SuggestMe: um sistema de recomendação utilizando web semântica para evitar o cold start

Abstract

Recommender systems are increasingly present in the daily lives of users on the Internet, they are used in e-commerce and various other applications, automating the natural social process of recommendation and influecing in the final decision of the user. However, the recommendation systems have many challenges to make recommendations at the beginning, because there is not sufficient data. This problem is called cold start. To solve this problem, a common solution is to apply questionnaires to obtain initial user data, however, this creates another barrier for user registration, introducing extra steps to create a user account on a system. This work proposes a recommendation system to solve the problem of cold start using data available on the web. You can obtain the user data through a social network and therefore, perform the recommendation using semantic web. It was implemented a prototype solving the problem of cold start getting Facebook data and performing recommendations using semantic web via Freebase.