Knowledge-based multimedia information retrieval applied to plant diagrams

A discussion is given on a new method of multimedia information retrieval, which is applied to plant information systems. The method is based on a retrieval model which formulates each retrieval action as a transition of viewpoint in information space, where the space consists of knowledge space and data space. Following good features are derived; various types of media can be treated in an integrated way by means of knowledge space representation, and easy operations are enabled by means of visual communication between users and the system. The method is especially applicable for plant information systems because there are various types of information such as plant diagrams, sensed data and experts' knowledge, and because the man-machine interface is important to enable quick response by operators. Methods of retrieval based on similarity and on relationships which are useful as decision support facilities are discussed.<>