Content uploaded by Ossama Embarak
Author content
All content in this area was uploaded by Ossama Embarak on Nov 09, 2022
Content may be subject to copyright.
1
An adaptive paradigm for smart education
systems in smart cities using the internet of
behaviour (IoB) and explainable artificial
intelligence (XAI)
Dr. Ossama Embarak
Higher Colleges of Technology
Fujairah, UAE
ORCID: 0000-0003-4504-8561
oembarak@hct.ac.ae
Abstract—The rapid shift towards smart cities, particularly in
the era of pandemics, necessitates the employment of e-learning,
remote learning systems, and hybrid models. Building adaptive
and personalized education becomes a requirement to mitigate the
downsides of distant learning while maintaining high levels of
achievement. Explainable artificial intelligence (XAI), machine
learning (ML), and the internet of behaviour (IoB) are just a few
of the technologies that are helping to shape the future of smart
education in the age of smart cities through Customization and
personalization. This study presents a paradigm for smart
education based on the integration of XAI and IoB technologies.
The research uses data acquired on students' behaviours to
determine whether or not the current education systems respond
appropriately to learners' requirements. Despite the existence of
sophisticated education systems, they have not yet reached the
degree of development that allows them to be tailored to learners'
cognitive needs and support them in the absence of face-to-face
instruction. The study collected data on 41 learner's behaviours in
response to academic activities and assessed whether the running
systems were able to capture such behaviours and respond
appropriately or not; the study used evaluation methods that
demonstrated that there is a change in students' academic
progression concerning monitoring using IoT/IoB to enable a
relative response to support their progression.
Keywords: Smart education, Explainable artificial intelligence
(XAI), the internet of behaviour (IoB) , Smart cities education,
adaptive learning, future AI-based education.
1. Introduction
Education and learning are no longer confined to the confines
of schools and classrooms. Technology's advancement has
altered the way educational institutions operate. The Internet
of Things (IoT) has the ability to connect a wide range of
objects to the Internet. In a wireless context, artificial
intelligence (AI) and machine learning (ML) have the ability
to make computers think like and mimic people. These
technologies have an impact on every aspect of people's life.
They have a wide range of applications in education,
manufacturing, healthcare, transportation, smart cities, and
energy.
The education/learning area is not immune to technological
advancement. The Internet of Things has played a significant
role in engaging learners in the classroom. The Internet of
Things has assisted instructors in making their classrooms
more dynamic and interesting. IoT makes education more
accessible in terms of ability, geography, and socioeconomic
condition. Immersion, for example, is a strategy used to learn
a foreign language based on real-time input. Learning a
language in its home country is simply because the speaker's
feedback is openly available. It is difficult to construct similar
environments outside of the country, but IoT aids in creating
simulation environments in which students can be watched
and teachers can provide real-time feedback to pupils. IoT
networking can facilitate task-based learning. Students learn-
by-doing in doing task-based learning, which allows teachers
to support and monitor a student's performance
automatically. IoT has been a big help to impaired pupils by
giving technical assistance.
Along with IoT, AI and ML have made significant
contributions to advancements in research and education. AI
and ML aid in the automation of time-consuming class
operations such as attendance and grading. Educational
Software can be customized to meet the demands of the
student. IoT and AI aid with e-learning courses by allowing
students to access lessons and videos from any location and
at any time. Discussion groups and forums also allow
students to discuss questions, ideas, and information.
The role of education in propagating knowledge has grown
in importance over the last several years due to the
fulminatory proliferation of knowledge. Meanwhile, the
educational process's paradigm is transforming, requiring
different pupils to complete their learning in a variety of
ways. As a result, a smart educational environment is
promoted. It integrates a variety of information and
communication technologies to energize the learning process
and adapt to the unique needs of individual pupils. The
quality of students' learning processes can be improved by
continuously monitoring and evaluating their states and
actions via information sensing devices and information
processing platforms in order to provide feedback on their
learning processes. The Internet of Things is committed to
achieving significant variation in life, individual well-being,
and organizational productivity. The IoT has the potential to
enable expansions and enhancements to critical utilities in a
variety of industries by creating a unique environment for
application development. Applying the Internet of Things
concept to any educational environment would improve the
quality of education by allowing students to learn more
quickly and teachers to do their duties more efficiently. It is
delicate to attempt to influence and modify user behaviour,
2
since it may face resistance and other psychological issues
connected to comfort and trust. Emerging technologies such
as XAI will aid in the user's comprehension and trust of any
system that makes use of AI models. XAI's goal is to employ
methods and approaches to communicate to the user what an
AI model does and why resulting in a complete understanding
of system operations. As a result, tracking, analyzing, and
influencing user behavior will become significantly simpler
[1] .
