Learning Analytics (LA) is a recent research branch that refers to methods for measuring, collecting, analyzing, and reporting learners’ data, in order to better understand and optimize the processes and the environments. Knowledge Tracing (KT) deals with the modeling of the evolution, during the time, of the students’ learning process. Particularly its aim is to predict students’ outcomes in order to avoid failures and to support both students and teachers. Recently, KT has been tackled by exploiting Deep Learning (DL) models and generating a new, ongoing, research line that is known as Deep Knowledge Tracing (DKT). This was made possible by the digitalization process that has simplified the gathering of educational data from many different sources such as online learning platforms, intelligent objects, and mainstream IT-based systems for education. DKT predicts the student’s performances by using the information embedded in the collected data. Moreover, it has been shown to be able to outperform the state-of-the-art models for KT. In this paper, we briefly describe the most promising DL models, by focusing on their prominent contribution in solving the KT task.

Deep learning for knowledge tracing in learning analytics: an overview

Luca Grilli;Pierpaolo Limone;Domenico Santoro;
2021-01-01

Abstract

Learning Analytics (LA) is a recent research branch that refers to methods for measuring, collecting, analyzing, and reporting learners’ data, in order to better understand and optimize the processes and the environments. Knowledge Tracing (KT) deals with the modeling of the evolution, during the time, of the students’ learning process. Particularly its aim is to predict students’ outcomes in order to avoid failures and to support both students and teachers. Recently, KT has been tackled by exploiting Deep Learning (DL) models and generating a new, ongoing, research line that is known as Deep Knowledge Tracing (DKT). This was made possible by the digitalization process that has simplified the gathering of educational data from many different sources such as online learning platforms, intelligent objects, and mainstream IT-based systems for education. DKT predicts the student’s performances by using the information embedded in the collected data. Moreover, it has been shown to be able to outperform the state-of-the-art models for KT. In this paper, we briefly describe the most promising DL models, by focusing on their prominent contribution in solving the KT task.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11369/398698
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 4
  • ???jsp.display-item.citation.isi??? ND
social impact