See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/313734208

UTiLearn: A Personalised Ubiquitous Teaching and Learning System for

Smart Societies

Article  in  IEEE Access · February 2017

DOI: 10.1109/ACCESS.2017.2668840

CITATIONS READS
19 178

6 authors, including:

Rashid Mehmood Furqan Alam

King Abdulaziz University King Abdulaziz University
102 PUBLICATIONS   835 CITATIONS    10 PUBLICATIONS   210 CITATIONS   

SEE PROFILE SEE PROFILE

Iyad Katib Aiiad Albeshri

King Abdulaziz University King Abdulaziz University
54 PUBLICATIONS   365 CITATIONS    28 PUBLICATIONS   219 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Internet of things and smart cities workshop call View project

High Performance Computing and Big Data: Technologies and Applications View project

All content following this page was uploaded by Iyad Katib on 13 April 2017.

The user has requested enhancement of the downloaded file.

 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

IEEE ACCESS, MANUSCRIPT ID 1

UTiLearn: A Personalised Ubiquitous

Teaching and Learning System for Smart
Societies

Rashid Mehmood, Senior Member, IEEE, Furqan Alam, Nasser N. Albogami, Iyad Katib, Member,
IEEE, Aiiad Albeshri and Saleh M. Altowaijri

Abstract—The education industry around the globe is undergoing major transformations. Organisations such as Coursera are
advancing new business models for education. A number of major industries have dropped degrees from the job requirements.
While the economics of higher education institutions are under threat in a continuing gloomy global economy, digital and lifelong
learners are increasingly demanding new teaching and learning paradigms from educational institutions. There is an urgent
need to transform teaching and learning landscape in order to drive global economic growth.

The use of distance eTeaching and eLearning (DTL) is on the rise among digital natives alongside our evolution towards smart
societies. However, the DTL systems today lack the necessary sophistication due to several challenges including data analysis
and management, learner-system interactivity, system cognition, resource planning, agility, and scalability. This paper proposes
a personalised Ubiquitous eTeaching & eLearning (UTiLearn) Framework that leverages IoT, big data, supercomputing, and
deep learning to provide enhanced development, management and delivery of teaching and learning in smart society settings. A
proof of concept UTiLearn system has been developed based on the framework. A detailed design, implementation and
evaluation of the UTiLearn system, including its five components, is provided using eleven widely used datasets.

Index Terms— Big data, Computational and artificial intelligence, Distance learning, High performance computing, Internet of
Things

——————————  ——————————

1 INTRODUCTION
course (MOOC ) organisations such as Coursera, edX and

T he education industry around the globe is undergoing

major transformations. A number of universities are
now providing free access to their courses. It all began with
the publication of free video lectures online in 1999 by the
University of Tübingen, Germany, for its Tübingen Inter-
net Multimedia Server (TIMMS) initiative [1], and then the
MIT OpenCourseWare in 2002. Massive open online
————————————————
 R. Mehmood is with High Performance Computing Center, King Abdulaziz course certificates gained online. Maggie Stilwell, manag-
University, Jeddah, KSA. E-mail: RMehmood@kau.edu.sa.
 F. Alam is with Department of Computer Science, Faculty of Computing
and Information Technology (FCIT), King Abdulaziz University, Jeddah,
KSA. E-mail: fmohammed0026@stu.kau.edu.sa.
 N.N. Albogami is with Department of Computer Science, Faculty of Com- fessional qualifications undertaken” [3].
puting and Information Technology (FCIT), King Abdulaziz University,
Jeddah, KSA. E-mail: nalbugami@kau.edu.sa.
 I. Katib is with Department of Computer Science, Faculty of Computing
and Information Technology (FCIT), King Abdulaziz University, Jeddah,
KSA. E-mail: iakatib@kau.edu.sa.
 A. Albeshri is with Department of Computer Science, Faculty of Compu-
ting and Information Technology (FCIT), King Abdulaziz University, Jed- cational institutions. The future of higher education and
dah, KSA. E-mail: aaalbeshri@kau.edu.sa.
 S.M. Altowaijri is with Faculty of Computing and Information Technol-
ogy, Northern Border University, Arar, KSA. E-mail: saltowai-
jri@nbu.edu.sa.
Khan Academy are advancing new business models for
education. PricewaterhouseCoopers, Ernst and Young,
and Penguin Random House has dropped degrees from
the requirements for job applicants [2]. Big companies such
as IBM have started recruiting young workforce with no
bachelor’s degree qualifications. Industry is increasingly
focusing on nurturing specific skills in the workforce as op-
posed to, or in addition to, a university degree. They are
hiring candidates by giving significant weightage to the
ing partner for talent, Ernst and Young, said, “… It found
no evidence to conclude that previous success in higher ed-
ucation correlated with future success in subsequent pro-
While the economics of higher education institutions
(HEIs) are under threat in a continuing gloomy global
economy, digital and lifelong learners are increasingly de-
manding new teaching and learning paradigms from edu-
education in general, is unclear in this unprecedented era
of innovation and competition. There is a clear and urgent
need to transform teaching and learning landscape in or-
der to drive economic growth globally.