However, very few studies focus on how to leverage the
internet of behaviours (IoB) to personalize learning content
and how IoB can be used to monitor students' behaviour that
affects their attainment, progression, and performance. This
work will demonstrate a proposed paradigm for integrating
explainable artificial intelligence (XAI) and the internet of
behaviour (IoB) to tailor learning content to students'
cognitive abilities and automate academic progress tracking.
This article aims to employ a combination of the IoB and XAI
to create trustworthy and intelligible frameworks that operate
in the education domain where users' behaviour changes. It
offers an IoB-based XAI-based system for use in the
education sector to change users' behaviour toward an eco-
friendly one to maximize the learning attenuation and robust
student’s academic progression. This framework
incorporates IoT, AI, Data Analytics, Behavioral Science,
and XAI approaches to deliver smart, adaptable education.
The following summarizes the paper's contributions:
• Propose a trustworthy and understandable IoB-XAI-
based education system capable of influencing and
changing learners' habits and behaviour for their
educational benefit.
• Provide a scenario illustrating the learner's
behaviour inside the class, practicals, and other
related activities to create knowledge about the
importance of controlling a student's progression.
• Compare and contrast the primary characteristics of
existing systems and the proposed paradigm.
• Outline prospective advancements and the planned
system's future directions.
Even though various studies seek to explain how educational
improvements occur, this study examines how explainable
artificial intelligence and the internet of behaviours can be
used to tailor and modify learning resources to students'
cognitive abilities and requirements. This paper is organized
as follows. This paper begins with the abstract and
summarizes the main concern and discoveries of the work.
Second, the introduction summarises the major research gaps
and our main contributions and aims. Thirdly, the literature
review where the context in which the associated work is
presented and elaborated. Fourthly, the research employed a
methodology that includes a detailed description of the
research concept, participants, data collection techniques,
and materials used in the study. Suggestions and
contributions are made in light of the findings. Finally, the
expansion that will be performed in the future.
2. Literature review
The Internet of Things (IoT) is a concept that allows diverse
devices to connect and interact with each other. This can
enhance education, healthcare, transportation, and other
commercial operations [2]. They might be virtual
(information world) or physical (physical world). Objects can
self-identify and integrate with the communication layer.
From a psychological standpoint, the Internet of Behavior is
a subset of the Internet of Things (IoT) that tries to address
users' behaviours [3]. There is a growing body of research on
the Internet of Behavior (IoB) and how it might help
businesses understand their customers better [4]. The IoB
platform helps academic institutions fully comprehend their
pupils. By connecting all phones in the app, users can see
their swing and stroke problems and get visual suggestions
on how to fix them. The connection of gadgets creates new
data points that go beyond the IoT. (IoT). Client data is shared
across linked devices in real-time.
The COVID-19 pandemic has influenced various aspects of
consumer behavior, including brand loyalty, staff
productivity, and customer engagement. The environment,
technology, and human health are all affected. As a result,
tracking people's behavior is crucial in bad situations. Using
machine learning for activities like mask recognition might
help enforce laws and detect negligence [5]. There will be
chances to profit from the information obtained by assessing
the history of patterns in various companies, societal, health,
political, and other realms. A person's behavior might impact
their willingness to cooperate or collaborate. Regardless of
the other four factors: cognition, emotion, personality [6], and
interpersonal communication, behavior is responsible for the
tendency to act. In this way, focusing on behavior might help
us understand how to impact and treat the individual [7]. The
Multi-layered paradigm promotes generating learning
materials based on learners' personalities and learning
patterns [8]. Phone sensors can now assess biometrics and
healthy behavior, allowing businesses to develop
applications that care about users' health. For example, health
applications track sleep, heart rate, and respiration. After
analyzing the user's behavior, such applications provide
alerts, messages, and suggestions to improve sleep and
motivate them to achieve daily objectives [9].
