A Case Study on The Application of Artificial Intelligence

in Education Industry

Qingyun Tang

The School of Business Admistration

South China University of Technology
Guangzhou, China

Email:158697435@qq.com

Abstract. In recent years, with the development of the internet, the application
of artificial intelligence has gradually expanded, particularly in the field of edu-
cation. In China, where the internet has been rapidly growing, the online educa-
tion industry has been flourishing. This emerging industry cannot be ignored.
However, most of the current research on the application of artificial intelligence
in education focuses on offline classrooms and theoretical studies, leaving a gap
in empirical research. Therefore, to fill this gap, this study is based on case anal-
ysis to explore how artificial intelligence is applied in the online education in-
dustry. The conclusions of this research will provide valuable insights for both
the academic community and the online education industry.

Keywords:artificial intelligence,application,education industry, recommenda-

tion algorithms

1 INTRODUCTION
Artificial intelligence has become an increasingly important topic in today's digital age.
With the explosion of digital information, businesses and organizations are now able to
collect vast amounts of data from various sources,which enhances the developement of
artifical intelligence in China.
China has experienced tremendous growth in its internet industry over the past few
decades, with over 989 million internet users, making it the country with the largest
number of internet users in the world. The internet infrastructure in China is highly
developed, with fast internet speeds and widespread connectivity. As a result of this
booming internet industry, a massive amount of user data has been generated, providing
companies in China with access to extensive data on consumer behavior, preferences,
and trends.By utilizing tagging tools and big data, Chinese companies applicate artifi-
cial intelligence on education industry.
The education system in China comprises 6 years of primary school, 3 years of junior
school, and 3 years of high school. During the transitions from junior school to high
school and from high school to university, students have to take exams that can signif-
icantly impact the trajectory of their lives, such as the entrance examination for high
© The Author(s) 2024
Y. Chen et al. (eds.), Proceedings of the 2023 3rd International Conference on Modern Educational Technology
and Social Sciences (ICMETSS 2023), Advances in Social Science, Education and Humanities Research 784,
https://doi.org/10.2991/978-2-38476-128-9_12
100 Q. Tang

school and the college entrance examination. To excel in these exams, students study
diligently and practice previous exam questions to prepare. However, a common chal-
lenge for students is identifying their weaknesses and strengths among the knowledge
they have learned in school and allocating their time effectively to overcome their
weaknesses and maximize their scores in limited time.
Some Chinese companies, such as Yuanfudao and Zuoyebang, have keenly recog-
nized this problem and have addressed it by applying artificial intelligence techniques.
Firstly, they collect exam questions and build a comprehensive database. Next, they
provide free online services for searching and practicing previous exam questions. Sub-
sequently, they process the behavioral data and question data to generate personalized
learning performance analysis reports. Lastly, they generate personalized learning plans
by applying recommendation algorithms. These learning plans automatically highlight
the knowledge points that are in the students' weakness zone, and suggest where stu-
dents should focus more effort to overcome their weaknesses. Additionally, if students
have a goal of entering a top university in China, the recommendation algorithms can
generate a tailored plan to help them achieve that goal.
This thesis aims to use case study to explain:
i)The technique way:How Chinese companies applied artificial intelligence in edu-
cation industry.
ii)The implications:What could we learn from the case.

2 METHODOLOGY

This article utilizes a case study method as the primary approach to study the applica-
tion of artificial intelligence in learning analytics. While much of the current academic
research on this topic remains in the conceptual stage, there is a lack of empirical evi-
dence on implementation cases in real-world settings. The case study method is well-
suited to fill this gap by providing in-depth insights into how companies actually im-
plement learning analytics and summarizing their practical experiences in this area.

