Expert Systems with Applications 39 (2012) 142153

Contents lists available at ScienceDirect

Expert Systems with Applications

Logistics tool selection with two-phase fuzzy multi criteria decision making: A case study for personal digital assistant selection
Glin Bykzkan a,, Jbid Arsenyan b, Da Ruan c
_ Department of Industrial Engineering, Galatasaray University, ragan Caddesi No: 36, Ortaky, Istanbul 34357, Turkey Department of Industrial Engineering, Bahesehir University, ragan Caddesi Osmanpasa Mektebi Sokak No: 4 6, Besiktas, Istanbul 34100, Turkey _ c The Belgian Nuclear Research Centre (SCKCEN), Boeretang 200, Mol 2400, Belgium
a b

a r t i c l e

i n f o

a b s t r a c t
Efcient logistics and supply chain management are enabled through the use of efcient information technologies (IT). The mobile logistics tools represent the IT interface in the supply chain. This paper aims to aid decision makers to identify the most appropriate mobile logistics tools and to achieve this aim, several evaluation criteria are identied to evaluate logistics tools, and a fuzzy axiomatic design (FAD) based group decision-making method is adopted to perform the evaluation in two phases. In the rst phase of pre-assessment, alternatives that cannot meet basic requirements and the dened threshold are eliminated. In the second phase of selection, the remaining alternatives are more meticulously evaluated. Criteria weights are determined using fuzzy analytic hierarchy process (AHP) and another fuzzy multicriteria decision-making (MCDM) technique, namely fuzzy technique for order preference by similarity to ideal solution (TOPSIS), is applied in the second phase to compare the outcome of FAD. A case study is provided in order to demonstrate the potential of the proposed methodology. Personal digital assistants (PDAs) with integrated barcode scanner that are available in the Turkish market are evaluated. 2011 Elsevier Ltd. All rights reserved.

Keywords: Fuzzy axiomatic design Group decision-making Fuzzy AHP Fuzzy TOPSIS Logistics tool selection Logistics industry

1. Introduction Rapid development of mobile communication technology holds an increasing importance in supply chain and logistics industry. Mobility generates new opportunities to improve operations in logistics applications. As real time data is crucial in supply chain management, mobile logistics tools are widely used and popular in logistics activities such as warehouse automation, distribution, inventory control and tracking. Mobile technologies have made possible for people to work from anywhere at any time via wireless communication network. With the emerging demand, various models with cutting edge technology and diverse features such as barcode scanner, RFID technology, etc. are entering the market. Selecting the right tool among different product families with different specications is an important decision problem for logistics companies. This paper proposes a two-phase decision framework that provides an effective evaluation of the mobile logistics tool alternatives. The evaluation criteria for mobile logistics tool are suggested and a two-phase fuzzy multi-criteria decision making (MCDM) with group decision methodology is introduced. Also a case study is proposed to evaluate the personal digital assistants (PDAs) with integrated barcode scanner, which are hand-held

Corresponding author. Tel.: +90 212 227 4480.

computers used for mobile data applications in different industries to improve the effectiveness of the organizations (Hoffer, 2005; Lee, Cheng, & Cheng, 2007; Madria, Mohania, Bhowmick, & Bhargava, 2002). MCDM is a powerful tool widely used for evaluating and ranking problems containing multiple, usually conicting criteria (Pomerol & Romero, 2000). Given the multi-dimensional characteristics of a mobile technology, MCDM provides an effective evaluation framework for mobile logistics tools. MCDM is one of the rare operations research techniques where decision makers (DMs) are involved in the solution process. This feature brings two signicant advantages: the problem can easily be structured corresponding to DMs requests and the results are adopted more easily by DMs, because, they are actively involved in the solution procedure (Masud & Ravindran, 2007). Literature offers many applications of axiomatic design (AD) methodology to design products, systems, organizations and software (Suh, 2001). However, AD principles also provide a powerful tool to measure how well system capabilities respond to functional requirements (FRs) and therefore, presents an opportunity for MCDM. AD consists of two axioms: the independence axiom and the information axiom. The latter proposes the selection of a proper alternative that has minimum information content. The conventional information axiom approach is employed in many applications such as decision making in design (Harutunian, Nordlund, Tate, & Suh, 1996), design for manufacturing (Goncalves-Coelho &

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

143

Mourao, 2007), and marine design (Jang, Yang, Song, Yeun, & Do, 2002). However, it cannot be used with incomplete information, since the expression of decision variables by crisp numbers would be ill dened (Kahraman & Kulak, 2005). The experts evaluate the importance of evaluation criteria or the performance of the providers with uncertain, imprecise, or subjective judgments (Liu & Wang, 2009). Hence, the subjectivity and vagueness in the assessment process are dealt with fuzzy logic (Zadeh, 1975). Multiple DMs are often preferred rather than a single DM to avoid any possible bias and to minimize the partiality in the decision process (Herrera, HerreraViedma, & Chiclana, 2001). Group decision-making is thus another important concern in this study. In order to integrate various experiences, opinions, ideas, and motivations of each DM, it is preferable to convert the linguistic estimation into fuzzy numbers, making fuzzy logic essential in resolving the problems of group decision-making (Lin & Wu, 2008). In consequence, a fuzzy axiomatic design (FAD) based group decision-making approach is applied to evaluate the alternatives according to the criteria considered and rank them in an efcient way. The methodology includes two phases. In the rst phase, which is a pre-assessment phase, the alternatives are evaluated by the group of DMs with respect to the main criteria, and FAD is applied to assess the alternatives. The rst phase aims to consider as many available tools as possible and to eliminate the ones that cannot meet the basic requirements. Also, a threshold is dened in the pre-assessment phase in order to reduce the alternative set that undergoes the detailed evaluation. In the second phase of selection, the remaining alternatives are evaluated more meticulously by the same group of DMs with respect to the sub-criteria, and since the structure of the problem is hierarchical, the hierarchical FAD methodology (Kahraman & Cebi, 2009) is applied to determine the scores of the alternatives. Both phases and the pre-evaluation process where alternatives, criteria, and criteria weights are determined, include the same DMs in order to assure consistency of the judgments. Finally, a case study evaluating the PDA barcode scanners is given to demonstrate the potential of the methodology. Another MCDM technique, namely fuzzy technique for order preference by similarity to ideal solution (TOPSIS), is applied to compare the outcome of the second phase given that FAD and Fuzzy TOPSIS operate on the geometrical principles. As the study includes incomplete information due to subjective judgments, and FAD is preferred to the conventional AD, Fuzzy TOPSIS is also preferred to the conventional TOPSIS in order to assure coherence in a fuzzy environment. This papers contribution lays in the two-phase pre-assessment and selection system that uses integratedly Fuzzy MCDM techniques with group decision. FAD emerges as an efcient fuzzy MCDM methodology. On the other hand, Fuzzy Delphi and fuzzy analytic hierarchy process (AHP) are commonly used Fuzzy MCDM techniques and these techniques are employed in the rst phase to calculate criteria weights. However, these techniques are generally used separately and for different purposes in various decision problems. This paper integrates these techniques in one methodology, aggregating subjective judgments of the DMs with Fuzzy Delphi, calculating the criteria weights with Fuzzy AHP and nally, assessing, eliminating and selecting the alternatives with FAD. The proposed methodology benets from the strength of each technique in order to solve the multi-criteria problem in hand in the most efcient way. Also, the selection of mobile logistics tools seems to be a neglected subject in the logistics literature. This paper applies the proposed methodology to the PDA Barcode Scanner selection problem with a case study in Turkey. The rest of the paper is organized as follows. Section 2 describes the details of the proposed evaluation methodology and the techniques employed in the study. Section 3 introduces the evaluation criteria for the mobile logistics tools. A case study is given in Section 4 with all detailed steps included to validate the model and to examine its effectiveness. Section 5 is dedicated to the comparison of the

FAD outcome with another Fuzzy MCDM technique and to the sensitivity analysis. Section 6 concludes the paper with some remarks.

