工业技术与职业教育2025，Vol.23Issue(5)：60-68,9.
基于符号图卷积网络的高职教育隐性学习困境学生群体行为模式探究与预测
Behavioral Pattern Analysis and Prediction of Latent Learning Difficulties in Vocational Education Using SGCN
摘要
Abstract
In the context of vocational education,latent underachievers often exhibit seemingly normal learning behaviors while lacking intrinsic motivation and effective strategies,making early identification challenging.This study analyzes online learning behavior data from 417 students in a vocational college,constructing a signed student relationship graph and employing a Signed Graph Convolutional Network(SGCN)to classify and predict latent underachievers.By distinguishing positive and negative relational paths through friend-enemy aggregation,the model effectively captures group behavior patterns and improves identification accuracy.Experimental results demonstrate that SGCN outperforms traditional GCN and MLP models in accuracy,F1 score and interpretability.The findings offer practical insights for data-driven academic early warning systems and targeted educational interventions,contributing to quality enhancement in vocational education.关键词
隐性学困生/在线学习行为/符号图卷积网络/图神经网络/学业预警Key words
latent underachievers/online learning behavior/signed graph convolutional network/graph neural networks/academic early warning分类
社会科学引用本文复制引用
高智良..基于符号图卷积网络的高职教育隐性学习困境学生群体行为模式探究与预测[J].工业技术与职业教育,2025,23(5):60-68,9.基金项目
湖南省社科基金项目青年项目"人工智能就业替代效应下高职教育的适应性研究"(课题编号:23YBQ142),主持人高智良 （课题编号:23YBQ142）
湖南省教育厅科学研究项目"基于符号图卷积网络的在线学习行为预警SSL模型研究"(课题编号:23C1114),主持人高智良. （课题编号:23C1114）
