Educational Data Classification Based on Deep Learning

doi:10.3969/j.issn.1671-1122.2023.03.010

Abstract

Abstract:

The continuous development of big data technology and the frequent occurrence of data leakage incidents have created an urgent need to protect data security in the education industry. In the education industry, it contains precise information on personal education and growth, which is of great value. Therefore, protecting educational data security has become an urgent need. To solve this problem, an educational data classification method based on deep learning is proposed in this paper. First, according to the role of data subject, three categories of personal data, organizational data, and business data were defined. Then, a Bi-LSTM neural network model combining based on word mixed embedding was proposed and implemented for automation and intellectualization of educational data classification. Finally, this paper validated the proposed classification method through experiments on two universities’ datasets. The experiment results show that the accuracy of our model can reach 95%, and all performance metrics are optimal compared with baselines.

Key words: educational data, data classification, deep learning, char-word mixture word representation, Bi-LSTM

CLC Number:

TP309

TAN Liuyan, RUAN Shuhua, YANG Min, CHEN Xingshu. Educational Data Classification Based on Deep Learning[J]. Netinfo Security, 2023, 23(3): 96-102.

Figures/Tables 8

References 16

[1]	STATISTA. Volume of Data Information Created, Captured, Copied, and Consumed Worldwide from 2010 to 2025[EB/OL]. (2021-06-07)[2022-06-10]. https://www.statista.com/statistics/871513/worldwide-data-created/.
[2]	ITRC. ITRC’s 2021 Q3 Data Breach Analysis and Key Takeaways[EB/OL]. (2021-10-06)[2022-06-10]. https://www.idtheftcenter.org/publication/2021-q3-data-breach-analysis/.
[3]	HONG Wei, REN Jianhong, XU Lina, et al. Classification of Educational Data Under the Background of Data Security Law[J]. Chinese Journal of ICT in Education, 2022, 28(3): 41-50.
	洪伟, 任剑洪, 徐丽娜, 等. 数据安全法实施背景下的教育数据分类分级研究[J]. 中国教育信息化, 2022, 28(3): 41-50.
[4]	WANG Zhen. Research on Data Classification and Grading-Proceed from Article 19 (1) of the Data Security Law of the People’s Republic of China (Draft)[D]. Beijing: Beijing Foreign Studies University, 2021.
	王真. 数据分级分类研究——从《中华人民共和国数据安全法(草案)》第19条第1款出发[D]. 北京:北京外国语大学, 2021.
[5]	BI Yunshan. Research of Chinese Text Classification Based on Deep Learning[D]. Hangzhou: Zhejiang University of Science and Technology, 2021.
	毕云杉. 基于深度学习的中文文本分类研究[D]. 杭州: 浙江科技学院, 2021.
[6]	WU Hongping, LIU Yuling, WANG Jingwen. Review of Text Classification Methods on Deep Learning[J]. Computers, Materials and Continua, 2020, 63(3): 1309-1321. doi: 10.32604/cmc.2020.010172 URL
[7]	MIKOLOV T, CHEN K, CORRADO G, et al. Efficient Estimation of Word Representations in Vector Space[J]. Computer Science, 2013, 40(1): 1-10.
[8]	PENNINGTON J, SOCHER R, MANNING C. GloVe: Global Vectors for Word Representation[C]// ACL. 2014 Conference on Empirical Methods in Natural Language Processing(EMNLP). New York: The Association for Computational Linguistics, 2014: 1532-1543.
[9]	DEVLIN J, CHANG Mingwei, LEE K, et al. BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding[C]// ACL. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. New York: The Association for Computational Linguistics, 2019: 4171-4186.
[10]	HOCHREITER S, SCHMIDHUBER J. Long Short-Term Memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[11]	SCHUSTER M, PALIWAL K K. Bidirectional Recurrent Neural Networks[J]. IEEE Transactions on Signal Processing, 1997, 45(11): 2673-2681. doi: 10.1109/78.650093 URL
[12]	GB/T 35273-2020 Information Security Technology-Personal Information Security Specification[S]. Beijing: Standards Press of China, 2020.
	GB/T 35273-2020 信息安全技术个人信息安全规范[S]. 北京: 中国标准出版社, 2020.
[13]	SUN Junyi. Jieba[EB/OL]. (2020-01-20)[2022-06-10]. https://github.com/fxsjy/jieba.
[14]	ZHANG Mohan. CNN-LSTM for Short Text Classification Based on Hybrid Word-Char Embeddings[J]. Information Technology and Informatization, 2019(1): 77-80.
	张默涵. 基于字词混合向量的CNN-LSTM短文本分类[J]. 信息技术与信息化, 2019(1): 77-80.
[15]	YANG Min, CHEN Xingshu, LUO Yonggang, et al. An Android Malware Detection Model Based on DT-SVM[J]. Security and Communication Networks, 2020, 1: 1-11. doi: 10.1002/(ISSN)1939-0122 URL
[16]	YOON K. Convolutional Neural Networks for Sentence Classification[C]// ACL. 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). New York: The Association for Computational Linguistics, 2014: 1746-1751.

实验环境	具体信息
操作系统	Windows 10
CPU	Intel(R) Core(TM) i7-9700
GPU	无
内存	16 GB
开发语言	Python 3.7.0
开发平台	PyCharm 2021.2.2

编号	数据项名	中文简称	类型	长度	约束	解释/举例	引用编号
JCXS010101	XH	学号	C	20	M	Complied by the school	JY/T 1002
JCXS020101	RXNY	入学年月	C	6	M	YYYYMM	—
JCTB010407	ZY	专业	C	12	0	—	—

数据集	阶段	个人数据/条	机构数据/条	业务数据/条
大学A	训练	231	201	1558
大学A	测试	92	68	693
大学B	训练	210	165	1107
大学B	测试	78	74	484

数据集	词嵌入	精确率	召回率	准确率	F1值
大学A	One-hot	75%	76%	88%	0.76
	Word2vec	88%	78%	92%	0.82
	GloVe	91%	55%	87%	0.63
	FasText	82%	77%	91%	0.79
大学B	One-hot	76%	83%	87%	0.79
	Word2vec	84%	82%	90%	0.83
	GloVe	71%	50%	80%	0.55
	FastText	82%	80%	90%	0.81

数据集	分类模型	精确率	召回率	准确率	F1值
大学A	DT-SVM	81%	80%	91%	0.80
	LSTM	88%	78%	92%	0.82
	CNN	82%	77%	90%	0.80
	Bi-LSTM	89%	82%	93%	0.85
	字词向量混合的Bi-LSTM	89%	86%	95%	0.88
大学B	DT-SVM	80%	81%	89%	0.80
	LSTM	86%	80%	91%	0.82
	CNN	84%	79%	90%	0.81
	Bi-LSTM	84%	82%	91%	0.83
	字词向量混合的Bi-LSTM	87%	86%	92%	0.86