2. Methodology MCDM has proven to be an effective methodology for solving a large variety of multi-criteria evaluation and ranking problems (Hwang & Yoon, 1981). MCDM techniques are applied in many different areas such as strategic decisions, economic evaluation, and technological investment. These techniques are also employed in planning, evaluating, and selecting information technologies (IT). In evaluating the decision alternatives in new problem settings, the assessment data for the criteria weights and for the performance ratings of the alternatives on qualitative criteria are often not available and have to be assessed subjectively by the decision makers, the stakeholders or the experts (Yeh & Chang, 2009). The subjective assessment process involves two types of judgment. Comparative judgments are often used for identifying the importance of criteria whereas absolute judgments are generally used to evaluate the alternatives. On the other hand, both comparative and absolute judgments include imprecision and subjectivity as it involves decision makers thoughts and bias. Fuzzy sets are used in this paper in order to deal with the imprecision and subjectivity involved in the evaluation process (Zadeh, 1975) and manipulates uncertain criteria of the problem (Royes, Bastos, & Royes, 2003). Fuzzy MDCM provides means to evaluate decision alternatives using subjective judgments. In this paper DMs opinions are gathered using linguistic terms and fuzzy scales are employed to translate the linguistic opinions into triangular fuzzy numbers. On the other hand, as the evaluation criteria become more intangible and decisions become more complex for one DM to make, the methodology utilizes group decision making where aggregation operations play an important role (Musilek, Guanlao, & Barreiro, 2005). Fuzzy MCDM with group decision, where individual judgments of the DMs are aggregated, is increasingly employed in literature. For example, Chen and Cheng (2005) apply fuzzy MCDM with group decision to information systems personnel selection. Wang and Parkan (2008) considers the fuzzy preference aggregation problem in group decision and they apply it to the broadband internet service selection. Recently, Yeh and Chang (2009) have developed a hierarchical weighting method to assess the weights of a large number of evaluation criteria by pairwise comparisons. Dagdeviren, Yavuz, and Klnc (2009) use the geometric mean of the pairwise comparisons obtained from individual evaluations for AHP in a weapon selection problem. A two-phase elimination and selection methodology using MCDM is proposed in this paper. Main steps of the proposed methodology are recapitulated in Fig. 1. The methodology is basically composed of three sections. During the pre-evaluation process, DMs are gathered in order to establish the MCDM parameters; the selection criteria are determined and categorized to establish a hierarchy. Afterwards, Fuzzy AHP methodology is employed in order to calculate the main criteria and sub-criteria weights. Also, the alternatives are identied. The rst phase of the methodology, the pre-assessment phase, includes multiple DMs determining design and system ranges in linguistic terms. These linguistic terms are then fuzzied and aggregated in order to undergo the FAD methodology. Also, a threshold is dened by the group of DMs. The pre-assessment phase eliminates the alternatives in two ways:  First elimination occurs with the design range. Alternatives that cannot meet the limits set by the design range have unlimited information content, and therefore, are removed from the alternatives set.

144

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

Fig. 1. Main steps of the proposed methodology.

 Second elimination occurs with the threshold. DMs determine a threshold on the number of alternatives or information content in order to perform a more detailed and thorough evaluation in respect to the sub-criteria with fewer alternatives. The second phase of the methodology is basically the same with the rst phase, except that it involves fewer alternatives and more criteria, as the evaluation in the selection phase is performed with the sub-criteria. Again, fuzzy MCDM with group decision is employed. DMs opinions are aggregated, and then, FAD is performed. The alternatives are ranked in increasing order of information content. Finally, another fuzzy MCDM technique is employed on the DMs evaluations in order to compare the outcome of the FAD technique and underline the advantages and the drawbacks of the proposed methodology.

team of analysts who have a rich knowledge and expertise in logistics activities different, including experts from the organization such as sales, marketing, manufacturing, nance, and logistics, determine all possible evaluation criteria prior to provider selection (Liu & Wang, 2009). In this study the criteria is provided beforehand and the criteria is determined by the experts of the organization.

2.2. Determining the criteria weights Group decision is employed in determining criteria weights, as well as in the next steps of this study. An aggregation method is used to combine DMs judgments and opinions. These aggregated judgments are then employed in fuzzy MCDM techniques. Fuzzy Delphi method is employed as the aggregation method. Fuzzy aggregation enables the handling of linguistic as well as ordinary quantitative information to deal with the multi-criteria decision making problem (Hung, Julian, Chien, & Jin, 2010). The technique is adapted from Buyukozkan and Ruan (2008). On the other hand, the technique used to calculate the criteria weights is Fuzzy AHP, given that AHP is particularly useful for evaluating complex multi-attribute alternatives involving subjective criteria (Kwong & Bai, 2003). AHP solves complex decision problems that may have correlations among decision criteria based on three principles: decomposition, comparative judgments and synthesis of priorities (Chamodrakas et al., 2010). As independence axiom of AD guarantees the independence of the FRs, AHP is a suitable technique as the independence of evaluation criteria is also assured.

2.1. Determining and categorizing the logistics tool criteria The rst step in the methodology is to determine the main criteria and sub-criteria for logistics tool performance evaluation. These criteria are collected through literature review and veried by the industrial experts. The criteria are then categorized in a hierarchical structure. In some research, the customer selects the appropriate qualitative and quantitative criteria for his/her organization given that there is no common identication of factors guiding the supplier selection process and that decision criteria vary in relation with various characteristics of the buyer organization (Chamodrakas, Batis, & Martakos, 2010). In other researches, a

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

145

AHP (Saaty, 1980), which is a widely used MCDM method, offers the opportunity to tackle the complexity of the decision problem by means of a hierarchy of the decision layers. The model enables DMs to divide a decision into smaller parts, starting from the level of goal formulation into criteria, down to the level of the alternative control actions. Yet for the subjective judgments, a theory is needed to measure the ambiguity of these concepts. Because of the vagueness and uncertainty on judgments of the DMs, the crisp pairwise comparison in the conventional AHP seems to be insufcient and imprecise to capture the correct judgments of DMs. Fuzzy logic (Zadeh, 1975) can be introduced in the pairwise comparison of the AHP to make up for this deciency in the conventional AHP, referred to as Fuzzy AHP. Fuzzy AHP is one of the most commonly used MCDM methods and it takes the vagueness and uncertainty on judgments of DMs into consideration (Ayag & Ozdemir, 2007). Therefore, fuzzy extension of the AHP methodology is employed in this study to calculate criteria weights. 2.3. Determining the alternatives Determining the alternatives involves merely the identication of the basic requirements for the tool in question. A generic requirement set is dened in order to describe the tool with all the main characteristics. Larger the requirement set gets, smaller becomes the alternative set. All the available tools that have the dened characteristics are included in the alternative set. 2.4. Pre-assessment: fuzzy axiomatic design with group decision making In the pre-assessment phase, PDA alternatives and FRs are evaluated by the group of DMs using linguistic terms. The opinions are aggregated by the Fuzzy Delphi method and the evaluation is performed by FAD considering only the main criteria, because the preassessment phase involves a preliminary appraisal of alternatives. AD, a systematic method offering a scientic base for design, was introduced by Suh (1990) and its application areas include software design, quality system design, general system design, manufacturing system design, e-commerce strategies, ergonomics, engineering systems, and ofce cell design. AD is based on two axioms. Independence axiom states that the independence of FRs should be maintained and information axiom states that among the designs that satisfy the FRs, the design with the minimum information content is the best design. On the other hand, the information content, on which MCDM technique is based, represents a function of probability of satisfying a FR. Therefore, the design with the highest probability to meet these requirements is the best design. Information content Ii of a design with a probability of success pi for a given FRi is dened as follows:

Fig. 2. System-design ranges and common area in crisp environment.

