Basic Study of Psychological Analysis about "KANSEI Word"

形象技术或感性工程被定义为一个翻译系统的消费者的形象或感觉到真实的设计组件。在本研究中，我们提出了一些方法来分析消费者的形象或感觉的对象或产品方面的“感性词”。我们发展了收集。通过互联网对系统的数据进行了验证，并对所提方法的效率进行了评价.我们提出的分析方法及其性质如下。(1)我们提出了投影寻踪回归使用成对比较数据的方式类似于Guttman的量化方法的成对比较。我们收集了有关感觉和情感的数据，并使用所提出的方法分析了对象的设计组件之间的关系，相对于形容词“引人注目”和“安静”。(2)利用马氏距离离散化分析的有效性，提出了一种基于伪马氏距离和最大熵的模糊聚类算法 ...更多信息 它被广泛应用于各种模式识别问题。(3)提出了一种同时应用模糊c-均值聚类算法和对应分析的消费者心理感受分析方法。在我们的聚类算法中，成员的集群确定，不仅考虑到距离聚类中心的距离最小化，但也分配给每个类别和个人的数值之间的相关性比最大化。(4)强非线性多变量函数很难带来有关输入和输出之间局部关系的见解。对于心理感受的分析，应用模糊c-回归模型的弱非线性版本。每个模糊聚类中的数据被投影到一维空间中，可以发现独立观测值与其相应的相关观测值之间潜在的局部弱非线性关系。(5)作为一个实证实验，我们处理的“最新”的感觉CI（企业形象）符号标志，它具有类似的模式，日本国旗的设计组成部分是在白色浮雕的圆圈的位置，并分析了使用所提出的方法设计的印象。少

Image Technology or Kansei Engineering is defined as a translation system of a consumer's image or feeling into real design components. In this research we proposed some methods to analyze consumer's image or feeling for objects or products with respect to "KANSEI word". We developed the collecting. System of data through the Internet and evaluated the efficiency of the proposed methods. Our proposed analyzing methods and their properties are as follows.(1) We proposed projection pursuit regression using pairwise comparison data in the similar way to Guttman's Quantifying method of pairwise comparisons. We collected the data about feeling and emotion, and analyzed the relation between design components of objects with respect to the adjective words "conspicuous" and "quiet" using proposed method.(2) We proposed the convenient fuzzy clustering algorithm using pseudo Mahalanobis distances and maximizing entropy approach, since discreminant analysis with Mahalanobis distances has been eff … More iciently applied to various pattern recognition problems.(3) We proposed an approach to analyze the psychological feeling of consumer in which the Fuzzy c-Means clustering algorithm and correspondence analysis are simultaneously applied. In our clustering algorithm, membership to clusters are determined by considering not only the minimization of distances from cluster centers but also the maximization of correlation ratio between numerical values which are assigned to each category and individual.(4) Strongly nonlinear multi-variate functions hardly bring about insights about the local relationships between inputs and outputs. For the analysis of psychological feelings a weakly nonlinear version of Fuzzy c-Regression Models is applied. The data in each fuzzy cluster is projected in a single dimensional space and latent local weakly nonlinear relationships between independent observations and their corresponding dependent observations can be found.(5) As an empirical experiment, we dealt with "up-to-date" feeling of CI (Corporate Identity) symbol marks, which has similar pattern to Japanese national flag whose design components are the locations of circles relieved in white, and analyzed the impression for the design using the proposed method. Less