MM: Modular architecture for multimedia information retrieval

In this paper we present an overview of a modular architecture that provides a set of tools used to analyze, index and retrieve multimedia documents. It addresses the users communities interested into multimedia document indexing and retrieval, filtering and mining, especially when the interaction between heterogeneous data are of interest (multi-modal and cross-modal indexing). The objective of the full architecture is to index and retrieve heterogeneous multimedia documents containing texts, audio and still or moving images based on user query. This query may be a simple text, a more complex and structured query or an example of multimedia document. The modular aspect of our system allows to configure it according to the needs. Main features of the system are reviewed then a walk-through of the demo is presented.