An Agent System for Electronic Mail Filtering

O. Nouali (Algeria), B. Toursel (France), R. Azzouz, and C. Benyahia (Algeria)


Information filtering, electronic mail, E-mail, Baysian Inference Network, genetic algorithm, Filters, mail filtering, Internet, Distributed Artificial Intelligence, Agent, Multi-agents.


With advances in communication technology, it's easy to create and disseminate the information to the users. The communication via electronic mail is one of the most popular services of the internet. The amount of incoming messages to users will become increasingly important. Users are facing increasing difficulties in processing huge amount of messages. Obviously, there is a strong demand for building automatic tools that filter messages according to user's interests. SYMCOL is an intelligent and dynamic E-mail filtering tool. It can be used by professional and non professional users. With SYMCOL, it is easier to use and manage messages. It aids users to filtering their incoming messages and in the selection of what to read. A user can describe his interests by providing either simply a set of keywords or a set of messages which are supposed interesting for him. SYMCOL modular structure makes it portable and easy to adapt to new filtering applications. This paper presents SYMCOL main feature: A multi agent to drive the filtering process, Baysian Inference Network to model the user's profile, and the genetic algorithms for learning.

Important Links:

Go Back