The probability of success pi is equal to the common area Ac. Consequently, the information content can be expressed as follows:

Ii log2

  1 : Ac

Also, if the probability distribution function is uniform, the probability of success becomes:

pi

common range : system range   system range : common range

Therefore, the information content can also be written as:

Ii log2

Ii log2

  1 : pi

According to Suh (2001), logarithm is employed to calculate information contents for obtaining additivity. The probability of success is given by the design range (the requirements for the design) and the system range (the system capacity). Fig. 2 illustrates the design and system ranges as well as the common area. The intersection of the ranges offers the feasible solution. Therefore, the probability of success can be expressed as:

However, the conventional information content approach cannot be used with incomplete information, because under incomplete information, the expression of the system and design ranges by crisp numbers would be ill dened (Kahraman & Kulak, 2005). For this reason, the subjectivity and vagueness in the assessment process are dealt with the fuzzy logic (Zadeh, 1975). The information axiom of AD is utilized as a fuzzy MCDM technique by Kulak and Kahraman (2005). FAD applications for fuzzy MCDM can be summarized as shown in Table 1. Even though FAD literature on fuzzy MCDM shows various applications, no previous study exists on the mobile logistics tool evaluation and there are only a few applications of FAD in supply chain domain. The FAD methodology is based on the conventional AD. However, the crisp ranges are replaced by fuzzy numbers that represent linguistic terms as seen in Fig. 3. In this study, the triangular fuzzy numbers (TFNs) are employed. Information content is calculated as in a non-fuzzy environment. Intersection of TFNs representing the design and system ranges presents the common area (Kulak & Kahraman, 2005). Information content in a fuzzy environment is calculated as follows:

Ii

1; no intersection;log 2

area of system range common area

 ; otherwise;

pi

Z
l

pFRi dFRi ;

In this study, the weighted information content calculation of the hierarchical FAD is adapted from Kahraman and Cebi (2009). This model requires the determination of the weights of the criteria and sub-criteria. Total weighted information content for the rst level criteria is calculated as follows:

where l and u represent the lower and upper limits of the design range and where p represents the probability density function of the system for a given FRi.

n X i1

wi Ii ;
Pn
i1 wi

where n is the number of the rst level criteria and

1.

146 Table 1 FAD literature. Authors Kulak and Kahraman (2005) Kulak (2005) Kulak, Durmusoglu, and Kahraman (2005) Eraslan, Akay, and Kurt (2006) Cebi and Celik (2007) Ozel and Ozyoruk (2007) Cebi and Celik (2008) Cebi, Celik, Er, and Kahraman (2008) Ycel and Aktas (2008) Celik and Er (2009) Celik (2009) Celik, Kahraman, Cebi, and Er (2009) Celik, Cebi, Kahraman, and Er (2009) Cevikcan, Cebi, and Kaya (2009) Cicek and Celik (2009) Kahraman and Cebi (2009) Cebi and Kahraman (2010a) Cebi and Kahraman (2010b)

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

Application area Selection among transportation companies Choice of material handling equipments Multi-attribute equipment selection Ranking of intercity bus passenger seats Measuring customer satisfaction at ports Supplier decision Ship machinery installation Ship design project approval Evaluation for ergonomic design of electronic consumer products Model selection paradigm Integrated environmental management system (IEMS) design Shipyards docking performance evaluation model Strategy making towards container port development Comparison of fuzzy VIKOR and fuzzy axiomatic design versus to fuzzy TOPSIS and application to candidate assessment Selection of porous materials in marine system design Teaching assistant selection problem Discussions on the adaptation of the current AD principles into fuzzy sets theory Design of indicator panel for passenger cars based on the AD principles under fuzzy environment

2.4.3. Determining a threshold for the elimination phase A threshold is determined by the group of DMs for the preassessment phase. The threshold may be dened using either one of the following two ways:  A limit can be set on the number N of alternatives to undergo the detailed evaluation. This limit is dened by consensus and once the evaluation is made, the rst N alternatives with minimum information content are selected to undergo the selection phase.  A limit on the information content can be set for the alternatives to undergo the detailed evaluation. The information content is calculated as a function of the intersection of the design range and the system range. Therefore, the gap between the upper vertices of the triangular fuzzy numbers can be limited by dening a threshold on the information content, i.e., the maximum gap between the desired performance and the actual performance of the alternative can be dened. The threshold should be dened according to the level of technology and the number of alternatives available. For example, if the investment in question seizes a high-level cutting-edge technology, the alternatives would be relatively fewer and the threshold cannot be highly severe in the elimination phase. However, if the technology to acquire is ordinary and well-spread, then the numerous alternatives available in the market can be eliminated by a strict threshold. Hence, a threshold on the number of alternatives is suggested when a common technology is in question, whereas a threshold on the information content is preferable for the highlevel technology. 2.4.4. Evaluation of the alternatives with the main criteria and elimination of the unsuitable ones Once the information contents are calculated, alternatives are ranked in decreasing order of the information content. First, the alternatives with innite information content are eliminated automatically, by the methodology itself. Then, the threshold is employed to eliminate more unsuitable alternatives according to the opinion of the group of DMs. 2.5. Selection: hierarchical fuzzy axiomatic design with group decision making In this phase, the remaining alternatives are evaluated considering the sub-criteria, therefore hierarchical FAD is applied and

Fig. 3. System-design ranges and common area in a fuzzy environment.

Likewise, the total weighted information content for the second level criteria (sub-criteria for criterion i) is calculated as follows:

Ii

m X j1

wij Iij :

where m is the number of sub-criteria for criterion i and Pm for i 1; . . . ; n. j1 wij 1; The lower level information contents are calculated similarly. According to the information axiom, the alternatives are ranked with an increasing order of information content. Alternatives that cannot meet the FRs, which means the information content is innite, are eliminated.

2.4.1. Determining the design range for the main criteria The design range for the main criteria determines the acceptable interval for the DMs. Preferably, a lower limit is set for the benet criteria and an upper limit is set for the cost criteria. Nevertheless, an interval may also be set for all criteria. With FAD, linguistic terms are employed to determine the design range.

2.4.2. Determining the system range for the alternatives Determination of the system range includes the evaluation of each alternative by the DMs. In FAD, these evaluations are expressed as linguistic terms. Nevertheless, for the tangible criteria, the evaluations can be expressed as crisp numbers as well.