During the COVID-19 epidemic, various nations created
health smartphone applications to help residents stop the
virus's spread. For example, Code-based apps track a user's
travel, contact history, and physical biometrics such as
temperature. Then, a colourful QR code is created to identify
his health state. As a result, various constraints may apply to
the user and influence his behaviour, such as permitting travel
or quarantine at home or a central area [10]. Popular
recommendation systems employ user behaviour, notably
viewing history and clicks, to improve an app or website's
experience. Their technology uses deep neural networks to
evaluate data and recommend films and shows to users. Also,
virtually all social media programs provide feeds depending
on viewing habits, time spent on one account, or interest in
subscribing to certain channels [11]. In transportation, IoB
can assist both drivers and passengers. Uber employs
gamification to influence driver behaviour after a lengthy
history of driver conflicts and high turnover. Uber uses loss
aversion, recognition, and intrinsic motivation to penalize
drivers [12]. The Deep Reinforcement Learning algorithms
(Deep RL) were used to tackle the power consumption based
3
on the user's behaviours [13]. Some systems tried to
understand student’s learning styles to generate
recommendations [14]. IoB plays a critical part in
discovering learners' behaviours and therefore eliminating
any misleading ratings that may occur intentionally while
producing recommendations [15].
3. The proposed model
Learners vary In their capabilities and cognitive styles, which
Imposes the need to personalize learning resources to fit their
abilities and needs. The use of IoT is helpful to collect
different forms of data that could be textbase, images, or
audio data. This study focuses on collecting data about
students and teachers and the learning environment in a
university. IoB has just been identified as one of the top
technological trends. The COVID-19 outbreak is mostly
responsible for IoB becoming a trend since it has transformed
how consumers interact with brands, forcing businesses to
reevaluate how they connect with customers.
3.1 The role of IoB.
The IoB idea seeks to correctly evaluate data and apply that
understanding to construct and provide new things from the
perspective of human psychology. The IoB seeks to
comprehend data gathered from online user activities through
the lens of behavioural psychology. It tries to answer
concerns about how to evaluate data and use that knowledge
to design and deliver new services, all based on human
psychology. This new trend may influence Quality
Infrastructure since many firms may improve their
connectivity, resulting in higher user demand. As represented
by pyramid figure 1, the IoT changes data to information, and
the IoB may transform our knowledge into true wisdom.
Fig. 1. IoB and IoT In Smart layered System
The following demonstrates each layer's main function
1. Observe and track learner behavior: The Internet of
Things (IoT) is used to collect information about
students inside the classroom, their learning
activities, their volunteering activities and
behaviours, how they react to the exercises and
formative assessments, class collaboration work,
hands-on activities, and everything else that is
related to their academic progression.
2. Analyze students' behaviour: the data obtained from
IoT may be utilized to analyse the student's
behaviour appropriately. However, this analysis will
not deliver automatic suggestions through the
system but provide facts about learners' actions
throughout their study and provide some
explanation of some phenomena.
3. Recognize significance: Data acquired by the
Internet of Things has been enriched by behavioral
and psychological analysis (IoT). With the Internet
of Behaviour, companies and organizations
worldwide may better understand and anticipate
how certain behaviors lead to certain results. The
IoB concept allows academic institutions to measure
and predict future student performance. Academic
institutions will be able to respond to students'
behavior by making recommendations and tracking
their progress. The IoB will continue to collect data
on students' choices and patterns of behavior. It's
possible to acquire a far better picture of students'
behavior by analyzing this data than previously
possible.
4. Significance influence: the IoB mechanism can be
used by academic institutions to maintain an
adaptive system that recognizes the behaviours of
students and advises behavioural modifications that
will result in a better outcome at this level.
