View article

The problem of domain modeling is to represent domain content so that it can be efficiently authored, optimally delivered to students, and precisely tracked with respect to student mastery. While the previous section focused primarily on the representation aspect of domain modeling, the present section focuses on various methods and concerns related to authoring, delivery, and mastery.

An overarching method that unites most of the chapters in the present section is a data-driven approach to domain modeling that uses machine learning. This is consistent with the current zeitgeist of Big Data, enabled by Internet-scale data sets and cloud-computing resources. With these data, researchers are able to parameterize increasingly complex models from data, perform model selection on alternatives of such models, and even author content using crowdsourcing or semi-supervised machine learning.