A KANSEI Retrieval Method Using Factor Analysis and An Application to Comfortable Cities in China

Abstract

In recent years, as the amount of information in society increases, search methods are diversified and rapidly developing. In such situation, various methods have been proposed for Kansei search from this century to the present. Among these methods many search objects were data including Kansei information. In this research, we proposed a Kansei search method using factor analysis with data (sample data) that does not contain any Kansei information, and showed an example. First, a name (Kansei factor name) including was given to a factor obtained from factor analysis results of sample data. Next, a Kansei search query (search query word) was created by combining several modifiers and the Kansei factor names that representing the level of sensitivity. Then, based on the Kansei search query we made query data. Using this data with result from factor analyze as factor loading matrix and inverse correlation matrix, we calculated the factor score of query data. Finally, by calculating the distance of factor score between query data and sample data, we made those sample data near query data as search result. The proposed method was applied to sample data of 240 cities in China consisting of 24 variables, and performed the verification.