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

147

the functional requirements are dened as functions of sub-criteria. The opinions are again aggregated by Fuzzy Delphi. 2.5.1. Determining the design range for the sub-criteria Similar to the determination of the design range for the main criteria, the DMs are involved in dening an interval, a lower limit or an upper limit for each sub-criterion determined for each requirement. 2.5.2. Determining the system range for the remaining alternatives Similar to the determination of the system range for the alternatives, the DMs evaluate the alternatives with linguistic terms according to the sub-criteria. The linguistic terms are then translated into fuzzy numbers. 2.5.3. Re-evaluation of the alternatives with sub-criteria Information contents for each alternative according to each subcriterion are calculated and hierarchical FAD is applied to determine the weighted information contents. 2.5.4. Ranking of the alternatives The last step of FAD methodology is to rank the alternatives in decreasing order of information content and to select the best tool. Information contents display how well the alternative is responding to all the requirements. The alternative with the minimum information content is therefore the best alternative. 3. Evaluation criteria for mobile logistics tools A typical supply chain is a complex mixture of actors who need coordination, collaboration, and information exchanges among them in order to increase productivity and efciency (MartinezSala, Egea-Lopez, Garcia-Sanchez, & Garcia-Haro, 2009). The use of technology to drive business innovation, enhance customer service, and utilize available resources has become a strategic necessity in this respect (Chen, Yen, & Chen, 2009). Mobile logistics tools provide the exibility and the functionality required in supply chains. Mobile stands for positioning a mobile device as any terminal that can be used on the move like a PDA (Wamba, Lefebvre, Bendavid, & Lefebvre, 2008). Factors affecting successful technology adoption include technological factors, market-related factors, socio-economic factors, regulatory factors, and factors related to internal organization (Jeong, Yoo, & Heo, 2009). On the other hand, according to innovation diffusion theory, ve signicant innovation characteristics (relative advantage, compatibility, complexity, trial ability, and observables) are used to explain the user adoption and decision making process (Wu & Wang, 2005). Jeong et al. (2009) state that, as the market becomes more competitive and the customers expectations regarding services or products increase, the successful adoption of a new technology becomes more challenging. In todays market, there exist hundreds of IT applications for maintenance management purposes, and choosing the most appropriate one is a challenge unless the requirements are dened (Kans, 2008). In consequence, in order to dene the expectations from mobile logistics tools, ve main criteria and 14 sub-criteria are determined based on the literature review and the DMs suggestions. These criteria are displayed in Table 2. Product characteristics, functionality, cost, after sales services, and brand reliability are identied as the main criteria. Product characteristics involve physical characteristics including size, screen size, weight, design, etc., technical characteristics such as memory, operating system, etc., and safety standards including certication and warranty. On the other hand,

functionality represents the intangible product characteristics that cannot be measured but perceived, such as ease of use and exibility. Adaptability refers to the tool capability to adapt to various systems and functionality implicates the usefulness of the functions provided. According to the behavioral decision theory, the cost-benet pattern is signicant to both perceived usefulness and ease of use (Wu & Wang, 2005). However this study considers functionality and cost as completely independent categories. For PDA barcode scanners, the operating cost and product price are very low compared to major investments. However, as the tool in question represents a corporate investment, the cost is nevertheless considered as an important factor. According to Wu and Wang (2005), consumers must deal with non-negligible costs in switching between different brands of products or relative services in various markets. After Sales Service and Brand Reliability are the two remaining categories, comparable yet dissimilar. In general, the vendor and the brand are two separate companies and therefore, two different considerations have to be made in order to evaluate both the manufacturer and the vendor. After Sales Services category includes technical support of the vendor, vendor reputation and vendor capacity to provide service. Provision of after-sales services is one way that rms can differentiate themselves from competitors and these services are important in terms of expectations, protability, and customer loyalty (Zackariasson & Wilson, 2002). On the other hand, Brand Reliability includes market share of the manufacturer and brand reputation, which is different from vendor reputation as it represents the reputation engendered by the manufacturing rm.

4. Case study Logistics and supply chain is a rapidly growing industry in Turkey. Several enterprises are integrating automation systems into their activities. The integration of wireless technologies to support business processes within a supply chain could have a signicant impact on overall business operations, leading to competitive advantages in terms of cost reduction, supply chain responsiveness, and performance of supply chain functions, and therefore impact the strategic management of all rms involved in the supply chain (Wamba et al., 2008). Moreover, investments in IT have a positive correlation to company protability and competitiveness, thus indicating that IT is of strategic importance and poor investment decisions could result in failure, leading to costly consequences (Kans, 2008). This paper considers a PDA Barcode Scanner selection problem for a logistics company ABC to be employed in their regional warehouses and headquarters. Founded in 1990 as a transportation company, ABC is a leading logistics rm with 26 branches and 1500 corporate customers. They offer warehouse and distribution management solutions as well as freight solutions. ABC aims to improve its logistics performance by using mobile technologies. The company thus demands a purchase of 24 PDAs, which results in a large amount that can be considered as a strategic investment. Moreover, the company intends to increase eventually the number of its warehouses, consequently more PDAs are to be purchased potentially. The problem presents similarities to the supplier selection problem in supply chain system, which is in fact a group decision making combination of several and different criteria with different forms of uncertainty, and hence, requires MCDM methods for an effective problem-solving (Sanayei, Mousavi, & Yazdankhah, 2010). As mentioned in Fig. 1 of Section 2, two-phased pre-assessment and selection methodology is proposed to effectively decide on the best PDA Barcode Scanner alternative for the rm. First, pairwise comparisons are obtained from three DMs (IT manager,

148 Table 2 Mobile logistic tool evaluation criteria. Criteria C1 C2 Product characteristics Functionality Denition

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

Sources Tam and Tummala (2001), Lee et al. (2007), Isklar and Bykzkan (2007), Lin, Wang, Chen, and Chang (2008) Lehmann and Oshaughnessy (1982), Isklar and Bykzkan (2007), Lee et al. (2007), Lin et al. (2008), Chen et al. (2009), Jeong et al. (2009), Qi, Li, Li, and Shu (2009), Lee (2009), Sanayei et al. (2010) Lehmann and Oshaughnessy (1982), Tam and Tummala (2001), Bei, Wang, and Hu (2006), Isklar and Bykzkan (2007), Liu and Wang (2009), Lee (2009), Sanayei et al. (2010) Tam and Tummala (2001), Zackariasson and Wilson (2002), Hoffer (2005), Bei et al. (2006), Liu and Wang (2009), Lee (2009) Lehmann and Oshaughnessy (1982), Tam and Tummala (2001), Hoffer (2005), Bei et al. (2006), Lee et al. (2007), Liu and Wang (2009)

Physical (C11), technical (C12), safety standards (C13) Ease of use (C21), function diversity (C22), adaptability (C23), exibility (C24) Operating cost (C31), product price (C32) Technical support (C41), vendor reputation (C42), capacity (C43) Market share (C51), brand reputation (C52)

C3 C4 C5

Cost After sales services Brand reliability

warehouse manager and logistics manager) having importance of 0.4, 0.3, 0.3, respectively. The judgments are aggregated through Fuzzy Delphi, and the criteria weights are determined. In the rst phase of the study, the group of DM gathers information about all available PDA Barcode Scanners in the Turkish market and evaluates these alternatives with respect to ve rst level criteria as well as the FRs dened by the group. The alternatives that cannot meet the basic requirements are eliminated in this phase. Also, the threshold dened by the DMs is taken into account at this stage in order to eliminate the unsuitable alternatives. The second phase is performed with the sub-criteria to apply a more detailed evaluation. In this phase, the alternatives are ranked and the most appropriate tool is selected. Selection phase involves the same group of DMs, which determines the FRs on the sub-criteria and evaluates the alternatives. The alternatives are ranked in an increasing order of the information content. Also, Fuzzy TOPSIS is applied to compare the outcome of FAD with another MCDM technique. ABC Companys mobile logistics tool evaluation process rst starts by the identication of PDA Barcode Scanner alternatives. The PDA alternatives are selected from popular items of Turkish information system and logistics tool providers. Products are determined from different brands and they are all available in Turkey. Common features are sought for the selection of PDAs to evaluate the same product family. Table 3 displays the product name and the vendors website. Once the DMs decide on the PDA Barcode Scanner alternatives, a primary evaluation is made with the rst-level criteria and related FRs for the pre-assessment phase. The linguistic terms employed in evaluating the PDAs need to be translated into fuzzy numbers in order to quantify the judgments. In this study, a 11-level fuzzy scale is used to assess the alternatives and determine the criteria weights, and another 11-level fuzzy scale is used to assess the FRs, as a bare minimum is required to be achieved for FRs. Table 4 and Fig. 4 describe the linguistic terms, their abbreviations and fuzzy membership functions. The DMs pairwise comparisons on main criteria and sub-criteria are gathered in linguistic terms and then fuzzied, and Fuzzy

