We propose a multimedia information retrieval model named "Panoramic Information Retrieval Model" and describe implementation of a Web document retrieval system based on the model. The system, named "MIRACLES", extracts images and their related text from web pages and locates these images on a plane based on the similarities between image or text features. Users can retrieve web pages interactively by displacing the images to upper or lower side of the plane. The displacement is controlled by the user's "cross-media" inputs, that is, keyword, key-text, key-image. We confirm the effectiveness of the proposed model through some experiments with the system.
展开▼
