of typical HR activities. Also, thisresearch adopts a new qualitative approach to interpretfuzzy linguistic variables to weight and concept values inorder to support further the soft computing characteristicsof the tool. It is the belief of this paper that the fuzzy rea-soning capabilities enhance considerably the usefulness ofthe proposed mechanism while reducing the effort of iden-tifying quantitative impact measurements. As far as theadded value of this research is concerned, the proposedmethodology offers an alternative approach to HRM basedon shareholder value analysis and modeling. Preliminaryexperiments indicate that the proposed approach can beeffective and realistic, without employing detailed quantita-tive calculations.

This paper consists of six sections. Section 2 presents ashort literature overview. Section 3 presents an overview ofthe proposed system, Section 4 discusses the new approachto HRM modeling and Section 5 discusses the major advan-tages of the proposed tool. Finally, Section 6 concludes thispaper and briefly discuses future research activities.

2. Literature overview

2.1. Contemporary HRM

As firms become increasingly aware that people areamong their most valuable strategic assets, they are reap-praising the way in which they manage their human capital.The emphasis is shifting from personnel management tothe wider, strategic concept of human resource manage-ment in which human resource policies and activities,including training and development, are linked closely tobusiness strategy. HR specialists who wish to develop astrategic approach to people management must establishcredibility with top management as the key figures toachieve successful results (Handy et al., 1989).

Strategic human resource management addresses a num-ber of key issues. Typical examples (loosely adapted fromBaker, 1999) include internal integration of personnel pol-icies, their external integration with overall strategy, linemanagement responsibility for HR implementation, indi-vidual rather than collective employee relations, HR com-mitment, HR initiatives, the managerial role of ‘‘enabler’’,‘‘empowerer’’, and ‘‘facilitator’’, etc.

Following a resource-based view of an enterprise, firmscan develop sustainable competitive advantage by creatingvalue in a way that is rare and difficult for competitors toimitate (Baker, 1999). Although traditional sources of com-petitive advantage such as natural resources, technology,economies of scale, value creation, and so forth, theresource-based argument is that these sources are increas-ingly easy to imitate, especially in comparison to a complexsocial structure such as an employment system. If that is so,HR strategies may be an important source of sustainablecompetitive advantage (Lado & Wilson, 1994).

But one may ask how HR management can deliver orga-nizational excellence and competitive advantage. Accord-ing to Ulrich (1997) there are four ways to do so:

• HR management could become a partner with seniorand line managers in strategy execution, helping to moveplanning from the conference room to the marketplace.

• HR management could become an expert in the waywork is organized and executed, delivering administra-tive efficiency to ensure that costs are reduced whilequality is maintained.

• HR management could become a champion for employ-ees, vigorously representing their concerns to seniormanagement and at the same time working to increaseemployee contribution; that is, employees’ commitmentto the organization and their ability to deliver results.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2019

• Finally, HR management could become an agent of con-tinuous transformation; shaping processes and a culturethat together improve an organization’s capacity forchange.

Research (Schuler & Jackson, 1996) presents furtherevidences that show how HR can contribute to competi-tive advantage. Recent consultancy studies show thatfirms that manage their human resource effectively havehigher levels of profitability, higher annual sales peremployee (productivity), higher market value, higher earn-ings-per-share growth, etc. In other words they meet theneeds of the stockholders and investors. Effective HRMcan also meet the needs of the employees in several ways:as firms survive, expand, and increase their profitability,they provide further employment security, job opportuni-ties, and high remuneration packages. Successful HRMserves the needs of society by elevating the standard of liv-ing, strengthening legal regulations and ethical guidelines,and controlling the effect of the firm on the surround-ing community; that is, it contributes positively to society,which in turn supports a favorable corporate image in themind of the public.

However, strategic HRM will not be taken seriouslyunless it can be shown that it is worth the return on invest-ment. Traditionally, the HR director could discussabstractly and conceptually about employee morale, turn-over, and commitment. To fulfill the business partner roleof HR, concepts need to be replaced with evidence, ideaswith results, and perceptions with impact assessments(Ulrich, 1997).

2.2. Decision tools as back end to HRM

Conceptually, a central aspect of ‘‘strategic HRM’’ isthe integration of the HR function with strategic deci-sion-making (Bennett et al., 1998). Research like (Cook& Ferris, 1986; Huselid, 1995; Wright & McMahan,1992) reason that organizations linking HRM to strategicdecision-making will outperform those that do not. How-ever, without building an information system one cannotprovide individual system users with integrated informa-tion. This integration is achieved through HumanResource Information Systems (HRIS).

HRIS can be defined as integrated systems, whichgather, store and analyze information regarding an organi-zation’s human resources. According to Kavanaugh et al.(1990) the first attempt to manage personnel informationwas bounded to employee names, addresses and briefemployment history. Between 1945 and 1960, organiza-tions became more aware of human capital issues andbegan to develop formal processes for employee selectionand development. At the same time, organizations beganto recognize the importance of employee morale on thefirm’s overall effectiveness. While this period of change inthe profession did not result in significant changes inHRIS, it set the pace for consequent changes.

In the next 20 years (1960–1980) HR was integrated intothe core business mission, and governmental and regula-tory reporting requirements for employees also increasedsignificantly. The HR department became one of the mostimportant users of the exceptionally costly computing sys-tems of the day, often edging out other functional areas forcomputer access. Although HRIS systems were computer-ized and grew extensively in size and scope during this per-iod, they were merely simple record-keeping systems(Kavanaugh et al., 1990).

Over the last two decades, firms have increasingly reliedon the HR function to provide management solutions thatincrease the effectiveness of human capital. HRIS systemshave evolved into complex tools designed not only to man-age a rich variety of information about the firm’s humancapital, but to also provide analytical tools to assist in deci-sion-making about the management of those assets (Kava-naugh et al., 1990).

Recent research activities like (Tannenbaum, 1990)decompose HRIS into of three systems, namely Expert Sys-tems (ES), Decision Support Systems (DSS) and ExecutiveInformation Systems (EIS).

2.2.1. Expert systems (ES)

Expert systems are designed to model the knowledge ofdomain experts in order to help users derive expert-qualitysolutions. ES are expected to support HR managers in ana-lyzing and solving verbally stated problems, which requireHRM expertise. They model the human HR decision-mak-ing process by applying questions and heuristics used byHR domain experts (Stefik, 1990). Hence, non-expertscan use ES to produce solutions by accessing to the expertknowledge base (Lawler & Eliot, 1996). However, thereexist only a few attempts to develop ES in the HR field(for example (Byun & Suh, 1994; Sturman et al., 1996).