Table 3 PDA barcode scanner alternatives. Product name A1 A2 A3 A4 A5 A6 A7 A8 A9 Zebex PDL 20-16 Opticon H19 Mobile Compia M3 Pidion BIP-5000 Symbol MC50 SCC SC600 Casio IT600 Bitatek IT7000 CipherLab 9570CE Vendor website www.bilkur.com.tr www.mlshop.biz www.baykod.com www.sempa.com.tr www.hepsibilgisayar.com www.sistembilisim.com www.mobit.com.tr www.teknoshop.com.tr www.mobilish.com

Delphi is employed to aggregate the pairwise comparisons. Criteria weights determined by Fuzzy AHP are displayed in Fig. 5. Table 5 displays Expert 1s judgment on alternatives as well as on the FRs. The scales described in Table 4 are used for the evaluations. The DMs judgments on FRs and alternatives are gathered and translated into fuzzy numbers. Fuzzy Delphi is then applied to aggregate the evaluations. Table 6 demonstrates the aggregated opinions of DMs based on the main (rst level) criteria. The aggregated evaluations undergo FAD methodology in order to calculate the information contents presented in Table 7. The information contents and the main criteria weights are used to compute the weighted information contents displayed in Table 8. As can be seen in Table 8, FAD methodology itself eliminates two alternatives, A1 and A6, given that these alternatives cannot meet the design range for criteria C2. Hence, the bare minimum dened on FR cannot be satised and these alternatives are eliminated. On the other hand, the pre-assessment phase aims to eliminate as many unt alternatives as possible in order to evaluate and select from a limited set of suitable candidates. Therefore, DMs set a threshold on information content. Alternatives that result in weighted information content WICTOT P 1 are eliminated in order to have the most suitable PDAs to analyze in detail and select from. Four alternatives (A3, A2, A7, and A5) are proven to be satisfactory enough to undergo the selection phase. Alternatives A4, A8, and A9 have insufcient information contents and hence, are eliminated. On the other hand, alternatives A1 and A6 are eliminated from solution set by the FAD methodology itself. The selection phase represents a more detailed evaluation with 14 sub-criteria. The same group of DM evaluates the remaining four alternatives with the same 11-level scale, as well as the FRs in respect to each sub-criterion. Table 9 displays Expert 1s opinions on the performance of the alternatives according to the subcriteria as well as his judgment on the design ranges. As in the pre-assessment phase, the DMs judgments are aggregated as seen in Table 10. These aggregated judgments are employed to compute the information contents for each criterion, displayed in Table 11. Then the hierarchical FAD is applied to calculate the weighted information contents in order to obtain the total weighted information contents shown in Table 12. The selection phase does not seek to eliminate the alternatives, unless the methodology itself eliminates them based on the minimum requirements. Therefore, once the total weighted information contents are calculated, the alternatives are ranked in increasing order of information content. The ranking of the four remaining alternatives are shown in Table 13. 5. Comparative validation This section of the paper presents the comparison of the FAD outcome with another MCDM technique, namely Fuzzy TOPSIS, as well as a sensitivity analysis conducted on variations of FRs.

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153 Table 4 Linguistic terms and membership functions for system and design ranges. Term None Very low Low Fairly low More or less low Medium More or less good Fairly good Good Very good Excellent Abbreviations N VL L FL ML M MG FG G VG E Membership function (0.00, 0.00, 1) (0.00, 0.10, 0.20) (0.10, 0.20, 0.30) (0.20, 0.30, 0.40) (0.30, 0.40, 0.50) (0.40, 0.50, 0.60) (0.50, 0.60, 0.70) (0.60, 0.70, 0.80) (0.70, 0.80, 0.90) (0.80, 0.90, 1.00) (0.90, 1.00, 1.00) Term At At At At At At At At At At At least least least least least least least least least least least none very low low fairly low more or less low medium more or less good fairly good good very good excellent Abbreviations LN LVL LL LFL LML LM LMG LFG LG LVG LE

149

Membership function (0.00, 1.00, 1.00) (0.05, 1.00, 1.00) (0.10, 1.00, 1.00) (0.20, 1.00, 1.00) (0.30, 1.00, 1.00) (0.40, 1.00, 1.00) (0.50, 1.00, 1.00) (0.60, 1.00, 1.00) (0.70, 1.00, 1.00) (0.80, 1.00, 1.00) (0.90, 1.00, 1.00)

Fig. 4. Membership functions for system and design range.

Fig. 5. Evaluation criteria weights.

5.1. Comparison of the obtained results with Fuzzy TOPSIS methodology In this study, another MCDM method is applied in order to compare the performance of the presented methodology. TOPSIS was initially proposed by Chen and Hwang (1992) and the basic principle is that the optimal solution should have the shortest distance from the positive ideal solution and the farthest from the negative ideal solution. In order to assure the consistency of the study, TOPSIS is adapted to the fuzzy environment. Fuzzy TOPSIS is chosen as the additional technique due to its simplicity and wide applications (Qureshi, Kumar, & Kumar, 2008; Shih, 2008) as well as due

Table 5 Expert 1 rst phase evaluation on alternatives and design range. FR C1 C2 C3 C4 C5 LL LM LVL LL LL A1 FL VL VG ML VL A2 M FG M G FG A3 ML FG G M MG A4 MG ML MG MG ML A5 FL M FL FG G A6 VL VL G M FL A7 FL M MG G VG A8 FL MG M ML VL A9 M FL VL ML ML

150 Table 6 Aggregation on expert evaluations. C1 FR A1 A2 A3 A4 A5 A6 A7 A8 A9 0.07 0.26 0.40 0.39 0.47 0.20 0.00 0.20 0.20 0.40 1.00 0.36 0.50 0.49 0.57 0.30 0.10 0.30 0.30 0.50 1.00 0.46 0.60 0.59 0.67 0.40 0.20 0.40 0.40 0.60 C2 0.37 0.06 0.54 0.60 0.27 0.40 0.06 0.40 0.47 0.26

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153

C3 1.00 0.16 0.64 0.70 0.37 0.50 0.16 0.50 0.57 0.36 1.00 0.26 0.74 0.80 0.47 0.60 0.26 0.60 0.67 0.46 0.11 0.80 0.34 0.70 0.47 0.23 0.67 0.50 0.40 0.06 1.00 0.90 0.44 0.80 0.57 0.33 0.77 0.60 0.50 0.16 1.00 1.00 0.54 0.90 0.67 0.43 0.87 0.70 0.60 0.26

C4 0.16 0.30 0.67 0.34 0.53 0.63 0.40 0.73 0.36 0.27 1.00 0.40 0.77 0.44 0.63 0.73 0.50 0.83 0.46 0.37 1.00 0.50 0.87 0.54 0.73 0.83 0.60 0.93 0.56 0.47

C5 0.11 0.09 0.60 0.47 0.36 0.70 0.26 0.86 0.03 0.30 1.00 0.19 0.70 0.57 0.46 0.80 0.36 0.96 0.13 0.40 1.00 0.29 0.80 0.67 0.56 0.90 0.46 1.00 0.23 0.50

