The Evolution of a Practical Agent-based Recommender System

Download

Postscript Download (2547Kb)

Description/Abstract

This paper presents a multi-agent framework that has been developed to assist Web users within organisations in finding relevant documents and identifying colleagues conducting related work. The system exploits the idea that information acquired and created by one user navigating the information space is something that can be utilised to assist other users. The design of the system builds on the practical experiences of an earlier agent-based recommender system that provided navigation assistance based on information about which documents other users had visited. The system presented here has an improved interface and agent architecture, and it significantly extends the functionality. The interface successfully employs bookmarks as a familiar and effective interaction mechanism. Added services within the architecture include handling of multimedia documents and integration with other open hypermedia systems, and the system is also capable of augmenting document content on delivery. The framework is open and extensible to incorporate new functionality.