自动阅卷系统仿真  

The Simulation of Automatic Marking System

作　　者：李彩虹[1] 尹督荣 LI Cai-hong;YIN Du-rong(School of Information Science and Engineering,Lanzhou University,Lanzhou Gansu 730000,China)

机构地区：[1]兰州大学信息科学与工程学院,甘肃兰州730000

出　　处：《计算机仿真》2020年第9期157-161,共5页Computer Simulation

摘　　要：随着近年来深度学习技术的快速发展和应用,提出了一个基于汉语试卷主客观题的阅卷系统仿真方案并且研究了其用到的关键技术。通过利用手写板笔迹生成算法模拟考试作答手写笔迹,在对试卷作答部分进行扫描后,将手写文本图像切分为单个汉字图像;然后经预处理送入卷积神经网络进行识别和提取,同时利用循环神经网络训练自动阅卷时的参考答案;最后基于同义词词林,把提取到的答案和训练得到的参考答案进行语义相似度匹配并进行打分。实验结果表明,所提方法在一定程度上可以对手写汉语试卷的主客观题进行评判。With the rapid development and application of deep learning technology in recent years,this paper pro⁃posed a simulation scheme of the marking system based on the subject-objective questions of Chinese examination pa⁃pers,and studied the key technologies used.The handwritten handwriting as simulated by using the handwriting gen⁃eration algorithm of the handwriting board.After scanning the writing part of the paper,the handwritten text image was divided into a single Chinese character image.Then it was preprocessed and sent to CNN for identification and extrac⁃tion.At the same time,the RNN was used to train the reference answers for automatic marking.Finally,based on the synonym word forest,the extracted answers and the trained reference answers were matched for semantic similarity and scored.The experimental results show that the proposed method can be used to evaluate the subjective and objec⁃tive questions in Chinese test paper to some extent.

关 键 词：自动阅卷 手写文本识别 自然语言处理 卷积神经网络 循环神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]