Shunji Ota

The semantic network of KANSEI words

We have developed a KANSEI (feeling or sensitivity in Japanese) dictionary system applicable to the semantic network by making use of the similarity of each word. This dictionary system can construct the numerical value of KANSEI words each of which is quite ambiguous. We introduce KANSEI words into our current information systems that manage subjective information. This dictionary handles more than 800 KANSEI words, the contents of which suffice for general purposes.

Image Retrieval System Using KANSEI Features

We have developed an Image Retrieval System using KANSEI(feeling or sensitivity) features. This system can search the same sensuous image from a large image storage not using text or word but an image. Therefore it doesn’t need indexing on each image for preparing image retrieval. Out system extracts the KANSEI features from each image, and sets adequate weights for combining those features. In order to decide the weights, we introduce a new value called “adaptability”. It judges how much the features are extracted from the image. As a result, adaptability makes it possible to construct a KANSEI model according to each image and to calculate similarity between images.