A Hybrid Architecture for Filtering Information on the Web

The system described in this paper has been developed with the aim of increasing the efficiency of Web Information Filtering processes and is based on a Case-Based approach. It currently acts as an intelligent interface for Web search engines. A primary characteristic of the system is its ability to adapt its filtering processes to a particular user based on the specific interests of the latter. This ability is achieved by means of a hybrid architecture, i.e., a sub-symbolic module integrated into a Case-Based reasoner. This dual component system serves to create and maintain individual user models that are stored in a library of cases for long-term reference of the system itself. The first experimental results of the system are encouraging.