Learning analytics with multimodal data through the lens of AI in education: A systematic literature review

Abstract

Learning analytics (LA) collects, processes, analyzes and displays various types of learner data through various means, to give insights that support learning. Combining different data sources with possibly different modalities can further enhance these insights. This field of research is called multimodal learning analytics (MMLA). The availability of substantial amounts of data opens up possibilities for AI to further analyze MMLA data and use these analyses to help learners. In this paper, we systematically collect and analyze MMLA studies that make use of AI technologies to tackle research questions related to improving learning processes. Our systematic literature review gives an overview of the current state of the art of MMLA, provides results about the types of research questions, data types and how they are integrated, their target audiences, and the AI algorithms and methods used. We finish with providing insights about the current challenges in the field and possible future directions.