Sistemas de Recomendação: abordagens utilizando dados implícitos

Abstract

Due to the e-commerce growth, the crave for mechanisms capable of personalizing user experience has been facing a continuous growth. Considering that fact, the use of Recommender Systems have been one of the biggests bets to build stronger and loyal relations between clients and the commerce itself. Getting relevant recommendations, however, is a common struggle, especially when it’s not possible to get user preferences in an explicit way. Due to this demand, this study aimed to explore differents recommen dation models and generate recommendations with similar or better hit ratios than the recent researches, with models that are able to observe user preferences even when not explicit. By using Recommender Systems built with neural networks found on the state of the art, this study achieved promising and positives results when it comes to good recommendations, even with hidden user preferences.