3.2 Explainable AI and IoB for a smart learning
education system
As the fourth industrial revolution begins, we are seeing a
rapid and broad deployment of artificial intelligence (AI),
contributing to a more algorithmic society. Even with these
extraordinary developments, lack of transparency is a major
barrier to AI-based systems' usage. They can make
tremendous predictions, but they can't be explained. This has
rekindled the debate about explainable AI (XAI) [16]. A
study field might significantly improve AI-based system trust
and transparency. It is required for AI to evolve steadily and
unhindered. Explainable AI helps use AI for data mining
while maintaining transparency. To make AI discoveries
more intelligible to humans, researchers have developed
XAI, a new field of study. Researchers have been studying
the subject of explainable for decades when they looked at
explanations for expert systems [17]. A direct result of
AI/relentless ML's widespread effect on critical decision-
making processes, but without providing explicit information
about the reasons behind specific judgements, proposals,
projections or actions. Increasing social, ethical, and legal
pressures necessitate new AI techniques that can explain
decisions.
4
Technically, there is no accepted definition of explainable AI.
Rather than a specific technology concept, XAI refers to a
movement, initiatives, and actions addressing AI
transparency and trust. Thus, XAI is a new breed of
artificially intelligent companions that humans can
comprehend and control while keeping great learning
performance (prediction accuracy) [18].
3.2.1 Data collection: data can be collected from various
related IoT devices such as wearable watches,
mobiles, cameras, campus cameras, labs monitor
devices, temperature sensors, etc. All forms of
students' activities such as volunteering works,
public speeches, seminars, presentations, project
shows/exhibitions etc., must be under spot to
understand students' behaviours and hence used as
input for analysis and predictions using XAI
models.
Fig. 2. IoB via IoT for a smart learning education system
3.2.2 IoB via XAI in a smart education system
The collected data is then used for applying XAI models,
which aims to
1. Automate the mapping of learner behaviour to
particular symptoms based on gathered data.
2. Establish behavioural patterns for tailoring learning
materials to individual requirements.
3. Justify rationales for specific model outcomes.
4. Provide all the necessary information for managing
patterns to avoid any projected faults or
vulnerabilities influencing model performance.
5. Revealing unanticipated patterns, rules, and reasons
for certain occurrences.
6. Create a smarter model based on a better knowledge
of the rationales for certain results.
Fig. 3. IoB objectives in smart learning education
Automate learner’s behaviour mapping: The main purpose
of IoB in education is to map particular behaviour to a
specific intellectual style. Several studies tried to find the
learning differences between learners [19]. The use of IoB
will automate the process to build smarter education
systems.
Establish behavioural patterns: Understanding learners'
behaviours will help maintain patterns of each behaviour
and better understand learners' perceptions and intellectual
styles.
Justify model outcomes: Several debates have erupted in
recent years about AI/ML-enabled systems producing
biased or discriminating outcomes [20]. That means there
will be a greater need for explanations to guarantee that AI-
based judgments are not made incorrectly. We mean the
need for explanations or justifications for that specific
conclusion rather than describing the inner workings or
logic of reasoning underlying the decision-making process.
Using XAI systems gives the necessary knowledge to justify
results, especially when making unexpected judgments. It
also assures an auditable and demonstrable mechanism to
explain algorithmic conclusions as fair and ethical, leading
to the development of trust.
Control the model: It is crucial not just defending decisions
but to provide the required prevention of errors. Indeed,
learning more about system behaviour offers better insight
into undiscovered vulnerabilities and defects and aids in
speeding and correcting errors in low-criticality scenarios
(debugging) for better model controlling.
Reveal discoveries: The suggested automated model, which
is based on student behaviour, would aid in the discovery of
numerous hidden patterns and rules that will lead to
improved performance.
5
Improve smarter model: Maintaining automated
behavioural patterns through XAI will help build a smarter
system that is upgraded regularly. A model that is easy to
describe and understand can be improved more readily.
Users will be able to make the system smarter since they
understand why it generated specific outcomes. Thus,
IoB/XAI serve as the cornerstone for continual human-
machine iteration and development.