2.2.2. Decision support systems (DSS)

The important role of DSS is to provide information tousers in order to analyze situations and make calculateddecisions (Poe et al., 1998). Research efforts like (Mohanty& Deshmukh, 1997) discuss a comprehensive HR planningsystem for large multi-divisional, multi-locational petro-leum companies. Such efforts adopt a resource supportapproach, which primarily focuses on providing the criticaltools and resources needed for HRM at the tactical level toplan for manpower requirements, recruitment, placement,promotion and appraisal. Research (Bellone et al., 1995)proposes the ISPM decision support system to assist in per-sonnel career management. The ISPM system has beendesigned mainly to deal with the establishment of the train-ing a human resource should undergo before occupying ajob position. Hence this system addresses mainly tacticalmanagement issues, rather than strategic level choices.Research (Niehaus, 1995) discusses the results of a multi-year effort in implementing a series of human resourceplanning DSS applications in the US Navy shipyard com-munity. These applications span both corporate (integrated

Number of customer

Operating cost

W41

Product price

W13

W12

W23

W52

W35

W34

W63

W46

W56

Product risk

Company profitability

Sales volumes

Fig. 1. Simple FCM.

2020 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

management of eight shipyards Navy-wide with 78,000employees) and local (individual shipyards with 9000employees) perspectives. This research effort first concen-trates on the development, implementation and introduc-tion of a DSS into a large organization that is goingthrough a personnel-downsizing process. Then, a discus-sion is provided of the implementation of complex humanresource planning models in a DSS that can be used bya mid-level staff person rather than at strategy-levelexecutives.

2.2.3. Executive information systems (EIS)

EIS first appeared on the decision support scene in themiddle of 1980s (Schendel & Hofer, 1979). The aim ofEIS is to ‘‘provide executives with easy access to internaland external information that is relevant to their criticalsuccess factors’’ (Watson & Howdeshel, 1997). Severalcommercial applications implement HR models, howeverthe ‘‘source mechanism’’ of such EIS is not always avail-able to public.

2.3. FCMs as a modeling technique

Fuzzy cognitive maps is an intelligent modeling method-ology for complex decision systems, which originated fromthe combination of Fuzzy Logic (Zadeh, 1965) and NeuralNetworks. An FCM describes the behavior of an intelligentsystem in terms of concepts; each concept represents anentity, a state, a variable, or a characteristic of the system(Dickerson & Kosko, 1997).

FCM nodes are named by such concepts forming the setof concepts C = {C1,C2, . . . ,Cn}. Arcs (Cj,Ci) are orientedand represent causal links between concepts; that is howconcept Cj causes concept Ci Arcs are elements of the set(A = {Cj,Ci)ji} � C · C. Weights of arcs are associatedwith a weight value matrix Wn·n, where each element ofthe matrix wji 2 [�1, . . . , 1] � R such that if (Cj,Ci) 62 Athen wji = 0 else excitation (respectively inhibition) causallink from concept Cj to concept Ci gives wji > 0 (respec-tively wji < 0). The proposed methodology frameworkassumes that [�1, . . . , 1] is a fuzzy bipolar interval, bipolar-ity being used as a means of representing a positive or neg-ative relationship between two concepts.

In practice, the graphical illustration of an FCM is asigned graph with feedback, consisting of nodes andweighted interconnections (e.g. �!Weight

). Signed and weightedarcs (elements of the set A) connect various nodes (ele-ments of the set C) representing the causal relationshipsamong concepts. This graphical representation illustratesdifferent aspects in the behavior of the system, showingits dynamics (Kosko, 1986) and allowing systematic causalpropagation (forward/backward chaining).

Positive or negative sign and fuzzy weights model theexpert knowledge of the causal relationships (Kosko,1991). Concept Cj causally increases Ci if the weight valuewji > 0 and causally decreases Ci if wji < 0. When wji = 0,concept Cj has no causal effect on Ci. The sign of wji

indicates whether the relationship between concepts is posi-tive ðCW j;i

j ! CiÞ or negative ðCW j;ij !� CiÞ, while the

value of wji indicates how strongly concept Cj influencesconcept Ci. The forward or backward direction of causalityindicates whether concept Cj causes concept Ci or viceversa (e.g. Fig. 1).

Simple variations of FCMs mostly used in business deci-sion-making applications may take trivalent weight values[�1,0,1]. This paper allows FMCs to utilize fuzzy wordweights like strong, medium, or weak, each of these wordsbeing a fuzzy set to provide complicated FCMs. In con-trast, (Kwahk & Kim, 1999) adopted only a simple relativeweight representation in the interval [�1, . . . , 1]. To thisextend, (Kwahk & Kim, 1999) offered reduced functional-ity since it does not allow fuzzy weight definitions.

Generally speaking FCM concept activations take theirvalue in an activation value set V = {0,1} or {�1,0,1} if incrisp mode or [�d, 1] with d = 0 or 1 if in fuzzy mode. Theproposed methodology framework assumes fuzzy modewith d = 1. At step t 2 N, each concept Cj is associated withan inner activation value at

j 2 V , and an external activationvalue et

aj2 R. FCM is a dynamic system. Initialization is

a0j ¼ 0. The dynamic obeys a general recurrent relation

atþ1 ¼ f ðgðeta;W

TatÞÞ, "t P 0, involving weight matrixproduct with inner activation, fuzzy logical operators (g)between this result and external forced activation andfinally normalization (f). However, this paper assumes noexternal activation (hence no fuzzy logical operators),resulting to the following typical formula for calculatingthe values of concepts of FCM:

atþ1i ¼ f

Xn

j¼1;j 6¼i

wjiatj

!for i ¼ 1; . . . ; n; ð1Þ

where atþ1i is the value of concept Ci at step t + 1, at

j thevalue of the interconnected concept Cj at step t, wji is theweighted arc from Cj to Ci and f : R! V is a thresholdfunction, which normalizes activations. Two normalizationfunctions are usually used. The unipolar sigmoid functionwhere k > 0 determines the steepness of the continuousfunction f ðxÞ ¼ 1

1þe

�kx. When concepts can be negative

(d < 0), function f(x) = tanh(x) can also be used.

