產品形態特徵與構成關係影響消費者感性評價之研究——以水壺的設計為例

Exploring the Relationship between the Product Form Features and Feature Composition and User's Kansei Evaluation

感性工學的發展，使設計者能藉由定量的分析，了解產品形態特徵與消費者感性的相關性，以有效地掌握消費者偏好與感受。然而在相關感性研究中，造形分析模式的建構大多以產品元件的形態特徵為導向。根據認知心理學中相關形狀辨識的理論發現，單純依賴物件特徵的辨識，並不能完整地解釋人類的認知行為；構成關係－－元件與元件間的相對關係，也扮演重要的角色。換言之，以產品特徵為導向的造形分析模式，並不能完整地將消費者的感性認知予以轉換。因此，本研究以感性工學之人因技術為基礎，結合人類圖樣識別之認知特質，提出一整合形態特徵與構成關係之造形分析模式，並以水壺的設計為探討對象。研究中利用線性複回歸分析進行感性工學模式的建構，並針對特徵導向模式與整合模式進行績效的比較。研究結果指出，藉由分析結果的比較與測試樣本的驗證，整合模式能明顯助益於消費者感性評價行為的預測。同時，本研究亦導入非線性的類神經網路，進行感性工學模式的建構，並針對線性與非線性分析模式之結果加以比較。在此案例中，複回歸分析的線性模式在感性評價的預測準確率，優於非線性的類神經網路模式。

In the viewpoint of cognitive psychology, people perceive and recognize objects by identifying their features as well as feature composition, the compositional relationship among these features. However, most studies of Kansei engineering have only discussed the relation between users' Kansei evaluation and products' form features. In this study, instead, we proposed a modified kansei engineering system, by using teapot designs as examples, to explain how the form element features and feature composition together will affect users' kansei. Firstly, a SD evaluating survey was conducted to some selected teapot designs. Factor analysis was conducted on the SD evaluation data, and three main factors of users' image perception and expression, including stable factor, intense factor and aesthetic factor, were extracted. A multiple linear regression then was applied to analyze the relationship between the users' feeling （SD evaluation） on these designs （dependent variables） and design features of these designs （independent variables）. There were two models of determining design features, a form element feature-oriented model and a modified model based on form element features and feature composition, adopted in this analysis. In the first model, only form element features were considered as independent variables; while in the second one, both form element features and feature composition were considered as independent variables. A series of statistical analyses was conducted to compare the different performance between the two models. The result indicated that the modified kansei system is more valid on explaining users' kansei information, especially in aesthetic factor. Furthermore, neural network was applied to construct a non-linear modified Kansei system based on design features including form element features and feature composition. A series of statistical analyses then was conducted to compare the different performance between the linear model and the non-linear one. As revealed by the result, the linear model is more valid on explaining users' kansei information than the non-linear model.