between the target achievements with respect to different Kansei attributes. In this case, the attitudinal character Ω will not affect the aggregation value for each priority level, thus the aggregation results depend only upon priority hierarchy of the seven Kansei attributes. For purposes of simplicity, we assume that the prioritization has been made in order of the index of Kansei attributes, denoted as
X4 X10 X11 X15X21 X25X26.
Then the ranking list of the top 5 Kanazawa gold leaf that best meet consumer’s preferences is:
A12A3 A11 A14A10.
8.5.3 Discussions
In this study, a preparative study is conducted first to gather the Kansei assessments for the thirty products of Kanazawa gold leaf. Although it is difficult and time-consuming to gather the Kansei assessments, the preparative study is a common step in Kansei engineering. In Kansei engineering, Kansei evaluation is an essential step. As Kansei engineering can be used by designers as a design aid to develop products that are able to match consumers’ Kansei. In this sense, this case study provides a possible solution to consumer-oriented design in Kansei engineering research for the Kanzawa gold leaf.
On the other hand, as Kansei engineering can be used by consumers to select products based on their Kansei requirements, this case study also provides a possible solution for the recommender systems. Japanese traditional crafts usually have the following properties:
low purchasing frequency, high price, high involvement in selection of preferred crafts.
As the aesthetic aspect (brand image, pattern, personal aesthetics, current trends of fashion etc.) plays a crucial role in consumers’ perceptions of traditional crafts, Kansei information is essential and necessary for the consumers. Thus a preparative study is quite necessary in choice of traditional crafts. In fact, in recommender systems, product rating techniques have been well developed. We can use the rating techniques in recommender systems to fulfil the preparative study step. In addition, it is not so easy to find consumers’
preference information on different Kansei attributes of the traditional crafts, consumers’
feeling target provides a good solution to find the preferred crafts.
Chapter 9
Contributions and Future Work
9.1 Main Contributions of the Thesis
In this thesis, we have presented a study of multi-attribute target-oriented decision analysis and its applications to Kansei evaluation problems. Among the nine chapters of the thesis, the main chapters are Chapter 3, Chapter 4, Chapter 5, Chapter 6, Chapter 7, and Chapter 8. The main contributions of this thesis are as follows.
1. The first main contribution is that we propose two methods to target-oriented decision model with different target preferences and extend those two methods to fuzzy target-oriented decision analysis.
(a) The first sub-contribution in this part is that we develop two methods for target-oriented decision analysis with different target preferences.
In most studies on target-oriented decision making, monotonic assumptions are given in advance to simplify the problems, e.g., the attribute/criteria wealth.
However, there are three types of target preferences. Two methods have been proposed to model the different target preference types: cumulative distribu-tion funcdistribu-tion (cdf) based method and level set based method. No matter which method is selected, both of these two methods can induce four shaped value functions: S-shaped, inverse S-shaped, convex, and concave, which represents decision maker’s psychological preference. The main difference between these two methods is that the level set based model induces a stricter value function than the cdf based model.
(b) The second sub-contribution in this part is that we extend those two random target-oriented decision analysis to fuzzy uncertain targets.
Target-oriented decision model assumes that target has a random probability distribution. Fuzzy uncertainty is considered by decision makers to linguisti-cally specify their uncertain targets. In this thesis, two methods of fuzzy target-oriented decision analysis with respect to different target preferences have been proposed. Firstly, a thorough analysis of possibility-probability transforma-tions is given, and then the proportional approach is properly used to trans-form a possibility distribution into the probability distribution. Secondly, two methods of fuzzy target-oriented decision analysis have been obtained from the random target-oriented decision model. Finally, some widely used fuzzy
targets used in the pioneering work by Bellman and Zadeh [12] are selected to illustrate the fuzzy target-oriented model. Our research outperforms better in terms of three aspects.
2. The second main contribution is that we develop a non-additive multi-attribute target-oriented decision model based on fuzzy measure and fuzzy integral, and put forward a prioritized aggregation operator to model the prioritization between targets/attributes.
