PEER REVIEWED PAPERS BIG DATA, COGNITIVE COMPUTING AND INNOVATIVE TEACHING MODELS BIG DATA FOR SOCIAL MEDIA LEARNING ANALYTICS: POTENTIALS AND CHALLENGES Stefania Manca 1 Luca Caviglione 2 Juliana Elisa Raffaghelli 3 1 Institute of Educational Technology, National Research Council of Italy, stefania.manca@itd.cnr.it 2 Institute for Intelligent Systems for Automation, National Research Council of Italy, luca.caviglione@ge.issia.cnr.it 3 Independent researcher, jraffaghelli@gmail.com Keywords: MOOCs, social media, social learning analytics, open datasets, big data, privacy & security, ethics, data anonymization. Today, the information gathered from massive learning platforms and social media sites allows to derive a very comprehensive set of learning information. To this aim, data mining techniques can surely help to gain proper insights, personalise learning experiences, formative assessments, performance measurements, as well as to develop new learning and instructional design models. Therefore, a core requirement is to classify, mix, ilter and process the big data sources involved by means of proper learning and social learning analytics tools. In this perspective, this paper investigates the most promising applications and issues of big data for the design of the next-generation of massive learning platforms and social media sites. Speciically, it addresses the methodological tools and instruments for social learning analytics, pitfalls arising from the usage of open datasets, for citations: Journal of e-Learning and Knowledge Society Je-LKS The Italian e-Learning Association Journal Vol. 12, n.2, 2016 ISSN: 1826-6223 | eISSN: 1971-8829 Manca S., Caviglione L., Raffaghelli J.E. (2016), Big data for social media learning analytics: potentials and challenges, Journal of e-Learning and Knowledge Society, v.12, n.2, 27-39. ISSN: 1826-6223, e-ISSN:1971-8829