Dialogue Techniques for Supporting Multimedia Information Retrieval

Advanced information retrieval and database management systems employ a variety of elaborate data representation, indexing and retrieval techniques, but usually apply less elaborate – often simplistic – models of the human-computer interaction. Effective interaction, however, is essential for effective information retrieval. To be able to assist and guide the user actively through the entire information-seeking process a cooperative system must rely on an elaborate model of the interaction, or, dialogue. In this paper we describe a prototype of a multimedia information retrieval system, which monitors and guides the user-system interaction based on an explicit representation of the dialogue. The retrieval engine employs a novel approach to content-oriented retrieval based on abductive reasoning. It will be shown by examples, how the retrieval engine and the dialogue manager interact with each other to generate situation-dependent interaction possibilities and system responses to create a coherent interaction structure as the dialogue evolves.