Abstract

Computer-assisted marking can reduce the work pressure on teachers. This paper briefly introduced the automatic English subjective question marking algorithm combined with the convolutional neural network + long short-term memory (CLSTM) algorithm. An attention mechanism was introduced to improve the CLSTM algorithm. The improved CLSTM-based automatic marking algorithm was simulated and tested, and it was compared with the convolutional recurrent neural network (CRNN)-based automatic marking algorithm and the traditional CLSTM-based automatic marking algorithm. The results indicated that the improved CLSTM-based automatic marking algorithm effectively recognized the text in the handwritten answer images and accurately graded answers to subjective questions.

中文翻译：

---

基于深度学习的大学英语考试主观题自动阅卷算法研究

摘要

计算机辅助阅卷可以减轻教师的工作压力。本文简要介绍了结合卷积神经网络+长短期记忆（CLSTM）算法的自动英语主观问号算法。引入注意力机制来改进CLSTM算法。对改进的基于CLSTM的自动标记算法进行了仿真和测试，并与基于卷积循环神经网络（CRNN）的自动标记算法和传统的基于CLSTM的自动标记算法进行了比较。结果表明，改进的基于CLSTM的自动评分算法有效地识别了手写答案图像中的文本，并对主观问题的答案进行了准确的评分。