2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only.Published
xxxx-xxxx/0x/$xx.00 © 200x IEEE
Personal by
usethe
is also permitted, but republication/redistribution requires IEEE permission. See
IEEE Computer Society
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
2 IEEE ACCESS, MANUSCRIPT ID
Modern day learners demand increasingly sophisti-
cated, dynamic, anytime anywhere learning environ-
ments. While the use of eLearning and Distance learning is
on the rise among the digital natives and millennials, the
DTL1 systems and environments of contemporary times
lack the necessary sophistication. Physical learning spaces
enable interactive environments with a number of im-
portant features that are difficult to mimic in virtual and
distance learning environments. The distance learning en-
vironments of today are unable to connect teacher to the
learner at the same level and quality that physical spaces
could provide. For example, it is difficult to monitor and
manage virtual and distance learning environments due to
the absence of physical presence of the learner and teacher
at the same place. It may be possible to create realistic in-
teractions between the teacher, learner and the learning en-
vironments but these require a systematic approach for
collecting and harnessing data related to these learning en-
vironments and their users. Physical spaces and their in-
teractive environments are organized and conducted by
humans; it is difficult to mimic human intelligence in vir-
tual and distance learning environments. There is a need
to embed cognitive abilities in the DTL systems in order to
serve the needs of the next-generation learners. Also, a
DTL system would typically provide learning content via
videos and virtual reality media, in addition to text-based
and voice-based content. This would require IP and other
networks based communication. Therefore, planning of re-
sources and scalability of a distance learning system is an-
other major challenge given the large-scale data, computa-
tional and networking requirements discussed above. The
agility of the infrastructure is also important. Moreover,
the system should allow accessibility from a broad range
of devices and networks.
Another development that is relevant to the topic of
this paper is around the concept of smart cities because it
relates to the environment of the learner in which they live,
learn, work, and socialise. Smart cities provide the state of
the art approaches for urbanisation, having evolved from
the developments carried out under the umbrella of
knowledge-based economy, and subsequently under the
notion of digital economy and intelligent economy [4] .
Smart cities exploit physical and digital infrastructure, as
well as the intellectual and social capital, for urban and so-
cial development. A city can be defined as ‘‘smart’’ when
“investments in human and social capital and traditional
(transport) and modern (ICT) communication infrastruc-
ture fuel sustainable economic growth and a high quality
of life, with a wise management of natural resources,
through participatory governance” [5]. Komninos [6] de-
fines smart cities as, “territories with high capacity for
learning and innovation, which is built-in the creativity of
their population, their institutions of knowledge creation,
and their digital infrastructure for communication and
knowledge management”. Technically, smart cities are
complex systems of systems that rely on converged and
ubiquitous infrastructures.
1 In this article, we adopt the term “Distance eTeaching and eLearning”
with the abbreviation “DTL” to refer to the broader concepts incorporating
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
The notion of smart cities can be extended to smart so-
ciety; i.e. a digitally-enabled, knowledge-based society,
aware of and working towards social, environmental and
economic sustainability.
We note that knowledge and human and social capital
are at the heart of smart city and smart society develop-
ments. Therefore, we believe that the role of distance learn-
ing would extend beyond the mainstream “education for
employment” scope. It will extend to the notion of social
and collaborative governance where the society will col-
laborate to train each other in maintaining its knowledge,
moral fiber, good practice, resilience, competitiveness, and
for bringing innovation, and becoming a knowledge-based
economy. The key to such efforts would be a converged
ubiquitous infrastructure comprising many systems that
share information among them and collaborate with each
other as necessary.
Motivated by the disruptions and emerging require-
ments of the education industry, and considering the chal-
lenges discussed above in this area, we propose in this pa-
per a personalised UTiLearn (Ubiquitous eTeaching &
eLearning) Framework. UTiLearn is DTL framework de-
signed for enhanced development, management and de-
livery of teaching and learning in smart city and smart so-
