رویکرد مبتنی بر اولویت جدید روابط ناقص برای گسترش کارکرد کیفیت

Quality function deployment (QFD) is a widely-used methodology for developing a design quality aimed at satisfying the customer and translating the customer’s demand into design targets. As QFD methodology is based on the voices of customers, it requires a group of individuals, or decision makers (DMs), to express their preferences. However, there can be a bottleneck since DMs may not have a perfect knowledge of the problem to be solved. Thereby, the aim of this study is to improve the effectiveness of the evaluation process in QFD. The QFD methodology is extended by introducing a new group decision making (GDM) approach that takes into account incomplete information of DMs by means of the fuzzy set theory. The proposed approach is applied to a collaborative software development process to demonstrate its potential. Keywords

Quality function deployment (QFD) is an effective tool that can aid in moving towards a more proactive product development. It is a customer-driven quality management system which incorporates the “voice of the customer” into appropriate company requirements at each stage of product development, from product planning and process design to manufacturing and delivery, to create higher customer satisfaction for the product [31], [34] and [57]. As a proactive product development or planning methodology, successful applications of QFD have produced many benefits, including lower start-up costs, shorter design cycles, promotion of teamwork, and provision of documentation [7], [13] and [21]. The QFD methodology can be used for both tangible products and non-tangible services, including manufactured goods, service industry, software products, IT projects, business process development, government, healthcare, environmental initiatives, and many other applications [10], [20], [22], [53] and [72]. QFD is comprised of a major group decision making (GDM) processes. In practice, determining the weights of customer requirements (CRs) is a GDM process. This is mainly because of the risk of relying on a single decision maker (DM) with his/her limitations on experiences, preferences or biases about the issues involved, and the fact that individuals are often unable to clearly identify their own states. Multiple DMs, thus GDM, are often preferred to a single DM to avoid the bias and minimize the partiality in the decision process [15], [25], [40] and [75]. Generally, different and even subjective opinions are quite often in a GDM process due to the limitations of experience and imprecision. Obviously, the importance of each CR in QFD is determined, often with ambiguity, by a group of people. In addition, due to constraints as time pressure, lack of expertise in related issue, etc. DMs may develop incomplete preferences where some elements cannot be provided. Under such circumstances, fuzzy set theory [73] and incomplete preference relations [2], [37], [38], [61], [64], [65] and [66] can be applied to deal with group decisions when the information is imprecise and missing. This paper develops a new fuzzy logic based GDM approach to QFD with incomplete preference relations. Moreover, a collaborative software development (CSD) example is provided to show that the proposed GDM approach can be effectively used in QFD. As the study of incomplete preferences in fuzzy environments is not widespread, there is no study in the literature that applies it in CSD field. In our knowledge, there is only one research realized by Han et al. [36] that combines fuzzy and incomplete preference relations to construct extended QFD model. While the mentioned paper also considers the incomplete information case for QFD, the way it treats the problem is very different from our approach. In their work, Han et al. [36] have assumed that there can be missing CRs weights and/or the quantified relationship between CRs and design requirements (DRs). They also postulated that any method relying on pairwise comparisons can be used only when there is complete information. In our study, we show that this later assumption can be relaxed, in other words, incomplete information in pairwise comparisons could be handled. Therefore, all CRs weights can be obtained rigorously. Another criticism with this mentioned work is that within a GDM setting, it is not possible in practice to arrive at a case where there is no information about the weight of a CR. Even when multiple customers could not evaluate the importance of a CR, there would be others who provide this information. This case is not considered in Han et al. [36]. In our approach however, every customer evaluation is investigated separately and a group consensus is reached by the method. The paper is then organized as follows: In Section 2, the integrated concept of QFD is briefly presented. Section 3 details the computational procedure. A CSD example is then given in Section 4. Finally, Section 5 contains our concluding remarks

This study provides a different point of view to the evaluation the QFD applications. To our knowledge, no previous work has investigated this subject using this kind of integrated method. We studied the problem of CSD with incomplete preferences, and this approach can be applied for different kinds of product development problems. Determining the relative importance of CRs is a fundamental problem in QFD applications. Successful applications of QFD basically rely on effective communication among team members to reach a consensus and assigning importance levels that reflect each individual member’s preferences. Thus, in this study one of the primary aims was to apply GDM in QFD. Generally it is difficult for a DM to provide preferences for all pairs of factors because of time pressure, lack of knowledge or data, and his/her limited expertise. Therefore, another aim was to show the use of incomplete preference relations in GDM applications. The prominent characteristic of the proposed method is that it needs the least judgments provided by the DM to construct a consistent complete linguistic preference relation. The approach combines all individual preferences into the group preferences and aggregates the overall information to get the ranking of factors. No previous paper in the QFD literature has attempted to aggregate opinions of team members in the case where each individual has incomplete evaluation. To extend the proposed method, future work can involve the use of incomplete linguistic preferences [4] to estimate the missing values.