ABSTRACT
The availability of educational big data catalyzes the rising tide of research on learner modeling for characterizing, evaluating, and promoting student learning in digital learning systems. Learning outcome modeling, a critical feature of computer-based assessments, is used to infer what students know and what they lack from their interactions with the assessment system. In addition, it is also used to make inferences on how assessment items associate with the latent skills measured by the assessment. This chapter gives an overview of the mainstream learner models that are commonly used in computer-based assessments for learning, as well as recent advances in learning outcome modeling. Techniques covered by this chapter include latent trait models, Bayesian networks, Bayesian knowledge tracing, deep learning (e.g., deep knowledge tracing), and collaborative filtering. The technical fundamentals, advantages, and challenges of each approach are reviewed by this chapter.