Table 7 Information contents and total information contents for pre-assessment phase. IC1 A1 A2 A3 A4 A5 A6 A7 A8 A9 0.94 0.50 0.52 0.35 1.23 3.61 1.23 1.23 0.50 IC2 1 0.59 0.38 3.87 1.50 1 1.50 0.94 4.17 IC3 0.02 0.74 0.08 0.39 1.23 0.10 0.33 0.56 3.14 IC4 1.05 0.11 0.87 0.32 0.16 0.64 0.06 0.78 1.22 IC5 2.69 0.18 0.39 0.68 0.08 1.09 0.00 3.90 0.90 ITOT 1 2.12 2.24 5.61 4.19 1 3.12 7.40 9.93

negative-ideal (Sanayei et al., 2010). A requirement set is determined by the DMs for FAD, whereas FPIRP and FNIRP are generic; which explains the variations in the outcomes. The results obtained in Fuzzy TOPSIS represent a generic ranking, where the requirements for all criteria are the same; whereas the ranking of FAD methodology presents the performance of the alternatives according to the dened requirements. The criteria weights are the same for both methodologies; however the requirements dened on criteria change. On the other hand, as seen in Table 14, when the ranking is divided in three groups, it is observed that each group contains the same alternatives with FAD and Fuzzy TOPSIS, which clearly indicates that alternatives in each group embrace the same level of performance. 5.2. Sensitivity analysis Considering the difference in the outcome of the two methodologies, a sensitivity analysis is conducted in order to identify the cause of this difference. Given that FR set on the C2 (functionality) is the eliminating requirement, the sensitivity analysis is conducted on the variations of the FR2. Table 16 displays the outcome resulted from different values of FR2. As the minimum requirement on FR2 decreases, the weighted information contents of A1 and A6 decreases as well. It can be concluded that FAD methodology results in the same outcome as Fuzzy TOPSIS if the requirement on functionality lessens. With the design range on FR2 being (0.25, 1, 1), the FAD ranking is identical to Fuzzy TOPSIS except for the last three alternatives. As the FR set on C2 decreases, it approaches to fuzzy negative ideal reference point (FNIRP), which is (0; 0; 0). Therefore, the similarity between FAD and Fuzzy TOPSIS outcome increases. On the other hand, once the alternatives that are eliminated with the previous FRs are capable to meet the requirements for C2 with the last FR, the ranking of the alternatives changes considerably and approach the Fuzzy TOPSIS ranking. As seen in Table 16, as long as the criteria weights remain the same, FAD methodology results are similar to the outcome of Fuzzy TOPSIS technique as long as the FRs are close to FNIRP. This can be easily explained by the fact that FAD methodology considers the requirements as well as the criteria weights and evaluates the alternatives according to these requirements. On the other hand, Fuzzy TOPSIS evaluates the alternatives according to the ideal reference points, and therefore, the

Table 8 Weighted information contents and total weighted information contents for preassessment phase. WIC1 A1 A2 A3 A4 A5 A6 A7 A8 A9 0.24 0.13 0.13 0.09 0.31 0.90 0.31 0.31 0.13 WIC2 1 0.15 0.10 0.97 0.38 1 0.38 0.23 1.04 WIC3 0.00 0.15 0.02 0.08 0.25 0.02 0.07 0.11 0.63 WIC4 0.18 0.02 0.15 0.05 0.03 0.11 0.01 0.13 0.21 WIC5 0.35 0.02 0.05 0.09 0.01 0.14 0.00 0.51 0.12 WITOT 1 0.46 0.44 1.28 0.97 1 0.76 1.29 2.12 Ranking 2 1 5 4 3 6 7

to its compatibility with group decision (Wu, Lin, Kung, & Lin, 2007). Moreover, Fuzzy TOPSIS is based on geometrical principles, similar to FAD which also operates on a geometrical level. The technique is adapted from Chen (2000). Tables 14 and 15 exhibit the outcome of the experts opinions with the TOPSIS methodology, as well as the outcome of the FAD methodology. The comparison of the outcomes clearly suggests that the two methodologies differ considerably in evaluating the identied alternatives. Although the FAD and Fuzzy TOPSIS are similar in evaluating the alternatives with a reference to ideal point (FRs for FAD, FPIRP and FNIRP for Fuzzy TOPSIS), the description of the ideal reference point differs for the two methodologies. In TOPSIS the chosen alternative should have the shortest distance from the ideal solution and the farthest distance from the

Table 9 Expert 1 rst phase evaluation on alternatives and design range. C11 FR A2 A3 A5 A7 LL FG G G FG C12 LM FL ML MG M C13 LMG G FG FG FG C21 LM M G FG ML C22 LL G FG MG G C23 LFL MG M M MG C24 LMG FG M MG MG C31 LML MG FL ML G C32 LML ML G ML MG C41 LMG MG M M M C42 LML FG MG G MG C43 LML G FG FG FG C51 LM FG MG FG G C52 LML M MG MG M

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153 Table 10 Aggregation on expert evaluations. C11 FR A2 A3 A5 A7 0.09 0.54 0.58 0.61 0.51 C23 FR A2 A3 A5 A7 0.29 0.44 0.37 0.37 0.38 C42 FR A2 A3 A5 A7 0.27 0.57 0.50 0.61 0.50 1.00 0.67 0.60 0.71 0.60 1.00 0.77 0.70 0.81 0.70 1.00 0.54 0.47 0.47 0.48 1.00 0.64 0.57 0.57 0.58 1.00 0.64 0.68 0.71 0.61 1.00 0.74 0.78 0.81 0.71 C12 0.37 0.32 0.30 0.44 0.28 C24 0.5 0.57 0.37 0.41 0.41 C43 0.39 0.64 0.54 0.57 0.60 1.00 0.74 0.64 0.67 0.70 1.00 0.84 0.74 0.77 0.80 1.00 0.67 0.47 0.51 0.51 1.00 0.77 0.57 0.61 0.61 1.00 0.42 0.40 0.54 0.38 1.00 0.52 0.50 0.64 0.48 C13 0.41 0.70 0.57 0.57 0.54 C31 0.45 0.44 0.32 0.27 0.64 C51 0.49 0.66 0.5 0.63 0.58 1.00 0.76 0.60 0.73 0.68 1.00 0.86 0.70 0.83 0.78 1.00 0.54 0.42 0.37 0.74 1.00 0.64 0.52 0.47 0.84 1.00 0.80 0.67 0.67 0.64 1.00 0.90 0.77 0.77 0.74 C21 0.49 0.34 0.61 0.57 0.36 C32 0.27 0.33 0.70 0.30 0.50 C52 0.3 0.49 0.53 0.62 0.55 1.00 0.59 0.63 0.72 0.65 1.00 0.69 0.73 0.82 0.75 1.00 0.43 0.80 0.40 0.60 1.00 0.53 0.90 0.50 0.70 1.00 0.44 0.71 0.67 0.46 1.00 0.54 0.81 0.77 0.56 C22 0.19 0.61 0.54 0.41 0.61 C41 0.41 0.47 0.37 0.52 0.46 1.00 0.57 0.47 0.62 0.56 1.00 0.71 0.64 0.51 0.71

151

1.00 0.81 0.74 0.61 0.81

1.00 0.67 0.57 0.72 0.66

Table 11 Information contents for selection phase. IC11 A2 A3 A5 A7 0.25 0.19 0.16 0.29 IC31 A2 A3 A5 A7 1.85 4.73 8.34 0.38 IC12 1.93 2.35 0.43 2.83 IC32 1.40 0.12 1.68 0.53 IC13 0.18 0.56 0.56 0.70 IC41 1.14 2.43 0.80 1.23 IC21 5.61 0.59 0.83 4.64 IC42 0.34 0.53 0.25 0.53 IC22 0.20 0.32 0.67 0.20 IC43 0.30 0.64 0.52 0.41 IC23 0.81 1.21 1.21 1.14 IC51 0.38 1.47 0.50 0.76 IC24 0.25 0.19 0.16 0.29 IC31 1.85 4.73 8.34 0.38

Table 14 Performance indices for pre-assessment phase. Alternatives A3 A2 A7 A5 A4 A8 A9 A1 A6 IC 0.44 0.46 0.76 0.97 1.28 1.29 2.12 1 1 Alternatives A3 A2 A7 A4 A5 A8 A1 A9 A6 PI 0.14 0.14 0.13 0.12 0.12 0.10 0.09 0.09 0.08

