A Recommender System Based on a Multi-Agent System to Create Virtual Academic Web Communities

Silvio Cazella Computing Science Department University of Rio Grande do Sul Brazil

DEPARTMENT SEMINAR DATE: Thursday, June 17, 2004 TIME: 3:30 p.m. PLACE: Agriculture 2E25 *** Everyone is welcome ***

Abstract

The Internet has served as a virtual meeting place for people sharing common interests. These interest groups are called virtual web communities. In this presentation I address the problem of creating virtual academic web communities for researchers in Brazil, taking into consideration information from: a) a hybrid recommender system, b) the Brazilian e-government's system, Curriculum Vitae Lattes. This presentation provides an overview of the architecture of W-RECMAS (a RECommender system to Web based on Multi-Agent System for academic paper recommendation). W-RECMAS proposes an alternative solution to the data overload problem, by use of a recommender system, combined with a multi-agent system and data mining techniques, to identify and create virtual academic web communities. We also present a new concept named Recommender's Rank, which is used to facilitate the process of community creation.

About the speaker

Silvio Cesar Cazella is a PhD candidate from Computing Science Department at University of Rio Grande do Sul in Brazil. His research is concentrated on Artificial Intelligence and Data Mining. He has taught Computing Science and Information Systems courses in University of Vale do Rio dos Sinos in Brazil. He is currently continuing his research at the University of Alberta in an international exchange program.

[an error occurred while processing this directive]