Imagery Informatics and KANSEI information Processing

摘要

The computational formalization of the imagery informatics is probably the next focus of artificial intelligence. Computer agent has to generate a machine image that is similar to user's one through the communication to user. In such case, imagery information needs to have a structure. In this work, visual and audio image can be classified into three types. Agent might use physiological and psychological data also. These structured knowledge forms a network with many sub networks. KANSEI information processing deals with this networked knowledge and data obtained. The basic system is developed to process KANSEI information and applied to shape selection problem, cooking selection problem and communication with a virtual creatures. These examples were rather independent each other. However logical and KANSEI subsystems are now trying to be integrated to the whole system.