ciety settings.
The UTiLearn framework leverages on emerging tech-
nologies to provide an environment and ecosystem for so-
cieties to enrich their knowledge, anytime, anywhere, any
age, through improved processes for collection, develop-
ment, accessibility, pedagogy and cognition of knowledge.
Its users could be teachers, students, governments, busi-
nesses, third parties, and virtual organisations or groups
formed for specific purposes. A learner may register for a
course or programme; or she may register her goals, learn-
ing levels, and timelines with the system and the system
suggests the learner the available courses or programmes,
based on her personal preferences and monitors her
through to the completion of the learning task. In addition
to the usual DTL services that MOOCs typically offer, a
certain group or a government department can develop
specific material to reach out to a group of people to create
awareness of a topic for social intervention purposes, such
as to change people behavior towards better driving prac-
tices or healthy lifestyle.
The UTiLearn framework uses IoT as part of the digital
infrastructure needed to connect with the user for person-
alisation and optimisation of teaching and learning. Mo-
bile computing technologies are used to connect to the us-
ers, sense their continuous activities, and provide them an-
ytime anywhere opportunity for DTL. Big data technolo-
gies are used to deal with the challenges related to the vol-
ume, velocity, and other characteristics of the DTL data.
Computational Intelligence including deep learning is
used for analysis, prediction, personalisation, optimisation
and decision making. HPC technologies are needed to deal
with the high computational requirements of AI and other
algorithms. Cloud technologies are used to deal with the
agility, scalability, and ubiquity of the framework. The
Teaching & Learning (T&L) through ICT-based distance education.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES
UTiLearn Framework could be implemented on top of a
hybrid cloud IaaS (infrastructure as a service) and can offer
ubiquitous teaching and learning as a Service (UTLaaS).
Smart city technologies could be used to benefit from con-
verged social and national infrastructure in order to pro-
vide integrated, enhanced and optimized pedagogies, con-
tent delivery, and better cognition for the users.
Moreover, based on the proposed framework, we de-
velop a system, called the UTiLearn system. A prototype
implementation of the UTiLearn system comprising five
components is described and evaluated in this paper. The
UTiLearn framework and its algorithmic refinement com-
prising five components is discussed in Section 3, and the
subsequent five sections detail the function, the methodol-
ogy, implementation and evaluation of the five compo-
nents. The purpose of the first component, the Sentiment
Analyser, discussed in Section 4, is to analyse social media
for forming the business case of the decisions taken by the
system on strategic, tactical, operational and real time lev-
els. The User Activity Recognition & User Identification
(UARI) Component, detailed in Section , analyses user ac-
tivity data acquired from IoT devices to understand and
predict spatio-temporal activities and other patterns of the
user. The Adaptive Content Delivery Mode Adviser (AC-
DMA), in Section 6, leverages the user activity and identi-
fication related information produced by the UARI com-
ponent to optimize the DTL material delivery, and the sys-
tem and network performance. The System and Network
Provisioning & Demand Prediction (SNPDP) Component
uses the information produced by ACDMA to compute,
predict and manage the system and network resources re-
quired by the UTiLearn system (see Section 7). The Perfor-
mance Analysis, Prediction and Intervention (P2AI) Com-
ponent informs UTiLearn users about their performance
and intervenes as necessary. All five components of the
system are designed to optimise personalisation for indi-
viduals as well as performance for the UTiLearn he system.
We have carried out a detailed literature review of dis-
tance and eLearning systems and reported the challenges
that future DTL systems face today. We used this infor-
mation to develop the UTiLearn framework and a proof of
concept system based on the emerging technologies. To the
best of our knowledge, we have not found any DTL frame-
work or system that addresses the breadth of issues as ad-
dressed by UTiLearn.
Data, Algorithms, Software, and Hardware: We
have a used a total of 11 publically available datasets in-
cluding Twitter data, news media data, seven activity
recognition datasets, telecommunication networks and ap-
plications traffic data, and data from edX [7] (a collabora-
tive effort of Harvard and MIT). We have used the R sta-
tistical platform for modelling, analysis and decision mak-
ing of all the data and processes in our UTiLearn system.