(a) The first sub-contribution in this part is that we model the interdependence be-tween different targets based on λ-fuzzy measure and Choquet fuzzy integral.
The use of fuzzy measures and fuzzy integral in MADA enables us to model some interaction phenomena existing among different attributes. As we shall see, multi-attribute target-oriented function has a similar structure with fuzzy measure, and fuzzy integral does not assume the independence. The fact that fuzzy integral model does not need to assume the independence of each target, means it can be used in non-linear situations. Thus we use fuzzy measure and fuzzy integral to model the interaction among targets. Furthermore, even if, in an objective sense any two targets are independent, they are not necessar-ily considered to be independent from the DM’s subjective viewpoint. This explains why a fuzzy integral is more appropriate. Since the specification for fuzzy measures requires the values of a fuzzy measure for all subsets, theλ-fuzzy measure is used in order to reduce the difficulty of collecting information and the Choquet fuzzy integral is used to model the dependence in multi-attribute target-oriented decision analysis. A bisection search algorithm is also designed to identify the fuzzy measures of individual attributes group with a given λ value.
(b) The second sub-contribution in this part is that we put forward a prioritized OWA aggregation operator to model the prioritization between different targets.
Firstly the OWA operator is used to obtain the satisfaction degree for each pri-ority level. To preserve the tradeoffs among the attributes in the same pripri-ority level, the degree of satisfaction for each priority level is viewed as a pseudo attribute. Secondly, we suggest that roughly speaking any t-norm can be used to model the priority relationships between the attributes in different prior-ity levels. To keep the slight change of priorprior-ity weight, strict Archimedean t-norms perform better in inducing priority weight. As Hamacher family of t-norms provide a wide class of strict Archimedean t-norms ranging from the product to weakest t-norm, Hamacher t-norms are selected to induce the pri-ority weight for each pripri-ority level. Thirdly, considering DM’s requirement toward the higher priority levels, a benchmark based approach is proposed to induce priority weight for each priority level, i.e., “the satisfactions of the higher priority attributes are larger than or equal to the DM’s requirements”. We sug-gest that the weights of lower priority level should depend on the benchmark achievement of all the higher priority levels.
To illustrate the effectiveness and advantages of the prioritized OWA operator
mentioned above, we conduct several comparative analysis with previous work on prioritized aggregation.
3. The third contribution is that we develop a Kansei evaluation model based on prioritized multi-attribute fuzzy target-oriented decision anal-ysis. A case study is also conducted to illustrate the proposed Kansei evaluation model.
To overcome the those two above-mentioned problems in current research on Kansei evaluation, we put forward a Kansei evaluation model based on fuzzy target-oriented decision analysis and prioritized OWA aggregation operator. Firstly, like the tradi-tional Kansei evaluation method, a preparatory experiment study is conducted in advance to select Kansei attributes by means of semantic differential (SD) method.
In order to obtain Kansei data of products, a number of people are selected to as-sess products regarding these Kansei attributes. Secondly, these Kansei data are used to generate Kansei profiles for evaluated products by making use of the voting statistics. Thirdly, according to consumer-specified preferences on Kansei attributes, three main types of fuzzy targets are defined, to represent the consumers’ prefer-ences. Based on the principle of target-oriented decision analysis, we can obtain the satisfaction degrees (probabilities of meeting targets) regarding the Kansei at-tributes selected by consumers for all the evaluated products. Finally, considering prioritization of the Kansei attributes, the prioritized OWA aggregation operator is used to aggregate the partial satisfaction degrees for the evaluated products.
Kansei evaluation has been applied to consumer products with successful results, e.g., table glasses, housing assessment, telephones, cars, and mobile phones. How-ever, Kansei evaluation of traditional crafts has not been addressed yet In Japan, there are many traditional crafts such as fittings, textile, etc. These beautiful, el-egant and delicate products are closely related to and have played an important role in Japanese culture and life. Evaluations of these traditional crafts would be of great help for marketing or recommendation purposes. Thus the Japanese tradi-tional crafts are used as a case study to illustrate the proposed Kansei evaluation model. By using our model, consumers can choose their preferred crafts according to their preferences.