Nowadays, the Internet connects people, multimedia and physical objects leading to a new-wave of services. This includes learning applications, which require to manage huge and mixed volumes of information coming from Web and social media, smart-cities and Internet of Things nodes. Unfortunately, designing smart e-learning systems able to take advantage of such a complex technological space raises different challenges. In this perspective, this paper introduces a reference architecture for the development of future and big-data-capable e-learning platforms. Also, it showcases how data can be used to enrich the learning process.