Active learning driven by rating impact analysis

• DOI: 10.1145/1864708.1864782
• Publication Date: 2010-09-28T13:41:50Z
• Authors: Carlos Eduardo Mello; Marie-Aude Aufaure; Geraldo Zimbrao
• Keywords: Training set; Data set

Many works have been proposed in order to improve the recommendation accuracy. Algorithms aiming accuracy developed and evaluated. These algorithms usually work with training data sets which are learned used make predictions on users' tastes. The set choice is a difficult task not only due technical nature of algorithm used, but also because user issues associated acquisition their opinions, since consist opinions. In this we show importance understanding impacted when rating acquired by developing naive active learning criterion based number predictions. To do that, Rating Impact Analysis method for user-based collaborative filtering applied issue.