A system architecture for multimedia information retrieval

Abstract

This paper describes the architecture and the main features of DOMINO, a multimedia information retrieval system whose database is a collection of multimedia documents (MDs) con stituted of a mixing of texts and images. DOMINO data structure has been designed in such a way that both MD integrity and query symmetry with respect to the type of information involved (textual or pictorial) are preserved. In this paper we will focus on the general aspects of DOMINO architecture bringing out a thorough description of the textual data structure, based on an inverted file structure, and on its generation, inquiry and updating procedures. Spe cial effort has been made to design a data structure for which storage requirements and retrieval time are optimized depend ing on the type of search strategies the system should pursue.