Title: Using Support Vector Machines for feature-oriented profile-based recommendations

Authors: Angel Garcia-Crespo, Juan Miguel Gomez-Berbis, Ricardo Colomo-Palacios, Francisco Garcia-Sanchez

Addresses: Computer Science Department, Universidad Carlos III de Madrid, Av. Universidad 30, Leganes, 28911, Madrid, Spain. ' Computer Science Department, Universidad Carlos III de Madrid, Av. Universidad 30, Leganes, 28911, Madrid, Spain. ' Computer Science Department, Universidad Carlos III de Madrid, Av. Universidad 30, Leganes, 28911, Madrid, Spain. ' Computer Science Department, Universidad de Murcia, Campus de Espinardo, Espinardo, 30180, Murcia, Spain

Abstract: Recommendations have reached a new dimension in a ubiquitous computing environment. A vast amount of information is coming from pervasive services and devices, providing a potential source of value-added knowledge. However, this knowledge must be exposed through a classification technique such as Support Vector Machines (SVMs), which would allow categorising, classifying and evaluating this information based on features and profiles to foster the efficiency of recommendations of informational items. This paper presents an algorithm based on SVMs and an underlying architecture to extract and build a number of recommendations based on features and profile preferences from the user.

Keywords: recommendation systems; SVM; support vector machines; semantic technologies; artificial intelligence; ubiquitous computing; classification; user profiling; user profiles; user preferences.

DOI: 10.1504/IJAIP.2009.026762

International Journal of Advanced Intelligence Paradigms, 2009 Vol.1 No.4, pp.418 - 431

Published online: 25 Jun 2009 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article