Neural Deep Learning Models for Learning Analytics in a Digital Humanities Laboratory

Given the significance that technology-enhanced learning (TEL) receives in learning analytics, this study presents a neural deep learning architecture aimed to enhance a smart environment of a digital humanities laboratory. Presented a humanistic subject/controversy, the students interacted by producing multi-modal data (text, images, sound, movement) captured and processed with programmed devices. The smart environment induces emotional states during students' interactions and tasks in various scenarios, capturing images and facial expressions to classify student emotions. This study set up a preliminary experiment, testing the learning analytics model to show how it works and what outputs could be obtained. Leveraging the benefits of non-contact sensor technology, affective computing, and artificial intelligence, we collected biometric and neurocognitive data to establish correlations in the learning production of students studying humanistic and artistic topics. Despite the extended opinion expressed in the literature that the humanities do not have solid references on which to base learning analytics, this study shows that much quantitative data is available in such resources.