Table 12 Weighted information contents for sub-criteria. WIC11 A2 A3 A5 A7 0.05 0.04 0.03 0.06 WIC31 A2 A3 A5 A7 0.76 1.94 3.42 0.16 WIC12 0.39 0.47 0.09 0.57 WIC32 0.57 0.05 0.69 0.22 WIC13 0.04 0.11 0.11 0.14 WIC41 0.63 1.34 0.44 0.68 WIC21 1.40 0.15 0.21 1.16 WIC42 0.19 0.29 0.14 0.29 WIC22 0.05 0.08 0.17 0.05 WIC43 0.16 0.35 0.28 0.23 WIC23 0.20 0.30 0.30 0.29 WIC51 0.19 0.74 0.25 0.38 WIC24 0.05 0.04 0.03 0.06 WIC31 0.76 1.94 3.42 0.16

Table 15 Performance indices for selection phase. Alternatives A2 A7 A3 A5 IC 1.08 1.13 1.44 1.45 Alternatives A7 A2 A5 A3 PI 0.05 0.05 0.05 0.05

Table 16 Sensitivity analysis on FR2. Alternatives FR2 A1 A2 A3 A4 A5 A6 A7 A8 A9 WICTOT (0.37, 1, 1) 1 0.46 0.44 1.28 0.97 1 0.76 1.29 2.12 WICTOT (0.35, 1, 1) 1 0.45 0.43 1.15 0.93 1 0.72 1.27 1.99 WICTOT (0.3, 1, 1) 1 0.43 0.42 0.93 0.86 1 0.65 1.23 1.74 WICTOT (0.25, 1, 1) 3.45 0.41 0.41 0.77 0.81 3.86 0.60 1.21 1.56

alternatives are ranked in reference to their performance against each other. Therefore, FAD methodology is a more appropriate technique as a Fuzzy MCDM method for cases where dening the requirements is necessary. Fuzzy TOPSIS can be employed when merely a comparison of the alternatives is required.

Table 13 Weighted information contents and total weighted information contents for selection phase. WIC1 A2 A3 A5 A7 0.12 0.16 0.06 0.19 WIC2 0.47 0.42 0.38 0.58 WIC3 0.27 0.40 0.82 0.07 WIC4 0.17 0.34 0.15 0.20 WIC5 0.07 0.13 0.05 0.08 WITOT 1.08 1.44 1.45 1.13 Ranking 1 3 4 2

6. Conclusions This study proposes a fuzzy group decision-making framework for the mobile logistics tool evaluation problem, as well as a case study for PDA barcode scanner selection. Applications of FAD in recent studies proved this technique to be an appropriate support tool in decision-making. For this reason, both phases of this study

152

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153 Chen, S., & Hwang, C. (1992). Fuzzy multiple attribute decision making: Methods and applications. Berlin: Springer-Verlag. Chen, J., Yen, D., & Chen, K. (2009). The acceptance and diffusion of the innovative smart phone use: A case study of a delivery service company in logistics. Information & Management, 46, 241248. Cicek, K., & Celik, M. (2009). Selection of porous materials in marine system design: The case of heat exchanger aboard ships. Materials and Design, 30, 42604266. Dagdeviren, M., Yavuz, S., & Klnc, N. (2009). Weapon selection using the AHP and TOPSIS methods under fuzzy environment. Expert Systems with Applications, 36, 81438151. Eraslan, E., Akay, D., & Kurt, M. (2006). Usability ranking of intercity bus passenger seats using fuzzy axiomatic design theory. Lecture Notes in Computer Science, 410, 141148. Goncalves-Coelho, A., & Mourao, A. (2007). Axiomatic design as support for decision-making in a design for manufacturing context: A case study. International Journal of Production Economics, 109, 8189. Harutunian, V., Nordlund, M., Tate, D., & Suh, N. (1996). Decision making and software tools for product development based on axiomatic design theory. In The 1996 CIRP general assembly. Como, Italy. Herrera, F., Herrera-Viedma, E., & Chiclana, F. (2001). Multiperson decision-making based on multiplicative preference relations. European Journal of Operational Research, 129, 372385. Hoffer, R. (2005). Auditing PDA wireless devices in nancial services. EDPACS The EDP audit, control, and security newsletter, 33, 19. Hung, K.-C., Julian, P., Chien, T., & Jin, W.-H. (2010). A decision support system for engineering design based on an enhanced fuzzy MCDM approach. Expert Systems with Applications, 37, 202213. Hwang, C., & Yoon, K. (1981). Multiple attribute decision making: Methods and applications. New York: Springer-Verlag. Isklar, G., & Bykzkan, G. (2007). Using a multi-criteria decision making approach to evaluate mobile phone alternatives. Computer Standards & Interfaces, 29, 265274. Jang, B.-S., Yang, Y.-S., Song, Y.-S., Yeun, Y.-S., & Do, S.-H. (2002). Axiomatic design approach for marine design problems. Marine Structures, 15, 3556. Jeong, N., Yoo, Y., & Heo, T.-Y. (2009). Moderating effect of personal innovativeness on mobile-RFID services: Based on Warshaws purchase intention model. Technological Forecasting and Social Change, 76, 154164. Kahraman, C., & Cebi, S. (2009). A new multi-attribute decision making method: Hierarchical fuzzy axiomatic design. Expert Systems with Applications, 36(3), 48484861. Kahraman, C., & Kulak, O. (2005). Fuzzy multi-attribute decision making using an information axiom based approach. Fuzzy multi-criteria decision making. New York, USA: Springer, pp. 265278. Kans, M. (2008). An approach for determining the requirements of computerised maintenance management systems. Computers in Industry, 3240, 59. Kulak, O. (2005). A decision support system for fuzzy multi-attribute selection of material handling equipments. Expert System with Application, 29(2), 310319. Kulak, O., Durmusoglu, M., & Kahraman, C. (2005). Fuzzy multi-attribute equipment selection based on information axiom. Journal of Materials Processing Technology, 169, 337345. Kulak, O., & Kahraman, C. (2005). Fuzzy multi-attribute selection among transportation companies using axiomatic design and analytic hierarchy process. Information Sciences, 170, 191210. Kwong, C., & Bai, H. (2003). Determining the importance weights for the customer requirements in QFD using a fuzzy AHP with an extent analysis approach. IIE Transactions, 35, 619626. Lee, C.-C., Cheng, H., & Cheng, H.-H. (2007). An empirical study of mobile commerce in insurance industry: Task-technology t and individual differences. Decision Support Systems, 43, 95110. Lee, A. H. I. (2009). A fuzzy supplier selection model with the consideration of benets, opportunities, costs and risks. Expert Systems with Applications, 36, 28792893. Lehmann, D., & Oshaughnessy, J. (1982). Decision criteria used in buying different categories of products. Journal of Purchasing and Materials Management, 18, 914. Lin, M., Wang, C., Chen, M., & Chang, C. (2008). Using AHP and TOPSIS approaches in customer-driven product design process. Computers in Industry, 59, 1731. Lin, C.-J., & Wu, W.-W. (2008). A causal analytical method for group decisionmaking under fuzzy environment. Expert Systems with Applications, 34, 205213. Liu, H.-T., & Wang, W.-K. (2009). An integrated fuzzy approach for provider evaluation and selection in third-party logistics. Expert Systems with Applications, 36, 43874398. Madria, S. K., Mohania, M., Bhowmick, S., & Bhargava, B. (2002). Mobile data and transaction management. Information Sciences, 141, 279309. Martinez-Sala, A., Egea-Lopez, E., Garcia-Sanchez, F., & Garcia-Haro, J. (2009). Tracking of returnable packaging and transport units with active RFID in the grocery supply chain. Computers in Industry, 60, 161171. Masud, A., & Ravindran, A. (2007). Multiple criteria decision making. Operations research and management science handbook. Boca Raton, FL: CRC Press. Musilek, P., Guanlao, R., & Barreiro, G. (2005). Genetic programming of fuzzy aggregation operations. Journal of Intelligent & Fuzzy Systems, 16(2), 107118. Ozel, B., & Ozyoruk, B. (2007). Supplier selection with fuzzy axiomatic design. Journal of the Faculty of Engineering and Architecture of Gazi University, 22(3), 415423. Pomerol, J.-C., & Romero, S. (2000). Multicriterion decision in management: Principles and practice. Norwell: Kluwer Academic Publishers.

