Estudo da percepção humana e mineração de preferências contextuais na recomendação de imagens para o usuário

Abstract

This work presents a study on the use of information of visual perception in the image recommendation. Thus, it proposed a method that makes the grouping of users through your visual perception and their similarity. This method was denominated VP-Similarity. The VP-Similarity is implemented by extending the system PrefRec to use of visual perception data in conjunction with preference data item and the recommendation task, thereby forming the VP-PrefRec. For the validation of the system a database was created. This database contains preference and visual perception data of users. In addition, VP-Similarity was also applied on a social recommendation system, providing a visual perception network. The purpose of this network is to minimize the user cold start problem, which in most recommender systems. The objective of purpose of this paper is to show that the recommendation of images with data items and contextual preferences for a target user has better quality when considering only those users who have similar visual perceptions that target user, considering some prior evaluations of that user target. And it also shows that it is possible to improve the quality of images of recommendations for a new user who has never made use of the system (cold start), focusing only on users to visually perceive the similar images.