基于图注意力网络的自动化教学系统创新设计  

Design of automatic online examination scoring system based on graph attention network

作　　者：南姣鹏[1] NAN Jiaopeng(Xianyang Normal University,Xianyang Shanxi 712000,China)

机构地区：[1]咸阳师范学院,陕西咸阳712000

出　　处：《自动化与仪器仪表》2025年第3期215-219,共5页Automation & Instrumentation

基　　金：咸阳师范学院专项课题《共建共享下的学前教育数字化资源开发策略研究》(2024XSYH059)。

摘　　要：针对在线考试自动评分准确率低,导致自动化教学效果不佳的问题,提出设计一个基于B/S架构的在线考试自动评分系统。首先,对自动评分系统进行整体搭建;然后构建一种基于图注意力网络的考试自动评分模型,通过该模型进行词向量生成和特征向量提取;最后计算学生考试相似度,由此实现在线考试自动评分。结果表明,在相同测试集下,本模型的精确率、召回率和F1分数分别取值为93.14%、96.57%和95.02%,相较于传统的GCN模型、KNN模型和LCS自动评分方法,本模型的评分精度更高,满足自动化教学系统的自动准确评分需求,进一步验证了将人工智能与教育方式相结合,能够实现学前教育自动化教学系统的有效创新。Aiming at the low accuracy of automatic scoring and the poor effect of automatic teaching,an automatic scoring system based on B/S architecture is proposed.Firstly,the automatic scoring system is set up,then an automatic scoring model based on graph attention network is constructed,through the word vector generation and feature vector extraction;finally the student test simi-larity is calculated to realize automatic scoring of online examination.The results show that under the same test set,the precision rate,recall rate and F1 score of this model are 93.14%,96.57%and 95.02%,respectively.Compared with the traditional GCN model,KNN model and LCS automatic scoring method,the scoring accuracy of this model is higher,which meets the requirements of automatic and accurate scoring of automatic teaching system,and further proves that the combination of artificial intelligence and edu-cation mode can realize the effective innovation of automatic teaching system in preschool education.

关 键 词：人工智能 自动评分 抽象语法树 图注意力网络 教学系统 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]