The data analysis, data quality management, prediction
and decision making methods used in the prototype im-
plementation of the UTiLearn framework include Naïve
Bayes classifier, Random Forest, and deep learning artifi-
cial neural networks (DLANNs). The experiments were
carried out on a supercomputer, called Aziz, a Fujitsu
made machine with peak performance of 230 teraflops,
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
3
comprising 11,904 cores in 496 nodes. Aziz was ranked the
world’s top 491 supercomputer in November 2015 Top500
competition (http://www.top500.org/).
Section 2.2.1 gives a background on the technologies
used in this research and reviews the relevant literature.
Sections 3 to 8 introduce the UTiLearn framework and its
proof of concept, including the function, the methodology,
implementation and evaluation of the UTiLearn system.
Section 9 concludes this paper and gives future directions.
2 Background Material and Literature Review
We now give a brief background of the emerging technol-
ogies used in UTiLearn followed by a detailed literature
review in Section 2.2.
2.1 Technology Background
Internet of Things (IoT): The Internet of Things (IoT) refers
to “a global infrastructure for the information society, ena-
bling advanced services by interconnecting (physical and
virtual) things based on existing and evolving interopera-
ble information and communication technologies” [8]. IoT
would make our environments smarter by providing sens-
ing and control capabilities at a fine-grain level. The data
generated by IoT would grow exponentially as the number
of “things” is expected to reach the 50 billion mark by 2020
[9]. The IoT data would allow us to comprehend and man-
age complex environments, make better decisions, increase
automation, and enable higher productivity and efficien-
cies [10]. We will see in this paper how IoT could enhance
DTL theory. and practice.
Big Data: Big data technologies refer to “the emerging
technologies that are designed to extract value from data
having four Vs characteristics; volume, variety, velocity
and veracity” [11]. Big data is revolutionizing most aspects
of our lives, be it work or social interactions. IoT would be
a major contributor of big data. A major challenge is the
management of big data and timely analyses of big data in
order to extract reliable insights and other “value” from
big data. Big data would play an important role in next
generation DTL technologies.
AI and Deep Learning: The use of computational intel-
ligence, including artificial intelligence (AI), has seen tre-
mendous growth in the last decade, and this would con-
tinue to grow at a much faster pace in the coming years
(see e.g. [12]). This is mainly due to the increasing capaci-
ties, and decreasing costs, of computing infrastructure. We
have mainly focused on deep learning, a subfield of AI, in
this paper. Deep learning technologies, also known as deep
structured learning, have made remarkable progress re-
cently and its use has grown in diverse areas including
computer vision, bioinformatics, speech recognition, and
natural language processing. In certain application areas,
deep learning has produced significantly better results
than other AI methods including machine learning [10],
[13], [14].
High Performance Computing (HPC): High Perfor-
mance Computing (HPC) “most generally refers to the
practice of aggregating computing power in a way that de-
livers much higher performance than one could get out of
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
4 IEEE ACCESS, MANUSCRIPT ID
a typical desktop computer or workstation in order to
solve large problems in science, engineering, or business”
[15]. The other terms similar to HPC are supercomputing
and parallel computing. The term distributed computing
typically has a different focus but in future it is expected to
converge with HPC. The use of HPC initially was restricted
to some specialized applications such as in engineering, oil
and gas due to the high cost of HPC machines. However,
HPC is now being applied to many diverse areas including
social media mining. The use of HPC in the education in-
dustry is also expected to grow in the near future.
Mobile computing has already been widely used in
DTL. It allows close interactions with the user, collection of
leaner related preferences and other data, personalization
and improved delivery of learning. Cloud computing al-
lows system agility. The AI technologies and big data
would come together to extract novel insights and make
better decisions from data collected through IoT and other
means in smart environment settings, including the DTL
environments. Big data would affect the future network
traffic and hence the network architectures and this also
needs to be taken into account while designing a DTL sys-
tem. The use of the technologies (discuss above) in the
UTiLearn system would become clearer as we explain our
framework and system.