BusinessStrategy

Organisational Capabilities

Human ResourcePractices

ShareholderSatisfaction

CustomerSatisfaction

EmployeeSatisfaction

Fig. 2. Static integrated model connecting strategic HR framework to keyresults areas.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2021

2.4. Applications of fuzzy cognitive maps

Over the last 10 years, a variety of FCMs have been usedfor capturing–representing knowledge and intelligent infor-mation in engineering applications, for instance, geograph-ical information systems (Liu & Satur, 1999) and faultdetection (Ndouse & Okuda, 1996; Pelaez & Bowles,1995). FCMs have been used in modeling the supervisionof distributed systems (Stylioa & Georgopoulos, 1997).FCMs have also been used in operations research (Craigeret al., 1996), web data mining (Hong & Han, 2002; Leeet al., 2002), as a back end to computer-based modelsand medical diagnosis (e.g. Georgopoulos et al., 2002).

Several research reports applying basic concepts ofFCMs have also been presented in the field of business(e.g. Xirogiannis & Glykas, 2004; Xirogiannis & Glykas,2004) and other social sciences. Research in Axelrod(1976) and Perusich (1996) have used FCM for represent-ing tacit knowledge in political and social analysis. FCMshave been successfully applied to various fields such asdecision making in complex war games (Klein & Cooper,1982), strategic planning (Diffenbach, 1982), (Ramaprasad& Poon, 1985), information retrieval (Johnson & Briggs,1994), distributed decision process modeling (Zhanget al., 1994), performance-driven changes (Xirogiannis &Glykas, 2004). Research like (Lee & Kim, 1997) has suc-cessfully applied FCMs to infer rich implications fromstock market analysis results. Research like (Lee & Kim,1998) also suggested a new concept of fuzzy causal rela-tions found in FCMs and applied it to analyze and predictstock market trends. The inference power of FCMs hasalso been adopted to analyze the competition betweentwo companies, which are assumed to use differentialgames mechanisms to set up their own strategic planning(Lee & Kwon, 1998). FCMs have been integrated withcase-based reasoning technique in the field of knowledgemanagement (Noh et al., 2000). Research (Parenthoen &Reignier, 2001) proposed the use of FCM as a tool tomodel emotional behavior of virtual actors improvisingin free interaction within the framework of a ‘‘nouvellevague’’ scenario. Recent research adopted FCMs to sup-port the core activities of highly technical functions likeurban design (Xirogiannis et al., 2004). Summarizing,FCMs can contribute to the construction of more intelli-gent systems, since the more intelligent a system becomes,the more symbolic and fuzzy representations it utilizes.

In addition, a few modifications have been proposed. Forexample, the research in (Silva (1995)) proposed new formsof combined matrices for FCMs, the research in (Hagiwara(1992)) extended FCMs by permitting non-linear and timedelay on the arcs, the research in (Schneider et al. (1995)) pre-sented a method for automatically constructing FCMs.More recently, Liu & Satur (1999) has carried extensiveresearch on FCMs investigating inference properties ofFCMs, proposed contextual FCMs based on the object-ori-ented paradigm of decision support and applied contextualFCMs to geographical information systems (Liu, 2000).

3. Fundamentals of the methodology tool

3.1. An integrated framework

Typical HR decision-making can be achieved throughthe establishment of an integrated performance flow: creat-ing and improving HR value to enhance employee satisfac-tion, successful operational utilization of all HR entitiesand strategic effectives to achieve strategic shareholdervalue. Such a typical though static performance flow(Fig. 2) is usually utilized by most research attempts as pre-sented in Section 2.2.

This paper extends such static frameworks by proposinga dynamic feedback to assert and maintain employee satis-faction based on a continuous establishment of contempo-rary HR practices. Having said that, it is the proposal ofthis research that effective integration of HRM benefitsand shareholder value can be realized through a bi-direc-tional alignment:

• HRM perspective (‘‘HR practices! shareholder satis-faction’’): It is the view of this research that optimaldecision-making can be achieved through the establish-ment of an integrated performance flow: creating andimproving HR value to enhance employee satisfaction,successful operational utilization of all HR entities, stra-tegic effectives to achieve shareholder value and feed-back to assert and maintain employee satisfaction.

• Shareholder value perspective (‘‘shareholder satisfac-tion! HR practices’’): It is the view of this research thatoptimal strategic improvement can be realized throughthe alignment of customer-oriented service delivery withthe organizational capabilities of the enterprise, as wellas the alignment of HRM with the organizationalcharacteristics.

Fig. 3 illustrates the dynamic integration of the strategicHR framework with the business framework. This researchgroups the interrelationships between the core componentsinto two decision modeling chains compatible with the twoalignment perspectives:

• The forward chain assumes that HRM caters forimproved shareholder value (as far as HR is concerned).HR practices can be conceptualized as the key driversfor building organizational capabilities, enhancing

BusinessStrategy

Organisational Capabilities

Human ResourcePractices

ShareholderSatisfaction

CustomerSatisfaction

EmployeeSatisfaction

Fig. 3. Dynamic integrated model connecting strategic HR framework tokey results areas.

2022 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

employee satisfaction, and more innovatively, shapingcustomer satisfaction. Organizational capabilities alongwith customer satisfaction and the business strategyaffect the shareholder satisfaction and value.

• The backward chain assumes that shareholder valuedrives further changes to organizational characteristicsin an attempt to improve HR practices and maintainemployee satisfaction. This backward chain assumesthat strategic level decisions cater for proper manage-ment of human assets in return for increased share-holder satisfaction.

3.2. FCMs in decision modeling

3.2.1. Updated FCM algorithm

This research extends typical FCM algorithms (as usedfor example by Kwahk & Kim, 1999), by proposing the fol-lowing updated algorithm aiming at modeling more effec-tively the dispersed nature of the HR domain:

Atþ1i ¼ f k1At

i þ k�2Xn

j¼1;j 6¼i

W jiAtj

!: ð2Þ

This paper also assumes that k1 and k2 can be fuzzy sets,extending previous relevant research. The coefficient k1

represents the proportion of the contribution of the valueof the concept Ai at time t in the computation of the va-lue of Ai at time t + 1. In practice, this is equivalent toassume that Wii = k1. The incorporation of this coeffi-

Profitabi

Sales Depar

Sales & promotion expertise

Internet sales unit

Technical background

Direct sal

W1 W2 W1 W

Sales & promotion expertise

Fig. 4. Horizontal decompo

cient results in smoother variation of concept values dur-ing the iterations of the FCM algorithm. The coefficientk2 expresses the ‘‘influence’’ of the interconnected con-cepts in the configuration of the value of the conceptAi at time t + 1. In practice, it indicates the centralizedor decentralized importance of the concept Ai in compar-ison to other concepts of the map. Intuitively, the intro-duction coefficient k2 imposes three steps of analysis forestablishing the ‘‘influence’’ of causal HR relationships:

• Step 1: Estimation of the direct influence of an HRconcept Cj to a shareholder value concept Ci withthe actual weight wji of the relationship. In such casesboth Ci and Cj belong to the same HR domain andlevel (e.g. organizational unit, line of service, etc). Thisstep sets k2 = 1.

