Integrating a data mining engine into recommender systems

Abstract

History could be epitomised to a handful of events that changed the course of human evolution. Now, we found ourselves amid another revolution: the data revolution. Easily unnoticeable, this new outlook is shifting in every possible way how we interact with the internet and, for the first time in history, how the internet interacts with us. This new kind of interactions is defined by connections between users and consumable goods (products, articles, movies, etc.). And through these connections, knowledge can be found. This is the definition of data mining. Buying online has become mainstream due to its convenience and variety, but the enormous offering options affect negatively the user experience. Millions of products are displayed online, and frequently the search for the craved product is long and tiring. This process can lead to a loss of interest from the customers and, consequentially, losing profits. The competition is increasing, and personalisation is considered the game-changer for platforms. This article follows the research and implementation of a recommender engine in a well-known Portuguese e-commerce platform specialised in clothing and sports apparel, aiming the increase in customer engagement, by providing a personalised experience with multiple types of recommendations across the platform. First, we address the reason why implementing recommender systems can benefit online platforms and the state of the art in that area. Then, a proposal and implementation of a customised system are presented, and its results discussed.