Concept Prerequisite Relation Prediction by Using Permutation-Equivariant Directed Graph Neural Networks

TL;DR

This paper introduces a permutation-equivariant directed graph neural network model for concept prerequisite relation prediction, improving expressivity and prediction accuracy in educational graph tasks.

Contribution

The paper proposes a novel permutation-equivariant directed GNN incorporating the Weisfeiler-Lehman test, enhancing graph isomorphism handling for CPRP.

Findings

Outperforms state-of-the-art methods on three datasets

Demonstrates improved prediction accuracy

Enhances GNN expressivity for directed graphs

Abstract

This paper studies the problem of CPRP, concept prerequisite relation prediction, which is a fundamental task in using AI for education. CPRP is usually formulated into a link-prediction task on a relationship graph of concepts and solved by training the graph neural network (GNN) model. However, current directed GNNs fail to manage graph isomorphism which refers to the invariance of non-isomorphic graphs, reducing the expressivity of resulting representations. We present a permutation-equivariant directed GNN model by introducing the Weisfeiler-Lehman test into directed GNN learning. Our method is then used for CPRP and evaluated on three public datasets. The experimental results show that our model delivers better prediction performance than the state-of-the-art methods.