2.2 Literature Review
Modern day learning and teaching techniques are gone be-
yond the boundaries of conventional classroom teachings
of the schools and universities. ELearning which is con-
ducted via electronic media, specifically with the help of
internet is seen today as a key component to provide better
learning experience to individuals beyond any age bound-
aries and enhancing teaching skills of the tutors. Learning
management systems (LMSs) are the examples of popular
eLearning tools. Typically LMSs are web-based platforms
which provides the functionalities like recording lectures,
course content design, documents download and uploads,
student registration, feedbacks, student development
planning and skill gap analysis [16]. ELearning is becom-
ing popular, however should not be seen as a replacement
of traditional learning. In broader terms, ELearning is the
virtualization of the traditional learning and teaching via
internet with additional attributes like better planning and
management through data analysis [17]. ELearning can be
classified into three sub branches as illustrated in Fig. 1,
which are:
 Semi-classroom: In this form of learning, some part of
teaching is classroom based and remaining is remote
site (home) based.
 Distance Classroom: In this learning mode classroom
is at remote site other than actual university site. Tutor
is present at university site and teaching activates are
facilitate by electronic media using internet.
 Anytime and anywhere: There is no dedicated remote
classroom. Individual can access this lecture anywhere
which is independent of time. Anytime and anywhere
learning is the most flexible form of eLearning.
Cognitive, motivational and environmental factors
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
play critical role in assuring the effectiveness of the eLearn-
ing systems [18]. Cognitive factors describe learner deci-
sion making ability that regulate processes in controlled
learning environment [19]–[21]. Motivation-related factors
include personal goal orientation, personal attribution of
success, and personal attitude toward performing a learn-
ing task [22], [23]. While cognitive and motivation-related
factors have received increasing attention in recent re-
search, few studies have directly examined the environ-
mental context’s influence on learning effectiveness. Envi-
ronmental factors include the structure and content of a
course, as well as learning conditions that may moderate
the effectiveness of learner-controlled instruction [24], [25].
How to trap all these factors is challenging aspect of
eLearning applications and allot depends on these factors
that how learner perceive the systems.
The eLearning market is showing great potential
around the world. It is growing at a slow but steady rate
[26], [27] around the world. The growing acceptance rate
of eLearning solution both by institutions and students is
also attracting more research and developments by indus-
try.
Fig. 1. An eLearning taxonomy
2.2.1 Distance eLearning and Learner Cognition
Designing a DTL system which is efficient is essential for
its acceptance and its success. Moreover, with a carefully
designed system, student can perform better. Leach and
Walker in [28], review some of the DTL models while giv-
ing stress on internet based adult learning. Whereas Mayes
and Freitas in their in-depth review of eLearning models
suggested that cognition can play important role in pro-
ducing desired learning outcomes [29]. The objective of
their work, was to give a detailed account of the presuppo-
sitions about learning that support current eLearning mod-
els in practice. Thus the work tried to achieve mapping of
theoretical contexts onto pedagogical models. Authors en-
deavor to demonstrate how these mappings have been at-
tained from the three broad theoretical contexts: (1) associ-
ationist, (2) cognitive and (3) situative. The associationist
is defined as learning as activity and situative perspective
is defined as learning as social practice [29]. Thinking, per-
ception, learning, language and reasoning are the products
of student's consciousness, ability to remember and form-
ing processes through concepts has been seen as a shift in
psychology in 1960s. The approach is student dependent
rather than population dependent. This added a complete
new dimension to pedagogy which is student centric based
on students' conscious intellectual activity such as remem-
bering, reasoning and thinking termed as cognitive which
means learning as achieving understanding [29].
Further in [30], authors presents an extensive review of
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES
DTL developments based on technology, pedagogy, the-
ory, policy, accessibility, design, environment and man-
agement domains. This work illustrated the recent and
past DTL trends.
2.2.2 Distance eLearning Frameworks
Motivation plays an important in the learning process.
