Dynamic models for knowledge tracing & prediction of future performance

Large-scale data about learners' behavior are being generated at high speed on various online learning platforms. Knowledge Tracing (KT) is a family of machine learning sequence models that are capable of using these data efficiently with the objective to identify the likelihood of future learning performance. This study provides an overview of KT models from a technical and an educational point of view. It focuses on data representation, evaluation, and optimization, and discusses the underlying model assumptions such that the strengths and weaknesses with regard to a specific application become visible. Based on the need for advanced analytical methods suited for large and diverse data, we briefly review big data analytics along with KT learning algorithms' efficiency, learnability and scalability. Challenges and future research directions are also outlined. In general, the overview can serve as a guide for researchers and developers, linking the dynamic knowledge tracing models and properties to the learner's knowledge acquisition process that should be accurately modeled over time. Applied KT models to online learning environments hold great potential for the online education industry because it enables the development of personalized adaptive learning systems. Keywords-big data applications; educational data mining; knowledge tracing; sequential supervised machine learning.