Modelling User Preferences with Multi-Instance Genetic Programming

Amelia Zafra, Sebastián Ventura

In this paper we introduce a novel model for providing users with recommendations about web index pages of their interests. The approach proposed developes user profiles based on evolutionary multiinstance learning which determines what users find interesting and uninteresting by means of rules which add comprehensibility and clarity to user models and increase the quality of the recommendations. Experimental results show that our methodology achieves competitive results, providing high-quality user models which improve the accuracy of recommendations.

PDF full paper