3 LITERATURE REVIEW AND THEORETICAL

BASIS

3.1 Literature Review

There are several different perspectives in the academic community regarding the ap-
plication of artificial intelligence in education. Some are based on its application in
offline classrooms, such as research on the integration of artificial intelligence educa-
tion with kindergarten teaching [1] and research in primary and middle school stages.
For example, Zhu Zhe et al. conducted research on how the era of artificial intelligence
can be integrated with classroom teaching in the context of mathematics education [2].
Zhou Yajian et al. proposed suggestions for advancing the integration of artificial in-
telligence into teaching [3]. Some studies focus on adult education. Meng Yuanhang
explores the implementation of artificial intelligence in adult lifelong education [4],
 A Case Study on The Application of Artificial Intelligence 101

while Xu Xuetian researches the combination of vocational education and artificial in-
telligence [5]. There is also research on the opportunities and challenges of integrating
ideological and political education for university students with artificial intelli-
gence[6]And some research on the implications it can reveal for future of
schools[12].However some people fears about AI and its use in education[11].
In terms of theory, most of the literature on the application of artificial intelligence
in education mentions value logic, goal logic, and practice logic [7][8]. Based on these
logics, three main practical directions are virtual and real-world applications, personal-
ized education and learning assessment, and data and natural language-based pro-
cessing. Zhang Huifeng et al. conducted a detailed analysis of the application of artifi-
cial intelligence in teaching based on educational learning assessment and proposed
recommendations for precision teaching [9], including features like image-based ques-
tion search. Du Juan et al. [10] mentioned in their article that the integration of tech-
nologies such as natural language processing and image recognition into teaching has
made greater progress compared to traditional teaching methods.
Since the aforementioned articles are primarily based on research in offline class-
rooms or adult and vocational education, and the theoretical research lacks empirical
studies to some extent, the significance of this study lies in exploring the application of
artificial intelligence in online education teaching based on educational learning assess-
ment. The study aims to fill the research gap in empirical studies by investigating actual
company cases in the rapidly developing field of online education in China.

3.2 Theoretical Basis

Artificial Intelligence (AI) refers to the development and implementation of computer
systems or machines that can perform tasks that typically require human intelligence.
AI aims to simulate and replicate human cognitive processes, such as learning, reason-
ing, problem-solving, perception, and decision-making, using algorithms and compu-
tational models.
Artificial intelligence can be classified according to different methods. Here is a
common classification, including the concept of labeling:
1.Weak AI (Narrow AI): Also known as specialized artificial intelligence, it refers
to AI systems designed and trained for specific tasks or domains. These systems can
exhibit human-like intelligence in specific tasks but may perform weakly in other tasks.
Labeling and simple recommendations fall into the category of weak AI as they are
designed to handle specific datasets or user needs.
2.Strong AI (General AI): Refers to AI systems that possess human-level intelli-
gence and can exhibit human-like intelligence across a wide range of tasks and do-
mains. The goal of strong AI is to mimic human cognitive abilities and wisdom, includ-
ing understanding, learning, and reasoning capabilities.
3.Machine Learning: It is a branch of artificial intelligence that uses algorithms and
models to enable machines to learn and improve performance from data without explicit
programming instructions. Machine learning can learn and make predictions from train-
ing data and can be used for tasks such as classification, regression, and clustering. In
the context of labeling, machine learning algorithms can learn the associations between
102 Q. Tang

keywords and labels from annotated data, enabling automatic labeling of new unlabeled
data.
4.Deep Learning: It is a subfield of machine learning that mimics the structure and
functioning principles of the human brain's neural networks. Deep learning utilizes
multi-layered neural networks for learning and decision-making, and through training
on large-scale datasets, it can acquire more complex and advanced representation and
pattern recognition capabilities. Deep learning has achieved significant advancements
in areas such as image processing, speech recognition, and natural language processing
and can be applied in labeling for automated feature extraction and learning label asso-
ciations.
In labeling, artificial intelligence can automate the process of assigning labels by
utilizing machine learning and deep learning techniques. It can learn to extract features
from input data and map them to corresponding labels. This automated labeling process
improves efficiency and reduces the manual workload.

4 ANALYTICAL REVIEW OF THE SELECTED

CASES

This case study presented here is based on the application of Narrow AI and try to study
how Chinese online education companies utilize tagging tools to achieve intelligent
recommendations.

