A neural network approach to estimate student skill mastery in cognitive diagnostic assessments

• Author(s) / Creator(s)

  • Cui, Y.
  • Guo, Q.
  • Cutumisu, M.

• In computer-based tutoring systems, it is important to assess students’ mastery of different skills and provide remediation. In this study, we propose a novel neural network approach to estimate students’ skill mastery patterns. We conducted a simulation to evaluate the proposed neural network approach and we compared the neural network approach with one of the most widely used cognitive diagnostic algorithm, the DINA model, in terms of skill estimation accuracy and the ability to recover skill prerequisite relations. Results suggest that, while the neural network method is comparable in skill estimation accuracy to the DINA model, the former can recover skill prerequisite relations more accurately than the DINA model.

• Date created

  2017

• Subjects / Keywords

  • Prerequisite discovery
  • Student modeling
  • Cognitive diagnosis model
  • Skills
  • Neural network

• Type of Item

  Conference/Workshop Poster

• DOI

  https://doi.org/10.7939/R35H7C71D

• License

  Attribution-NonCommercial-NoDerivatives 4.0 International