Automating indexing of classes and conferences

The 'Cours Chemin' project aims to develop tools for the electronic capture and processing of lectures and conferences. More specifically, the main goal of this project is to produce methodologies and tools which should eventually minimize the acquisition cost, whilst facilitating content-based access through the Internet, with the help of XML-based multimedia documents technology on the one hand, and of natural language processing tools on the other hand. This paper describes the architecture of our system and presents sample data from on an on-going case-study in the field of legal conferences. After a short presentation of the specifics of the application domain, we will discuss, for each component of our system, the technological choices that have been made, their current limits and the possible ways for circumventing such limits. The paper also includes a preliminary quantitative evaluation of its various modules.