• Step 2: approximation of the indirect importance ofduplicate causal relationships spanning to ‘‘equivalent’’HR domains and levels, using coefficient k2 < 1.

• Step 3: approximation of the indirect importance of cau-sal relationships spanning to different HR domains andlevels using coefficient k2 < 1.

The three steps introduced above are used to estimate k2

under three different ‘‘influence’’ cases. Consider for exam-ple a typical sales department of an enterprise (Fig. 4) con-sisting of several sales units/teams, each offering the sameservices over different distribution channels (e.g. internetsales, direct sales, phone sales, etc). All units and channelsoperate under the same strategic framework. Let ‘‘Sales &promotion expertise’’ and ‘‘profitability’’ be interrelatedperformance concepts. As far as the second step is con-cerned, from a theoretical standpoint weight w1 shouldappear to be the same for all duplicate ‘‘Sales & promotionexpertise – profitability’’ relationships across all sales units.From a practical standpoint, sales units may serve differentnumber of customers (or even customers with differenttransaction volumes). In this case, coefficient k2 modelsthe fact that direct sales & promotion expertise can provemore useful to the sales persons of the ‘‘Direct sales unit’’simply because such sales persons may serve more custom-

lity

tment

es unit Phone sales unit

2 W1 W2

Technical background

Sales & promotion expertise

Technical background

sition of HR practices.

C1

C2 C3 C4

C5

Level 0

C6

W1

W2

W3Level 1

Level 2

Level 3

k2,(L1,L2)

k2 ,(L2,L3)

k2 ≈ k2,(L1,L2) + k2 ,(L2,L3)

Fig. 5. Top–down maturity decomposition.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2023

ers on a daily basis. In contrast, the sales & promotion expe-rience of the sales persons of the ‘‘Internet sales unit’’ maybe less important simply because they serve fewer customers(or even serve them off-line), despite the fact that sales andpromotion coordination is the same for all sales units.

Consider now the third step of influence analysis andFig. 5, which presents a generic organizational breakdownstructure, accompanied with a sample concept hierarchy.Regardless of weight values w1, w2, w3, coefficients 1 ¼k2;ðL1;L1Þ P k2;ðL1;L2Þ P k2;ðL2;L3Þ model the fact othat affect-ing HR concepts at level Li+1 (e.g. concept C5) are moreimportant in determining the value of affected shareholderconcepts at level Li (e.g. concept C3) in comparison toother affecting concepts at level Li+2 (e.g. concept C6).k2,(Li,Lj) stands for the value of coefficient k2 associated withlevels i and j.

Ideally, coefficient k2 could break down into two sepa-rate coefficients (say k2 ¼ x�kx

2 þ y�ky2), where kx

2 aligns indi-rectly the value of HR concept Ci with the nature ofshareholder concept Cj, while ky

2 aligns indirectly the valueof concept Ci with the significance of the hierarchical posi-tioning of concept Cj within the enterprise. Parameters x, y

could present the relative importance of kx2 and ky

2 in mixedinterconnection problems (e.g. concepts in different HRlevels participating in duplicate relationships). However,preliminary experiments showed that this separationimposed initialization overheads without increasing signif-icantly the accuracy of the FCM algorithm.

Two alternative but equally interesting interpretationsof coefficient k2 follows:

• If the set of identified HR concepts Cj, j 5 i, is incom-plete (e.g. incomplete maps, missing concepts, etc), thenthe estimation of the value of shareholder concept Ci

may prove imprecise. In this case coefficient k2 may indi-cate the sufficiency of the set of concepts Cj j 5 i, in thecalculation of the value of the concept Ci.

• If the information necessary to approximate the inputvalues of HR concepts Cj, j 5 i, is incomplete (e.g.

incomplete HR performance evaluation), then the esti-mation of the value of shareholder concept Ci may alsoprove imprecise. In this case coefficient k2 may indicatethe completeness of information utilized in the approxi-mation of the input values of concepts Cj during the cal-culation of the value of the concept Ci.

3.2.2. Sample FCMs demonstrationTo demonstrate intelligent performance modeling using

FCMs, consider Fig. 6, which depicts a graphical exampleof fuzzy relationships with no feedback loops followed bysample numerical calculations using formula (3), withk1 = k2 = 1 and k = 5 as the steepness of the normalizationfunction.

Setting the input value of ‘‘training‘‘ to 0.5 (1st scenario)triggers the FCM formula. The formula then calculates thecurrent values of all related concepts. A ‘‘zero’’ conceptvalue indicates that the concept remains neutral, waitingfor causal relationships to modify its current value. A gen-eric interpretation of the first scenario indicates that iftraining increases by 50% then deployment may increaseby 88.07% and the operations performance by 95.61%. Incontrast, if training increases by 20% (2nd scenario), thendeployment and operations performance may increase by59.86% and 89.04% respectively.

Fig. 7 presents a typical example of a feedback loop.Similarly to Fig. 6, changing the input value of ‘‘training’’triggers the FCM formula. The feedback loop dictates thatcalculations stop only when an equilibrium for all affectedconcepts has been reached, modifying all input valuesaccordingly.

4. Estimating the impact of HR practices

4.1. FCM categories

This research effort proposes a set of generic deci-sion maps that can supplement HRM modeling of typical

Scenario BScenario AScenario BScenario A

0

0

0,2

0

0

0.5

External input value (t=0)

0,9561

0,8807

0.5

Current Value

0,2Training

0,5986Deployment

0,8904Operations

performance

Concept Scenario BScenario AB

0

0

0,2

External input value (t=0)

0,9561

0,8807

0.5

Current Value

0,2

0,5986t

0,8904

Fig. 6. Sample FCM calculations with no feedback loop.

