Self-supervised contrastive learning with multi-dimensional graph structures for knowledge tracing

doi:10.1016/j.neucom.2026.133837

Self-supervised contrastive learning with multi-dimensional graph structures for knowledge tracing

2026-05-04

0

PRE

AI

DOI：10.1016/j.neucom.2026.133837

Original

Original request for help

Save

Abstract

En 中文

• Integrates heterogeneous and interaction-directed graphs to model learning behaviors. • Employs MAGNN and DGCNs to extract structural features from student interactions. • Introduces self-supervised graph contrastive learning to improve embedding quality.

Keywords:

Graph Contrastive Learning

Multi-dimensional Graphs

Knowledge Tracing

Student Interaction Modeling

Embedding Quality

Self-supervised contrastive learning with multi-dimensional graph structures for knowledge tracing

Abstract

Journal

Organization