4.1 Case Background

Yuanfudao and Zuoyebang are two prominent online education platforms in China that
utilize AI technology to assist students in their learning, particularly in artificial intel-
ligence homework tutoring.
Yuanfudao, founded in 2012, has grown rapidly to become a valued over $15 billion
company. Its platform offers personalized online courses, AI-powered homework tu-
toring, and one-on-one online tutoring services for K-12 students. Through adaptive
learning algorithms, Yuanfudao's AI technology tailors course content to each student's
learning style and progress, providing a personalized learning experience in subjects
such as mathematics, English, physics, chemistry, Chinese language arts, and extracur-
ricular classes in painting, piano, and chess. Yuanfudao aims to make quality education
accessible to everyone and has become one of the most popular and influential online
education companies in China.
Zuoyebang, founded in 2015, has quickly become one of the largest online education
platforms in China, valued at over $10 billion. Its main service is artificial intelligence
homework tutoring, where students can take a photo of their homework question and
receive a step-by-step explanation from a virtual tutor. The platform also offers online
classes and one-on-one tutoring in subjects such as math, English, physics, as well as
exam preparation courses for major standardized tests. Zuoyebang focuses on using
technology to improve access to education and has become a popular resource for stu-
dents seeking academic support in China.
 A Case Study on The Application of Artificial Intelligence 103

4.2 Overview of Functions

The popular features mentioned above are based on artificial intelligence applications:
1)artificial intelligence homework tutoring:students who take a photo of their home-
work question will receive a step-by-step explanation from a virtual tutor.
In this application scenario, students upload exam questions by taking photos with
their smartphones. The backend system first performs semantic analysis on the ques-
tions and matches them with the corresponding keyword tags in the tag library. Then,
the system retrieves the relevant exam questions by querying the corresponding tags (in
this process, it follows an exact-to-fuzzy search method).
2)Personalized online courses: develop adaptive learning algorithms that personalize
course content to each student's learning style and progress.
In this application scenario, when students use the app to do exercises,the app will
automatically captures the user's behavior, such as the time spent on answering ques-
tions and the number of times the answer option is revised. After the user submits the
answers, the system calculates the user's error rate and the error rate of different specific
knowledge points based on the tags of the exam questions. Using recommendation al-
gorithms, the system analyzes the student's weak knowledge points and provides per-
sonalized improvement plans accordingly.
Figure 1 organizes the implementation logic of the entire core technology. Tags and
algorithm recommendation are the key technical aspects that enterprises utilize to apply
artificial intelligence in education. Enterprises collect a large amount of exam question
data and assign relevant tags to each exam question when storing the data in the data-
base. These tags may include subject, grade, specific knowledge points, year of exam-
ination, and other relevant information.

Fig. 1. Exam Questions Recommendation System Deployment Diagram.

104 Q. Tang

4.3 Technical Implementation Details

Here is the database design and recommendation process.

1)Database and Table Design

In the realm of database design, it is common practice to establish separate data marts
for user-related, exam-related, and tag-related data. Corresponding tables are then
stored within their respective data marts.
For instance, taking the user profile data mart as an example, it is customary to create
a data table for user profile data(Figure 2) and another data table for user behavior-
related data(Figure 3) within this data mart. Here are some exemplary fields that could
be included in these data tables:

i)User Profile Table:.

Fig. 2. User Profile Data Table.

Field 1: Index - Index number for the user record

Field 2: User ID - Unique identifier for the user
Field 3: Grade - Grade level of the user
Field 4: Subject - Subject of study for the user
Field 5: Region - Geographic region associated with the user
Field 6: Tags - Tags or labels associated with the user
Field 7: Update Time - Time of last update for the user record
Field 8: Create Time - Time of creation for the user record

ii)User Behavior Data Table:.

Fig. 3. User Behavoir Data Table.