Scenario BScenario AScenario BScenario A

0

0

0,2

0

0

0.5

Initial input (t=0)

0,9666

0.9623

0,81

Current Value

0,5128Training

0,8860Deployment

0,9569Operations

performance

Concept Scenario BScenario AScenario BScenario A

0

0

0,2

0

0

0.5

Initial input (t=0)

0,9666

0.9623

0,81

Current Value

0,5128Training

0,8860Deployment

0,9569Operations

performance

Concept

Fig. 7. Sample FCM calculations with feedback loop.

Fig. 8. Map categories.

2024 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

financial sector enterprises. This intelligent modelingapproach may store concepts under different decisionmap categories (Fig. 8), for example:

• Business category: all concepts relating to core organiza-tional and HR utilization activities.

• Technical category: all infrastructure related conceptswith emphasis on technology infrastructure.

• Social category: all HRM related concepts, accompa-nied by external stimuli (e.g. external hire characteris-tics, etc).

• Integrated category: essentially all top-most concepts(e.g. a concept Ai with no backward causality such that"j: wji = 0), or concepts which may fall under more thanone main categories.

This categorization is compatible both with either the‘‘process view’’ or the ‘‘organizational view’’ (as adoptedby (Kwahk & Kim, 1999)) of the enterprise to allow greaterflexibility in modeling HRM drivers in contrast to othermodeling discussed in Section 2.2. The hierarchical decom-position of HRM concepts generates a set of dynamicallyinterconnected hierarchical maps. Each map analyzesfurther the relationships among concepts at the same hierar-chical level. Currently, the mechanism integrates more than250 concepts, forming a hierarchy of more than 15 maps. Itsdynamic interface allows its users to utilize a sub-set of these250 concepts by setting the value of the redundant conceptsand/or the value of the associated weights to zero.

This research effort can portray the HR performancemodel following either a holistic or a scalable approach.This view is analogous to seeing the shareholder satisfac-

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2025

tion and value either as a single, ‘‘big bang’’ event or as anongoing activity of targeting successive tasks of selectedsub-processes. The proposed mechanism can accommodateboth approaches, extending older models presented in Sec-tion 2.2. Also the current implementation allows:

• easy customization of the function f and easy re-config-uration of the formula Atþ1

i to adapt to the specific char-acteristics of individual enterprises,

• generation of alternative scenarios for the same skeletonFCM,

• automatic loop simulation until a user-defined equilib-rium point has been reached. Alternatively, step-by-stepsimulation (with graphical output of partial results) is alsoavailable to provide a justification for the partial results.

The following sections exhibit sample skeleton decisionmaps for the social and integrated categories, which providerelevance and research interest to this paper. While the pre-sentation of other map categories supports the completenessof the proposed methodology tool, they add little (if any atall) significance to the research activities presented by thispaper.

4.2. Social metrics

This research proposes a set of maps with generic HRMperformance concepts. For example:

• The ‘‘Deployment’’ map (Appendix, Fig. 10) essentiallyrelates concepts, which affect the operational effective-ness of the enterprise given the utilization characteristicsof its human capital.

• The ‘‘Staff morale’’ map relates all employment charac-teristics to the level of employee satisfaction. Also it sup-ports reasoning of the impact of knowledge management,training and employee satisfaction to the financial model.

• The ‘‘Benefits’’ map (Appendix, Fig. 11) relates conceptswhich affect the financial and operational effectivenessgiven a set comprehensive rewards and remunerationpractices.

• The ‘‘Training’’ map (Appendix, Fig. 12) reasons onthe impact of qualifications and experience enhance-ment to the internal operational effectiveness of theenterprise.

• The ‘‘Staff performance’’ map (Appendix, Fig. 13) sum-marizes concepts that affect the performance assessmentpractices of the enterprise.

4.3. Integrated metrics

The proposed integrated category consists of two differ-ent maps as follows:

• The ‘‘HR practices’’ map (Appendix, Fig. 14) essentiallyrelates concepts, which fall under more than one cate-gory, as well as top-most concepts.

• The ‘‘Integrated model’’ map (Appendix, Fig. 15) pro-vides an FCM-based implementation of the dynamicinterrelationships presented in Fig. 3.

Concepts denoted as ‘‘#’’ expand top–down to lower levelmaps. Similarly ‘‘"’’ denotes bottom–up causal propagation.

4.4. Business metrics

This research also proposes maps with generic strategicperformance concepts. For example, the ‘‘Differentiationstrategy’’ map reasons on the impact of strategic changeto the competitive identity and financial status of the enter-prise, the ‘‘Customers’ satisfaction’’ map reasons on theimpact of customer appreciation of the HR improvementeffort and consequent financial value of the enterprise,the ‘‘Productivity’’ map reasons on the impact of internalHR efficiency and organizational adequacy on the overallprofitability of the enterprise and so on.

4.5. Infrastructure metrics

This research may also integrate various maps consistingof technical concepts like centralization/decentralization ofIT infrastructure, IS/IT effectiveness, etc. While the presen-tation of such concepts supports the completeness of theproposed methodology tool, they add little (if any at all) sig-nificance to the research activities presented by this paper.

All the above-mentioned maps form a generic domain ofFCMs. This knowledge domain serves as the basis of theproposed approach and can be modified to comply withthe requirements of specific knowledge capturing exercises.For example, further maps in the technical category couldrelate concepts that relate the adequacy of the productioncycle with the expected financial performance of the enter-prise (Valiris & Glykas, 2000; Irani et al., 2002).

4.6. Assigning linguistic variables to FCM weights and

concepts

In order to define weight value of the association rules inan adaptive and dynamic manner, the following methodol-ogy is proposed. HR experts are asked to describe the inter-connection influence of concepts using linguistic notions.Influence of one concept over another, is interpreted as alinguistic variable in the interval [�1,1]. Its term setT(influence) = {negatively very very high, nega-

tively very high, negatively high, negatively

medium, negatively low, negatively very low,

negatively very very low, zero, positively

very very low, positively very low, positively

low, positively medium, positively high, pos-

itively very high, positively very very high}.This paper proposes a semantic rule M to be defined at

this point. The above-mentioned terms are characterized bythe fuzzy sets whose membership functions l are shown inFig. 9.

0 0.2 0.35 10.1

0.5

1

μ

influence

µpm µps µpvs µpvvs

0.5 0.65 0.8 0.9-0.1-0.2-0.35-0.5-0.65-0.8-0.9-1

µplµpvlµpvvlµnvvlµnvlµnlµnmµnhµnvhµnvvh µz

Fig. 9. Membership functions of linguistic variable influence.

