A Recommender System Based on a Machine Learning Algorithm for B2C Portals

Users of B2C (Business-to-Consumer) portals often lack of detailed knowledge about the state of the market related to the product they want to purchase. This leads to consumers purchasing the most popular product of a category, although perhaps that is neither the best suited for their requirements nor at the best cost. In this paper, a methodology to develop a recommender system is proposed. Our proposal is based on a supervised learning approach to infer knowledge that allows consumers to unveil what the existing patterns among the features that describe the searched product are. Such knowledge allows consumers to learn what they can buy and at what cost.