Positive motivation enhances better understanding of top-
ics among the students. MENTOR is one of the such multi-
media based adaptive online DTL framework [31]. It re-
tains positive emotions in students by proving user
friendly, personalised environment for learning which
based student’s emotion, sentiment and personality. An-
other DTL model is proposed in [32] where authors inte-
grate both constructivist and objectivist approaches to in-
structional design. When creating or redesigning courses
for Internet distribution, content and media developers,
faculty and researchers benefit from the identification of a
developmental model taking into account both learning
and design principles.
Bong and Zhang proposed read, reflect, display, and
do (R2D2) DTL framework particularly for online learning
[33]. Online learners have one universal characteristic that
they have diverse learning preferences. The first compo-
nent of R2D2 have methods to assist the learners to gain
knowledge through internet based media. For example
readings, virtual explorations, and online audio lesson and
podcasts. The second component of R2D2 model uses the
reflective activities of the learner such as online blogging
and self-assessment quizzes. The third component is based
on representations of the learning content such as anima-
tions to visually present the content. The fourth and the
last component accentuates what learners can accomplish
with the material in hands-on activities such as real-time
cases and scenarios. Further R2D2 model is explored in
[34]. As the finding suggests that R2D2 model is not effi-
cient in terms of data collection methods. However it
shows positive learning effects with vocabulary pronunci-
ation, acquisition, note taking and presentations by the stu-
dents. In China online learning faced problems initially.
First teachers find hard to adapt and secondly, students
particularly adult students feel over-burdened with allot of
learning content. To address these issues, in paper [35] a
framework is proposed which assist students in interacting
with an automated question answering system. This helps
the teacher to analyze different learning patterns of the stu-
dents. Further teachers are able to organize online learning
contents. Data mining is used to help both teachers and
students to personalize the services. Another DTL frame-
work proposed in [36] which exploits data mining to help
teachers to manage the learning contents on the web using
available infrastructure, needs and experience. Later learn-
ing content can be reorganized according to student's per-
formance.
2.2.3 Social Media and Sentiment Analysis
Social media (SM) has gained a great deal of popularity
during the last decade (see e.g. [37]). Now it is not merely
a platform that is used to do social communication but now
it is more like an online world which is used by everybody
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
5
such as individuals, businesses and governments. Now
businesses are using SM such as Twitter, Facebook, Insta-
gram and blogging to advertise their product and services,
to give special offers as per user preferences. Last year in
American elections we have already seen how SM is used
by people to form an opinion about particular candidate
[38]. A new dimension of SM is now shaping up to form a
completely new segment in its services i.e. distance
eLearning. In the work [39], found out that younger gen-
eration is too enthusiastic for using SM for distance
eLearning. The factors of this increasing enthusiasm is
based on SM eLearning perception, convenience, academic
reasons, easiness and collaboration by social networking
which are same as analysed in [40]. The work [39] also
highlights the interests of students and teacher towards us-
ing SM for eLearning as they believe it will ease and en-
hance communication and collaboration between different
eLearning entities. The results of [39], [40] see similar
trends as shown in earlier work [41] which highlight the
popularity of social networking and its possibilities as an
education tool in Romania. A more recent work [42] which
focuses on SM aspect in actively encouraging cognitive
and meta-cognitive learner development while using a
heutagogical teaching and learning approach. This work
also indicated that familiarity of students with SM is an im-
portant factor that encourages use of SM in eLearning.
From course design perspective the use of SM is increased
in preparing new eLearning courses and packages [43].
These literatures demonstrate that SM can play an im-
portant role in distance eLearning thus must be explored
further with new analyses and approaches.
Sentiment analysis (SA) refers to analyzing the text to
find out key insights about the opinion of the particular
entity with respect to specific topic. This is achieved by
computational linguistics algorithms that extract particu-
lar specific subject related information using social media