2026 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

• M (zero) = the fuzzy set for ‘‘an influence close to 0’’with membership function lz

• M (positively very very low) = the fuzzy set for‘‘influence close to 10%’’ with membership function lpvvl

• M (positively very low) = the fuzzy set for ‘‘influ-ence close to 20%’’ with membership function lpvl

• M (positively low) = the fuzzy set for ‘‘influenceclose to 35%’’ with membership function lpl

• M (positively medium) = the fuzzy set for ‘‘influ-ence close to 50%’’ with membership function lpm

• M (positively high) = the fuzzy set for ‘‘influenceclose to 65%’’ with membership function lph

• M (positively very high) = the fuzzy set for‘‘influence close to 80%’’ with membership function lpvh

• M (positively very very high) = the fuzzy setfor ‘‘influence close to 90%’’ with membership functionlpvvh

• Similarly for membership functions with negativeinfluence.

The membership functions are not of the same size sinceit is desirable to have finer distinction between grades in thelower and higher end of the influence scale. As an example,three managers/HR experts proposed different linguisticweights for the interconnection Wij from concept Ci toconcept Cj: (a) positively very low (b) positively strong(c) positively very strong. The three suggested linguisticsare integrated using a sum combination method and thenthe defuzzification method of center of gravity (CoG) isused to produce a weight wij = 0.549 in the interval[�1,1]. This approach has the advantage that experts donot have to assign numerical causality weights but todescribe the degree of causality. The same semantic ruleand term set can be used to define coefficients k1, k2 as wellas to assign values to concepts.

5. Preliminary experiments

5.1. The nature of the experiments

Two informal experiments were conducted by utilizingdecision metrics from actual HRM exercises in two major

financial sector enterprises. For each experiment, a teamof experts was engaged to:

• provide linguistic variables for the causal weights, theconcept values and the coefficients values to let theFCM algorithm reason about the impact of potentialHR performance changes,

• provide their independent expert estimates (using similarlinguistic variables) of the impact of HR practices to theoverall business model.

In both cases, the proposed tool iterated a subset ofapproximately 120 concepts spread over eight sample hier-archical maps in order to calculate their equilibriumvalues.

5.2. Discussion

5.2.1. Theoretical value

Various aspects of the proposed modeling technique arenow commented on. As far as the theoretical value is con-cerned, the proposed mechanism extends previous researchattempts by:

• allowing fuzzy node and weight definitions in the cogni-tive maps,

• introducing a specific interpretation mechanism of lin-guistic variables to fuzzy sets,

• proposing an updated FCM algorithm to suit better theHRM domain,

• supporting node linking to establish map hierarchiesand dynamic map selection during simulation,

• concentrating on the actual HR activity and the associ-ated HR model,

• allowing dynamic map decomposition andreconfiguration,

• integrating three modes of FCM simulation, namelybivalent (with a crisp activation set {0,1}), trivalent(with a crisp activation set {�1,0,1}) and linear (withan activation set in the fuzzy interval [�1, . . . , 1]).

• supporting the user with interface windows when loops,cycles and node conflicts are identified.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2027

5.2.2. Practical value

As far as the practical value of the proposed mechanismis concerned:

• When compared to the expert estimates, the mechanismdoes not provide fundamentally different ‘‘diagnosis’’.On the contrary, it provides reasonably good approxi-mations of the expert estimates.

• The justification of the ‘‘diagnosis’’ (essentially the met-rics decomposition) proved helpful in comprehendingthe sequence of complex decision interactions (essen-tially the performance roadmap).

• The concept-based decision approach did not restrict theinterpretation of the impact of HRM practices to theoverall business model and shareholder value. The fuzzyinterpretation of concept and weight values served asindications rather than precise arithmetic calculations.

• The hierarchical (or partial) traversal of performancemetrics improved the distributed decision monitoringthroughout the organizational levels of the enterpriseand stipulated targeted communication of the associatedHR performance.

• The adaptive nature of this modeling technique is alsoworth mentioning. For instance:– the approach can portray either a holistic or a scalable

view to shareholder value modeling to comply withthe decision modeling approach of the enterprise,

– knowledge categorization is compatible with eitherthe ‘‘process view’’ or the ‘‘organizational view’’ ofthe enterprise,

Fig. 10. Deploymen

– hierarchical (and/or decentralized) composition/decomposition of knowledge concepts couples effec-tively with the hierarchical (and/or decentralized)structures of an enterprise,

– skeleton decision maps accompanied by business cases(scenarios) improve the flexibility of the tool by allow-ing the user to generate alternative performance-baseddecision roadmaps with little extra effort,

– decision maps can expand/retract on demand allow-ing the user to utilize only the necessary subset of per-formance concepts,

– decision maps are dynamic, further performance con-cepts may be added to encapsulate further of knowl-edge interactions.

5.2.3. Added value

Having established the theoretical and practical value ofthe proposed mechanism, it is useful to discuss also theadded value of incorporating such a mechanism intoHRM exercises. It is the belief of this paper that the result-ing tool provides real value to the principle beneficiariesand stakeholders of such exercises. For example:

• The mechanism eases significantly the complexity ofderiving performance-based decisions at a strategiclevel. Informal experiments indicated that the timerequired by experts to estimate manually the extensiveimpact of major changes to HRM practices to the share-holder value could pose as a considerable overhead. Onthe other hand the elapsed time for automated decision

t FCMs.

2028 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

support using FCM can be insignificant, once the maphierarchies have been set up.

• To extend further this syllogism, realistic decision sup-port should involve continuous argument of changeoptions (e.g. application of best practices, alternativeHR practices, etc.) until an equilibrium solution hasbeen agreed upon. Informal discussions with the princi-ple beneficiaries of the two financial sector enterprisesrevealed that the proposed support can reduce signifi-

Fig. 11. Benefi

Fig. 12. Traini

cantly the impact estimation overheads, letting thestakeholders focus on the actual HRM exercise whileexploring in depth all alternatives and controlling effec-tively major change initiatives.

• The proposed modeling mechanism can also assist perfor-mance evaluation of the enterprise on a regular basis.FCMs may serve as a back end to performance scorecards(Bourne et al., 2000; Kaplan & Norton, 1996; Kaplan &Norton, 2001) to provide holistic strategic performance

ts FCMs.

ng FCMs.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2029

evaluation and management. However a detailed analysisof this extension falls out of the scope of this paper.

Summarizing, preliminary experimental results showedthat FCM-based management of shareholder value canbe effective and realistic. This is considered to be a majorcontribution of the proposed modeling methodology toactual decision-making exercises. Moreover, ex ante rea-soning of the impact of HRM practices to the overallshareholder model can be estimated with a moderatestart-up effort. Scenario building on the other hand canbe trivial once the skeleton FCMs have been established.

6. Conclusion

This paper presented an intelligent decision-modelingtechnique, which assessed the impact of contemporary

Fig. 14. HR practice

Fig. 13. Performance be

HRM practices to the shareholder value. This researchaddressed the problem of establishing HRM interrelation-ships in order to drive the overall business performancefrom the shareholder value perspective. The proposedmethodology offered an alternative approach to HRMbased on shareholder value modeling. The underlyingresearch addressed the problem of performance captureand representation in order to provide an implementationof an integrated HRM framework. The proposed method-ology utilized the fuzzy causal characteristics of FCMs togenerate a hierarchical and dynamic network of intercon-nected HR performance decision concepts. Also, genericmaps that supplemented the decision process presented aroadmap for integrating hierarchical FCMs into HR per-formance management techniques. The application ofFCMs as an intelligent (though soft) modeler of HRknowledge is believed to be novel.

high-level FCM.

nchmarking FCMs.

Fig. 15. Integrated model FCM.

2030 G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031

This paper extended typical FCM algorithms to adaptto the distributed nature of typical HR activities. Also, thisresearch adopted a new qualitative approach to interpretfuzzy linguistic variables to weight and concept values inorder to support further the soft computing characteristicsof the technique. It is the belief of this paper that the intel-ligent reasoning capabilities enhanced considerably the use-fulness of the mechanism while reducing the effort ofidentifying quantitative impact measurements.

The proposed mechanism should not be regarded onlyas an effective decision modeling support tool. Its mainpurpose is to drive HR change activities rather than limititself to qualitative simulations. Moreover, the proposedmechanism should not be seen as an ‘‘one-off’’ decisionaid. It should be a means for setting a course for contin-uous improvement (Langbert & Friedman, 2002).

Appendix

See Figs. 10–15.

References

Ansoff, H. I. (1975). Managing strategic surprise by response to weaksignals. California Management Review, XXVIII(2), 21–33.

Axelrod, R. (1976). Structure of decision: The cognitive maps of political

elites. Princeton, NJ: Princeton University Press.Baker, D. (1999). Strategic human resource management: perfor-

mance, alignment, management. Library Career Development, 7(5),51–63.

Bellone, M., Merlino, M., et al. (1995). A DSS for personnel careermanagement. Decision Support Systems, 15(3), 219–227.

Bennett, N., Ketchen, D. J., et al. (1998). Examination of factors

associated with the integration of human resource management and

strategic decision making. Human Resource Management (Vol. 37(1)).Spring, pp. 3–16.

Bourne, M., Mills, J., et al. (2000). Designing, implementing and updatingperformance measurement systems. International Journal of Operations

and Production Management, 20, 754–771.Byun, D., & Suh, E. (1994). Human Resource Management Expert Systems

Technology. Expert Systems with Applications, 11(2), 109–119.Cook, D. S., & Ferris, G. R. (1986). Strategic human resource manage-

ment and firm effectiveness in industries experiencing decline. Human

Resource Management, 25, 441–458.Craiger, J. P., Goodman, D. F., et al. (1996). Modeling organizational

behavior with fuzzy cognitive maps. Journal of Computational Intel-

ligence and Organisations, 1, 120–123.Dickerson, J. A., & Kosko, B. (1997). Virtual worlds as fuzzy cognitive

maps. In B. Kosko (Ed.), Fuzzy engineering (pp. 125–141). UpperSaddle River, NJ: Prentice-Hall.

Diffenbach, J. (1982). Influence diagrams for complex strategic issues.Strategic Management Journal, 3, 133–146.

Drucker, P. F. (1993). Harper Collins Publishers.Georgopoulos, V., Malandraki, G., et al. (2002). A fuzzy cognitive map

approach to differential diagnosis of specific language impairment.Journal of Artificial Intelligence in Medicine, 679, 1–18.

Grove, A. (1999). Only the paranoid survive: how to exploit the crisis points

that challenge every company. Bantam Book.Hagiwara, M. (1992). Extended fuzzy cognitive maps. In Proceedings of

the 1st IEEE International Conference on Fuzzy Systems, New York.Handy, L., Barham, K., et al. (1989). Beyond the personnel function: The

strategic management of human resource. Journal of European

Industrial Training, 13(1).Hong, T., & Han, I. (2002). Knowledge based data mining of news

information of the internet using cognitive maps and neural networks.Journal of Expert Systems with Applications, 23(1), 1–8.

Huselid, M. A. (1995). The impact of human resource managementpractices on turnover, productivity, and corporate financial perfor-mance. Academy of Management Journal, 38, 635–672.

G. Xirogiannis et al. / Expert Systems with Applications 35 (2008) 2017–2031 2031

Irani, Z., Hlupic, V., et al. (2002). Re-engineering manufacturingprocesses through simulation modeling. Journal of Logistics Informa-

tion Management, 13(1), 7–13.Johnson, R. J., & Briggs, R. O. (1994). A model of cognitive information

retrieval for ill-structured managerial problems and its benefits forknowledge acquisition. In 27th Annual Hawaii International Conference

on System Sciences, Hawaii.Kaplan, R. S., & Norton, D. P. (1996). Using the balanced scorecard as a

strategic management system (pp. 75–85). Harvard Business Review.January/February.

Kaplan, R. S., & Norton, D. P. (2001). Leading change with the balancedscorecard. Financial Executive, 17, 64–66.

Kavanaugh, M. J., Gueutal, H. G., et al. (1990). Human Resource

Information Systems: Development and Application.Kelly, K. (1998). New rules for the new economy: ten ways the network

economy is changing everything. London: Fourth Estate.Klein, J. C., & Cooper, D. F. (1982). Cognitive maps of decision

makers in a complex game. Journal of Operation Research Society,

33, 63–71.Kosko, B. (1986). Fuzzy cognitive maps. Journal of Man–Machine Studies,

24, 65–75.Kosko, B. (1991). Neural networks and fuzzy systems. Englewood Cliffs:

Prentice Hall.Kwahk, K. Y., & Kim, Y. G. (1999). Supporting business process redesign

using cognitive maps. Decision Support Systems, 25(2), 155–178.Lado, A. A., & Wilson, M. C. (1994). Human resource systems and

sustained competitive advantage: A competency-based perspective.Academy of Management Review, 19, 699–727.