 A Case Study on The Application of Artificial Intelligence 105

Field 1: Index - Index number for the behavior record

Field 2: User ID - User ID linked to the User Profile Table
Field 3: User Behavior - Type of user behavior, such as browsing, liking, etc.
Field 4: Duration - Duration of the behavior (if applicable)
Field 5: Question ID - ID of the question associated with the behavior
Field 6: Tags - Tags or labels associated with the behavior
Field 7: Create Time - Time of creation for the behavior record
Taking the exam question data mart as an example, it typically houses data tables
such as the question data table(Figure 4) and the user's mistaken questions data ta-
ble(Figure 5). Example fields for these data tables are as follows:

iii)Question Data Table:

Fig. 4. Questions Data Table.

Field 1: Index - Index number for the question record

Field 2: Question ID - Unique identifier for the question
Field 3: Subject - Subject of study for the question
Field 4: Grade - Grade level of the question
Field 5: Knowledge - Specific knowledge point of the question
Field 6: Question Content - Content of the question
Field 7: Options - Multiple-choice options for the question (if applicable)
Field 8: Answer - Correct answer for the question
Field 9: Difficulty Level - Difficulty level of the question
Field 10: Tag ID - Unique identifier for the tags or labels associated with the question
Field 11: Create Time - Time of creation for the question record

iv)User's Mistaken Questions Data Table:.

Fig. 5. User’s Mistaken Question Data Table.

106 Q. Tang

Field 1: Index - Index number for the mistaken question record

Field 2: User ID - User ID linked to the User Profile Table
Field 3: Question ID - ID of the mistaken question linked to the Question Data Table
Field 4: Mistake Time - Time when the user made the mistake
Field 5: Mistake Type - Reason or explanation for the mistake
Field 6: Tags ID - Unique identifier for the tags or labels associated with the mis-
taken question
Field 7: Create Time - Time of creation for the mistaken question record
Using the tag data mart as an example, this data mart typically involves cross-refer-
encing users and exam questions, and therefore categorizes tags based on categories.
There would have exam question tags data table(Figure 6) and user tags data table(Fig-
ure 7).Example fields for the data tables are as follows:

v)Exam Question Tags Table:

Fig. 6. Exam Question Tags Data Table.

Field 1: Index - Index number for the tag record

Field 2: Tag ID - Unique identifier for the tag
Field 3: Tag Name - Name of the tag
Field 4: Tag Property - Properties or attributes associated with the tag
Field 5: Update Time - Time of last update for the tag record
Field 6: Create Time - Time of creation for the tag record

vi)User Tags Table:

Fig. 7. User Tags Data Table

Field 1: Index - Index number for the user tag record

Field 2: Tag ID - Unique identifier for the tag
Field 3: Tag Name - Name of the tag
Field 4: Tag Property - Properties or attributes associated with the tag
 A Case Study on The Application of Artificial Intelligence 107

Field 5: Update Time - Time of last update for the user tag record
Field 6: Create Time - Time of creation for the user tag record

2)Recommendation Process
i)Tagging Exam Questions:Exam questions are batch-entered by staff in the
backend, and tags such as grade level, subject, region, knowledge point, and difficulty
are added to the exam questions.
ii)Collecting Behavioral Data:When users use the question answering function in the
app, the app records their behaviors and sends them to the cloud server, which then
records them in the user behavior data table.
iii)Insert Performance Data:After users complete their questions, the app sends their
performance to the cloud server and records it in the user error question data table.
iv)Tagging Users:At this point, the server will use a set of rule-based algorithms to
automatically tag users based on their behaviors and error question situations, such as
weak foundation in English, weak in relative clause knowledge, strong in gerunds, and
so on.
v)Recommendation:After tagging, the system will automatically match the relevant
tags associated with the exam question data table based on the user's tags, and generate
a study plan and an output plan based on rule-based algorithms, which will include the
exam questions that need to be reviewed and mastered for each module.

