Um sistema de recomendação para mídias baseado em conteúdo nebuloso

Abstract

This work presents the development of a content-based recommender system for media of the AMMO (Authoring and Manipulation of Multimedia Objects) environment s multimedia database. This environment keeps inexact semantic information that describes the media. This project s objective was to recommend media using this fuzzy information. Other informations that should be considered are the of user s interest evolution, relevance feedback and the inference of the user s interest, based on his navigation in the database s multimedia applications. These multimedia applications are presented by the system in the form of a graph, on an intuitive and easy to use graphical interface. The developed recommendation technique uses statistical tests, match algorithms and inexact knowledge processing through fuzzy rules.