employed FAD, which can be used to eliminate alternatives that cannot meet basic requirements in the pre-assessment phase, and also to rank the remaining alternatives according to their performance in the selection phase. An industrial case study is used to illustrate the approach. The outcome of FAD application is then compared to the outcome of Fuzzy TOPSIS, given that it is widely used MCDM technique. The main contribution of this paper is that the proposed methodology is a two-phase approach employing FAD and integrating various MCDM techniques. On the other hand, the industrial applications of FAD cover many areas such as shipbuilding industry, transportation and various selection problems; however it does not include IT evaluation, neither mobile/electronic tool selection such as PDAs and barcode scanners. To our knowledge, no previous work investigated such problem. Another contribution of this paper was to establish criteria and methodology for evaluating mobile logistics tools. Implementation of a case study in Turkish logistics industry veried the potential of the proposed methodology. For further research, another industrial case will be executed and an evaluation of RFID systems will be performed for the mentioned company employing the methodology of this paper. Acknowledgements The authors acknowledge the contribution of the industrial experts without which this study could not be accomplished. References
Ayag, Z., & Ozdemir, R. (2007). A combined fuzzy AHP-goal programming approach to assembly-line selection. Journal of Intelligent & Fuzzy Systems, 18(4), 345362. Bei, W., Wang, S., & Hu, J. (2006). An analysis of supplier selection in manufacturing supply chain management. Service Systems and Service Management, 2, 439444. Buyukozkan, G., & Ruan, D. (2008). Evaluation of software development projects using a fuzzy multi-criteria decision approach. Mathematics and Computers in Simulation, 77(56), 464475. Cebi, S., & Celik, M. (2007). Fuzzy axiomatic design approach for measuring of customer satisfaction level on Turkish container ports. In Proceeding of the international 11th IFAC symposium on computational economics and nancial and industrial systems (pp. 221226). Istanbul, Turkey. Cebi, M., & Celik, M. (2008). Ship machinery installation based on fuzzy information axiom: The case of compressed air system. In Proceedings of the 8th international FLINS conference on computational intelligence in decision and control. Madrid, Spain. Cebi, M., Celik, M., Er, I., & Kahraman, C. (2008). Structuring ship design project approval mechanism towards operator-system interfaces via fuzzy axiomatic design principles. In Proceedings of the 8th international FLINS conference on computational intelligence in decision and control (pp. 11111116). Madrid, Spain. Cebi, S., & Kahraman, C. (2010a). Extension of axiomatic design principles under fuzzy environment. Expert Systems with Applications, 37(3), 26822689. Cebi, S., & Kahraman, C. (2010b). Indicator design for passenger car using fuzzy axiomatic design principles. Expert Systems with Applications, 37(9), 64706481. Celik, M. (2009). A hybrid design methodology for structuring an integrated environmental management system (IE MS) for shipping business. Journal of Environmental Management, 90(3), 14691475. Celik, M., Cebi, S., Kahraman, C., & Er, I. (2009). Application of axiomatic design and TOPSIS methodologies under fuzzy environment for proposing competitive strategies on Turkish container ports in maritime transportation network. Expert Systems with Applications, 36(3), 45414557. Celik, M., & Er, I. (2009). Fuzzy axiomatic design extension for managing model selection paradigm in decision science. Expert Systems with Applications, 36(3), 64776484. Celik, M., Kahraman, C., Cebi, S., & Er, I. (2009). Fuzzy axiomatic design-based performance evaluation model for docking facilities in shipbuilding industry: The case of Turkish shipyards. Expert Systems with Applications, 36(1), 599615. Cevikcan, E., Cebi, S., & Kaya, I. (2009). Fuzzy VIKOR and fuzzy axiomatic design versus to fuzzy TOPSIS: An application of candidate assessment. Journal of Multiple-Valued Logic and Soft Computing, 15, 181208. Chamodrakas, I., Batis, D., & Martakos, D. (2010). Supplier selection in electronic marketplaces using satisfying and fuzzy AHP. Expert Systems with Applications, 37, 490498. Chen, C.-T. (2000). Extensions of the TOPSIS for group decision-making under fuzzy environment. Fuzzy Sets and Systems, 114, 19. Chen, L.-S., & Cheng, C.-H. (2005). Selecting IS personnel use fuzzy GDSS based on metric distance method. European Journal of Operational Research, 160, 803820.

G. Bykzkan et al. / Expert Systems with Applications 39 (2012) 142153 Qi, J., Li, L., Li, Y., & Shu, H. (2009). An extension of technology acceptance model: Analysis of the adoption of mobile data services in China. Systems Research and Behavioral Science, 26, 391507. Qureshi, M., Kumar, P., & Kumar, D. (2008). 3PL evaluation and selection under a fuzzy environment: A case study. The ICFAI Journal of Supply Chain Management, 5, 3853. Royes, G., Bastos, R., & Royes, G. (2003). Applicants selection applying a fuzzy multicriteria CBR methodology. Journal of Intelligent & Fuzzy Systems, 14(4), 167180. Saaty, T. (1980). The analytic hierarchy process. New York: McGraw Hill. Sanayei, A., Mousavi, S., & Yazdankhah, A. (2010). Group decision making process for supplier selection with VIKOR under fuzzy environment. Expert Systems with Applications, 37, 2430. Shih, H.-S. (2008). Incremental analysis for MCDM with an application to group TOPSIS. European Journal of Operational Research, 186, 720734. Suh, N. (1990). The principles of design. New York: Oxford University Press. Suh, N. (2001). Axiomatic design advances applications. New York: Oxford University Press. Tam, M., & Tummala, V. (2001). An application of the AHP in vendor selection of a telecommunications system. Omega, 29, 171182. Wamba, S., Lefebvre, L., Bendavid, Y., & Lefebvre, E. (2008). Exploring the impact of RFID technology and the EPC network on mobile B2B e-commerce: A case study

153

in the retail industry. International Journal of Production Economics, 112, 614629. Wang, Y.-M., & Parkan, C. (2008). Optimal aggregation of fuzzy preference relations with an application to broadband internet service selection. European Journal of Operational Research, 187, 14761486. Wu, W.-Y., Lin, C., Kung, J.-Y., & Lin, C.-T. (2007). A new fuzzy TOPSIS for fuzzy MADM problems under group decisions. Journal of Intelligent & Fuzzy Systems, 18(2), 109115. Wu, J.-H., & Wang, S.-C. (2005). What drives mobile commerce? An empirical evaluation of the revised technology acceptance model. Information & Management, 42, 719729. Yeh, C.-H., & Chang, Y.-H. (2009). Modeling subjective evaluation for fuzzy group multicriteria decision making. European Journal of Operational Research, 194(2), 464473. Ycel, G., & Aktas, E. (2008). An evaluation methodology for ergonomic design of electronic consumer products based on fuzzy axiomatic design. Journal of Multiple-Valued Logic and Soft Computing, 14(3-5), 475494. Zackariasson, P., & Wilson, T. (2002). Technology, after sales service and organizations. In Association of marketing theory and practice (AMTP). Georgia, USA. Zadeh, L. (1975). The concept of a linguistic variable and its applications to approximate reasoning. Information Sciences. 199249 (I), 301357 (II).