5 DISCUSSION

This case study is solely based on the analysis of two Chinese online education compa-
nies. There is a limitation in the number of case samples available. However, as these
two companies are leading players in the domestic online education industry in China,
this study considers the case to still be somewhat applicable. Moreover, the learning
objectives of users differ across different stages of education. For instance, in the stage
of quality education, the ultimate assessment goal is not solely based on scores, render-
ing AI recommendation algorithms based on incorrect answers inapplicable. In the fu-
ture, research will be conducted on the application of artificial intelligence based on
different stages of learning as a topic to continuously improve the overall research con-
tent.

6 CONCLUSIONS

The successful experience of Chinese enterprises in applying artificial intelligence to

education indicates that artificial intelligence will become an important tool in the ed-
ucation field, expected to help educational institutions better meet the needs of students
and the market, and improve the quality and efficiency of education. This also suggests
that the application prospects of artificial intelligence in the education field are vast.
108 Q. Tang

Starting from user needs, it's not always necessary for users to require very detailed
or high-end learning reports. The first step in applying artificial intelligence to educa-
tion is to provide basic statistics on the user's learning situation by integrating existing
data. Next, we can collaborate with teachers to integrate content related to knowledge
points and examination outlines, and use algorithms to calculate different learning plans
and breakthrough plans for different students at different stages.

REFERENCE
1. Wei Can. Reflections on AI Education Activities in Kindergartens in the Era of Artificial
Intelligence. China Modern Education Equipment, 2023(10):64-66+70. DOI:
10.13492/j.cnki.cmee.2023.10.018.
2. Zhu Zhe, Wang Minxia. Mathematics Education in the Era of Artificial Intelligence. Re-
search on Classroom Teaching in Primary and Secondary Schools, 2023(06):1-6.
3. Zhou Yajian, Lu Xiaohong. Intelligent Education in Primary and Secondary Schools in the
Era of Artificial Intelligence. China Education Journal, 2023(S1):6-8.
4. Meng Yuanhang. Exploration of Adult Lifelong Education in the Context of Artificial Intel-
ligence. Adult Education, 2022, 42(12):8-13.
5. Xu Xuetian. Development Opportunities and Challenges of Vocational Education in the
Context of Artificial Intelligence. Research on Continuing Education, 2023(08):88-92.
6. Hu Jian, Zhu Yulian. Opportunities, Challenges, and Countermeasures for Ideological and
Political Education of University Students in the Era of Artificial Intelligence. Journal of
Chongqing Three Gorges University, 2023, 39(04):21-29. DOI: 10.13743/j.cnki.issn.1009-
8135.2023.04.003.
7. Sui Yifan, Xing Taiqi. Challenges and Responses: Innovation and Development Logic of
Higher Education in the Era of Artificial Intelligence. Research on Science and Education
Development, 2023, 3(01):76-95.
8. Zhou Zibo, Zhang Xinhua. Empowering Online Education with Artificial Intelligence:
Logic, Mechanism, and Path. Adult Education, 2023, 43(07):52-58.
9. Zhang Huifeng, Chen Zhuxiang, Xu Shuijing. New Transformation of Higher Education
Models Empowered by Artificial Intelligence. Software Guide, 2022, 21(11):166-171.
10. Du Juan, Wang Linlin. Application Research of Artificial Intelligence Technology in Edu-
cation. Internet of Things Technology, 2023, 13(06):157-159. DOI: 10.16667/j.issn.2095-
1302.2023.06.042.
11. Latifa Douali,Sabah Selmaoui,Widad Bouab.Artificial Intelligence in Education: Fears and
Faiths.International Journal of Information and Education Technology. Volume 12 , Issue
7,2022,DOI：10.18178/IJIET.2022.12.7.1666.
12. Gocen Ahmet,Aydemir Fatih.Artificial Intelligence in Education and Schools. Research on
Education and Media. Volume 12 , Issue 1 . 2021. PP 13-21,DOI：10.2478/REM-2020-
0003.
 A Case Study on The Application of Artificial Intelligence 109

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-
NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/),
which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any
medium or format, as long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's
Creative Commons license, unless indicated otherwise in a credit line to the material. If material
is not included in the chapter's Creative Commons license and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder.