Towards cooperative information retrieval system with user modelling
Résumé
Finding information may be compared to problem solving. A user of an Information Retrieval System (IRS) has a need that constitutes an objective for him. Unfortunately, this objective may not be well defined when the user starts the IRS. Whatever the degree of understanding of this objective, the user translates it into a set of queries that the system can use for calculating the relevant solutions. Knowing the problem to solve does not necessarily mean knowing how to solve it. Observing the dialog between a user and a librarian which often produce better relevant solutions and faster than asking the user to use an IRS himself we propose an IRS that may be used to provide cooperative information retrieval. The architecture provides three functioning modes: autonomy, observation and cooperative. In cooperative mode, two users participate in the process of information retrieval, in real time. The two of them work on distant machines. In observation mode, user sees what another user is doing but without the possibility of intervening in that process. In autonomous mode, a user uses the system alone without anybody seeing him and he does not see anybody. Our proposals are implemented in a prototype (METIORE) used for accessing and analyzing bibliographic references of our research laboratory center