Langbert, M., & Friedman, H. (2002). Continuous improvement in thehistory of human resource management. Journal of Management

Decision, 40(8), 782–787.Lawler, J. J., & Eliot, R. (1996). Artificial intelligence in HRM: An

experimental study of an expert system. Journal Of Management, 22(1),85–111.

Lee, K. C., & Kim, H. S. (1997). A fuzzy cognitive map-based bi-directional inference mechanism: An application to stock investmentanalysis. Journal of Intelligent Systems in Accounting Finance &

Management, 6(1), 41–57.Lee, K. C., & Kim, H. S. (1998). Fuzzy implications of fuzzy cognitive

map with emphasis on fuzzy causal relationship and fuzzy par-tially causal relationship. Journal of Fuzzy Sets and Systems, 3,303–313.

Lee, K. C., Kim, J. S., et al. (2002). Fuzzy Cognitive Map approach toweb-mining inference amplification. Journal of Expert Systems with

Applications, 22, 197–211.Lee, K. C., & Kwon, O. B. (1998). A strategic planning simulation based

on cognitive map knowledge and differential game. Journal of

Simulation, 7(5), 316–327.Leopold, J., Lynette, H., et al. (1999). Strategic human resourcing-

principles, perspectives and practices. London, Financial Times: PitmanPublishing.

Liu, Z. Q. (2000). Fuzzy cognitive maps: Analysis and extension. Tokyo:Springer.

Liu, Z. Q., & Satur, R. (1999). Contexual fuzzy cognitive map for decisionsupport in geographical information systems. Journal of IEEE

Transactions on Fuzzy Systems, 7, 495–507.Mohanty, R. P., & Deshmukh, S. G. (1997). Evolution of a decision

support system for human resource planning in a petroleum company.International Journal of Production Economics, 51(3), 251–261.

Ndouse, T. D., & Okuda, T. (1996). Computational intelligence fordistributed fault management in networks using fuzzy cognitive maps.In IEEE International Conference Communication Converging Tech-

nology Tomorrow’s Application, New York.Niehaus, R. J. (1995). Evolution of the strategy and structure of a human

resource planning DSS application. Decision Support Systems, 14(3),187–204.

Noh, J. B., Lee, K. C., et al. (2000). A case- based reasoning approach tocognitive map driven -driven tacit knowledge management. Journal of

Expert Systems with Applications, 19, 249–259.Parenthoen, M., Reignier, P., et al. (2001). Put fuzzy cognitive maps to

work in virtual worlds. In Fuzz-IEEE’01 Proceedings, Australia, IEEE.Pelaez, C. E., & Bowles, J. B. (1995). Applying fuzzy cognitive maps

knowledge representation to failure modes effect analysis. In IEEE

Annual Reliability Maintainability Symposium, New York.Perusich, K. (1996). Fuzzy cognitive maps for political analysis. In

International Symposium on Technology and Society. Technical Exper-tise and Public Decisions, New York.

Poe, V., Klauer, P., et al. (1998). Building A Data warehouse for DecisionSupport.

Ramaprasad, A., & Poon, E. (1985). A computerised interactive techniquefor mapping influence diagrams (MIND). Strategic Management

Journal, 6(4), 377–392.Ritson, N. (1999). Corporate strategy and the role of HRM: critical cases

in oil and chemicals. Employee Relations, 21(2), 159–175.Schendel, D. E., & Hofer, C. W. (1979). Strategic Management: A New

View of Business Policy and Planning.Schneider, M., Schnaider, E. et al. (1995). Constructing fuzzy cognitive

maps. In IEEE Conference on Fuzzy Systems, New York.Schuler, R. S., & Jackson, S. E. (1996). Human resource management:

Positioning for the 21st century.Silva, P. C. (1995). New forms of combinated matrices of fuzzy cognitive

maps. In IEEE International Conference on Neural Networks, New York.Stefik, M. (1990). Introduction to Knowledge Systems.Sturman, M. C., Hannon, J. M., et al. (1996). Computerised decision aids

for flexible benefits decision: The effects of an expert system anddecision support system on employee intentions and satisfaction withbenefits. Personnel Psychology, 49, 883–908.

Stylios, C. D., Georgopoulos, V. C. et al. (1997). Introducing the Theoryof Fuzzy Cognitive Maps in Distributed Systems. In 12th IEEE

International Symposium on Intelligent Control, Istanbul, Turkey.Tannenbaum, S. I. (1990). Human resource information systems: User

group implications. Journal of Systems Management, 41(1), 27–32.Ulrich, D. (1997). Measuring human resources: an overview of practice and

a prescription for results. Human Resource Management (Vol. 36(3)).Spring, pp. 303–320.

Ulrich, D. (1998). A new mandate for human resources (pp. 124–134).Harvard Business Review January–February 1998.

Valiris, G., & Glykas, M. (2000). A Case Study on ReengineeringManufacturing Processes and Structures. Journal of Knowledge and

Process Management, 7(1), 20–28.Walls, J. G., Widmeyer, G. R., et al. (1992). Building an information

system design theory for vigilant EIS. Information System Research,

3(1), 36–59.Watson, H. J., Houdeshel, G. et al. (1997). Building Executive Informa-

tion Systems and Other Decision Support Applications.Wright, P. M., & McMahan, G. C. (1992). Theoretical perspectives for

strategic human resource management. Journal of Management, 18,295–320.

Xirogiannis, G., & Glykas, M. (2004). Fuzzy cognitive maps as a back endto knowledge-based systems in geographically dispersed financialorganizations. Knowledge and Process Management, 11, 1–18.

Xirogiannis, G., & Glykas, M. (2004). Fuzzy cognitive maps in businessanalysis and performance-driven change. IEEE Transactions on

Engineering Management, 13(17), 111–136.Xirogiannis, G., Stefanou, J., et al. (2004). A fuzzy cognitive map

approach to support urban design. Journal of Expert Systems with

Applications, 26(2), 257–268.Zadeh, L. A. (1965). Fuzzy sets. Journal of Information and Control, 8,

338–353.Zhang, W. R., Wang, W., et al. (1994). A-pool: An agent-oriented open

system for distributed decision process modeling. Journal of Organi-

sational Computing, 4(2), 127–154.