A Deep Learning System to Help Students Build Concept Maps

Nowadays, overwhelmed as people are by the amount of information available, it becomes more and more difficult to build an adequate cognitive process of knowledge building and discovery, on any one knowledge domain. Therefore, having knowledge-building tools at disposition, especially in the age of schooling, is of great importance. Knowledge building occurs by linking new concepts to already learned ones, thus connecting concepts together by means of semantic links representing their relationship. To accomplish this task, most of learners use Concept Maps, that is graphic tools, particularly suitable, to organize, represent and share knowledge. In fact, a Concept Map can explicitly express the knowledge of a person or group, about a given domain of interest: from primary school to university, and to professional/vocational training, Concept Maps can stimulate and unveil the occurrence of the so-called meaningful learning, according to the Ausubel’s learning theory. In this paper we investigate the use of a deep learning-based architecture, called TransH, designed for Knowledge Graph Embedding, to support the process of Concept Maps building. This approach has not been yet investigated for this particular educational task. Some preliminary case studies are discussed, confirming the potential of this approach.