Learning technologies such as Nearpods and BBL
collaboration were utilized to monitor the responses of 41
students (students in programming courses) to activities
such as polls and evaluate if the system supports related
activities. The collaborative work and game-
based/formative evaluations where students can answer
using their mobiles and tablets. Students were also asked to
complete a survey on their volunteer work; the following
table compiles the summary. Table 1 shows students'
capabilities and social keens in their academic performance;
we segment students into four groups, Elite (E), Fair(F),
Below-average (B), and Severely weak(S). We mined more
to understand these students' behaviours. The major goal is
to evaluate the system's ability to understand student
behaviour and personalise learning resources to each
student's strengths, needs, skills, and interests.
Table 1: Learner's capabilities and Social keens
symptoms
The following table illustrates several segments of
learners based on their speaking and writing
talents and their collaboration and volunteering
interests.
Table 2: Learner's segmentation
The following equations summarize the equation used to
measure if the system support
Where,
SP(ES) is the system attractiveness for elite learners;
calculated SP(ES) = 16%
SP(FS) is the system attractiveness for fair learners
calculated SP(FS) = 34%
SP(AS) is the system attractiveness for below-average
learners; calculated SP(AS) = 11%
SP(WS) is the system attractiveness for weak learners;
calculated SP(WS) = 39%
4. Discussions
IoT can help academic institutions gather data on student
reactions to events and understand academic
accomplishment. The learning process will help them
overcome their deficiency. In the first round, we evaluate the
system's ability to engage students in activities without their
6
knowledge and if instructor monitoring via technology
influences their academic achievement. SP(ES) = 16%,
SP(FS) = 34%, SP(AS) = 11%, and SP(WS) = 39%. Given
that the major purpose is to utilize IoB to tailor learning
resources and XAI to explain how the system reacts, the
system should help students by categorizing them and
picking evaluative elements. As part of the second cycle, we
taught students about assessment factors: Speaking
(challenging teachers, seldom asking questions), Writing
(Sentence with elaboration, Short sentences habits).
Volunteering, Collaboration (leading the work, quietness)
(Heavenly involved in activities, Focusing on the given task).
Quiet pupils and those who were focused on the activity were
also given recommendations to improve their elaboration
abilities. SP(ES) increased by 26%, SP(FS) by 38%, SP(AS)
by 19%, and SP(WS) by 17%. This shows that students may
conduct wisely in academic activities if they have a tailored,
regulated framework for learning.
5. The challenges and limitations
The study had several challenges due to many factors: time,
involved students, students selections, and measured
features. The following summarizes a few of these
limitations.
- Time constraints, since we collected data from 41
students over one semester.
- Few features were identified, and we intend to add
more in future study extensions.
- Some students were hesitant to participate in the
research after learning about it.
- The study solely includes students from the
computing division.
6. The fututre directions
The IoT is particularly useful for collecting data about
learners; the IoB is utilized to analyze how pupils react in
response to academic activities. We intend to consider more
factors and incorporate students from other divisions. In
addition, we will create an automated software that will allow
students to understand the elements influencing their
performance and tailor their learning materials. Use a smart
education system that limits students' development from one
level to the next based on their abilities and knowledge
leverage, as determined by automated algorithms that respect
their scraps of skills and knowledge.
7. The conclusion
The study tried to figure out how a smart education system
can make learning materials more personalized for each
student. The proposed method uses IoT and IoB to get
information about students and learn about their behaviour.
Also, using XAI to filter/select features that students can
choose to check their performance and get the help they need
from the education systems is a good thing. Students can
grow in the future by adding more features and making the
implementation broader so that more students from different
divisions can use it.
References
[1] D. Gunning, D.A.-A. magazine, undefined 2019, “DARPA’s
explainable artificial intelligence (XAI) program,” Ojs.Aaai.Org, 2019.
[2] M. Abdel-Basset, G. Manogaran, M. Mohamed, E. Rushdy, “Internet
of things in smart education environment: Supportive framework in the
decision-making process,” Concurrency and Computation: Practice
and Experience, 31(10), 2019, doi:10.1002/cpe.4515.
[3] A. Halgekar, A. Chouhan, I. Khetan, J. Bhatia, N. Shah, K. Srivastava,
“Internet of Behavior (IoB): A Survey,” Ieeexplore.Ieee.Org, 1–6,
2022, doi:10.1109/iscon52037.2021.9702450.
