Construction of a Global, Unified, Personalized Recommender System

16.03.2005

Abstract
This thesis presents WIRE, a recommender system for the WAIF project at the University of Tromsø. In this thesis we will describe a working prototype of a recommender system that provides its users with advices regarding web-resources. It is a distributed system that does not rely on centralized solutions to create a recommender system that produces recommendations with good precision and high recall. The design presented is derived to give a more general architecture for a recommender system in a push based environment which WAIF provides. Our solution allows users to connect and disconnect arbitrary. By utilizing the architecture, the system can connect similar users and provide them with recommendations that have good precision.