Ontology-based Indexing and Retrieval of Multimedia Documents on the WWW

This research addresses the issue of designing a precision-oriented information retrieval system to access the content of multimedia documents via natural language interaction on the WWW. The produced system will route user’s queries towards a dedicated information retrieval system (IRS) accessible online. It will use natural language interaction with the user in order to pre-fill a form used to send a query to the IRS. It will then process the results of this query before presenting them through a WWW interface. The objective of the research is to provide the system with capacities to adapt itself to the user, for a more effective and efficient interaction. In other words, the IRS is meant to be able to automatically and progressively tune its output to satisfy the user’s information needs. This technology is dedicated to personalized access to the multimedia information. The work will be directed towards adapting an information retrieval scenario to a specific user in order to provide him with information relevant to his needs, preferences and usage context. The direction of our research combines a set of technologies and knowledge related to the user context, with an aim of delivering the results which are the most relevant with respect to initial information needs. Our work aims at the integration of the document and query semantic contents as well as the contextual information in the information retrieval (IR) process. Accordingly, we summarize the goals of the PhD research: The integration of heterogeneous multimedia information sources stored in a distributed environment within a unique information retrieval scenario. The specification and design of a precision-oriented information retrieval system considering the semantics and context of both queries and indexed multimedia documents.