[4] H. Elayan, M. Aloqaily, M. Guizani, “Internet of Behavior (IoB) and
Explainable AI Systems for Influencing IoT Behavior,” Arxiv.Org,
2021.
[5] M. Moghaddam, H. Muccini, … J.D. preprint arXiv, undefined 2022,
“Designing Internet of Behaviors Systems,” Arxiv.Org, Mar. 2022.
[6] O.H. Embarak, “Three Layered Factors Model for Mining Students
Academic Performance,” in 2018 Fifth HCT Information Technology
Trends (ITT), IEEE: 219–226, 2018.
[7] X. Zhou, W. Liang, K.I.K. Wang, H. Wang, L.T. Yang, Q. Jin, “Deep-
Learning-Enhanced Human Activity Recognition for Internet of
Healthcare Things,” IEEE Internet of Things Journal, 7(7), 6429–6438,
2020, doi:10.1109/JIOT.2020.2985082.
[8] O. Embarak, “Multi-Layered machine learning model for mining
learners academic performance,” Advances in Science, Technology
and Engineering Systems, 6(1), 850–861, 2021,
doi:10.25046/aj060194.
[9] I. Sadek, A. Demarasse, M. Mokhtari, “Internet of things for sleep
tracking: wearables vs. nonwearables,” Health and Technology, 10(1),
333–340, 2020, doi:10.1007/s12553-019-00318-3.
[10] I. Sadek, S.U. Rehman, J. Codjo, B. Abdulrazak, “Privacy and Security
of IoT Based Healthcare Systems: Concerns, Solutions, and
Recommendations,” Lecture Notes in Computer Science (Including
Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 11862 LNCS, 3–17, 2019, doi:10.1007/978-3-030-
32785-9_1.
[11] A. Da’u, N. Salim, I. Rabiu, A. Osman, “Recommendation system
exploiting aspect-based opinion mining with deep learning method,”
Information Sciences, 512, 1279–1292, 2020,
doi:10.1016/j.ins.2019.10.038.
[12] N. Scheiber, “How Uber Uses Psychological Tricks to Push Its Driver,”
New York Times, 1, 2017.
[13] S.H. Kim, C. Lee, C.H. Youn, “An accelerated edge cloud system for
energy data stream processing based on adaptive incremental deep
learning scheme,” IEEE Access, 8, 195341–195358, 2020,
doi:10.1109/ACCESS.2020.3033771.
[14] O. Embarak, “A New Paradigm through Machine Learning: A
Learning Maximization Approach for Sustainable Education,”
Procedia Computer Science, 191, 445–450, 2021,
doi:10.1016/j.procs.2021.07.055.
[15] O. Embarak, “Demolish falsy ratings in recommendation systems,” in
2019 Sixth HCT Information Technology Trends (ITT), IEEE: 292–
295, 2019.
[16] O. Embarak, “Explainable Artificial Intelligence for Services
Exchange in Smart Cities,” Explainable Artificial Intelligence for
Smart Cities, 13–30, 2021, doi:10.1201/9781003172772-2.
[17] A. Das, P. Rad, “Opportunities and Challenges in Explainable
Artificial Intelligence (XAI): A Survey,” Arxiv.Org, 2020.
[18] F. Xu, H. Uszkoreit, Y. Du, W. Fan, D. Zhao, J. Zhu, “Explainable AI:
A Brief Survey on History, Research Areas, Approaches and
Challenges,” Lecture Notes in Computer Science (Including Subseries
Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 11839 LNAI, 563–574, 2019, doi:10.1007/978-3-
030-32236-6_51.
[19] O. Embarak, K. Saeed, M. Ali, “Perceiving Intellectual Style to Solve
Privacy Problem in Collaborative Systems,” Lecture Notes on Data
Engineering and Communications Technologies, 29, 401–410, 2019,
doi:10.1007/978-3-030-12839-5_37.
[20] R. Confalonieri, L. Coba, B. Wagner, T.R. Besold, “A historical
perspective of explainable Artificial Intelligence,” Wiley
Interdisciplinary Reviews: Data Mining and Knowledge Discovery,
11(1), 2021, doi:10.1002/widm.1391.