(SM) data like tweets, Facebook message, Tag lines, blogs
and online articles. The main goal of SA is to determine the
opinion, attitude and sentiment of a writer regarding spe-
cific topic. Current issues, trends and future prospects of
SA are discussed comprehensively in surveys [44], [45].
Papers also highlighted the importance of public opinion
mining in modern day world and its social and commercial
aspects. For example businesses are using SA to perceive
customer’s product buying choices by their online SM be-
haviors. This is achieved by analyzing tweets, Facebook
posts, blogs, tag-lines etc. In one [46] of such works, SA is
done for political events like elections. Psychometric in-
struments are used to extract six different types of senti-
ments such as anger, depression, confusion, fatigue, ten-
sion, vigor. Whereas in [47], nodes based SA using graphs
has been done that highlights social connection and their
correlation. Multi-language SA is always an issue because
most of the SA models are hard to extend for other lan-
guages. Addressing this problem in [48], which uses sim-
ple heuristics and minimal linguistic computation is per-
formed to make it easily extendible to other languages.
SVM classifier is further used to demonstrate high perfor-
mance of this model. Sentiment analysis is applied to vari-
ous fields to grasp the insights for mass opinions such as
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
6 IEEE ACCESS, MANUSCRIPT ID
finance [49], entertainment industry [50], [51], politics [47],
[52], sales [53] and business operations [54]. The literature
review given above clearly shows that SA and Social media
sentiment analysis (SMSA) are playing critical roles in
planning and policy making for businesses.
2.2.4 IoT and Remote Laboratories
With the emergence of Internet of Things (IoT), DTL can
take the leverage from IoT’s infrastructure and networking
technologies to provide any-time, any-where learning so-
lutions to individuals. Several different perspective of IoT
enabled learning activities are discussed by Adorni et al, in
their work [55]. Undoubtedly in recent years, IoT is ma-
tured and practical applications of IoT are successfully de-
ployed around the world. This is due to the two main rea-
sons: (1) IoT is getting wider acceptance from the users as
well as the industry and (2) hardware manufacturing cost
is coming down with mass productions of sensors, proces-
sors, actuator and other enabling devices. Now affordable
mobile devices are in market and introduction of Internet
Protocol version 6 (IPv6) provides far more bigger address-
ing space so that billions of IoT enabled "Things" can use
it. Both internet availability and use have increased several
folds and it is reaching the remote areas around the world.
The integration of IoT with DTL can be far more beneficial
than what we think today [56], [57]. We will see later in this
paper the use of IoT in our UTiLearn system for personali-
sation, system optimisation, and higher quality of experi-
ence.
An important area where IoT applications and infra-
structure can be used to enhance DL, is by developing re-
mote laboratories. It is often seen that schools and univer-
sities in urban areas have well established and well main-
tained laboratories. However in rural or under developed
areas laboratories are not available or available with lim-
ited and poorly maintained resources. Also, the idea of set-
ting laboratories in remote areas does not sound economi-
cally feasible due to the small number of students per geo-
graphic area. In the past, students who are enrolled in DTL
courses had to go to traditional and physical laboratories
to gain practical knowledge and skills. Currently, thanks
to the advances in networking and communication tech-
nologies, programmers programming languages, new
eLearning tools are available to make the best use of learn-
ing methodologies such as DL. Web-based remote labora-
tory (WBRL) is one of them. WBRL is an application which
allows students to work with real laboratory equipment’s
to perform their experiments. To facilitates this, learner
only need a computer with an Internet connection. How-
ever, emerging technologies and other hardware such as
mobile devices, cloud computing and IoT, require the de-
sign of different models to support innovative learning ex-
periences. For this reason, remote labs are being reconsid-
ered to smart device paradigm. Physically, the considered
smart device is made of the junction of a controller (com-
puter server) connected to physical equipment on the one
hand, and to the Internet on the other hand. In [58] authors
examine four emerging trends in the Internet that have the
potential to influence the development of WBRL: the real-
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
time web and one of them in IoT, geographically distrib-
uted collaboration, and scientific workflows. Each of these
trends is examined and discussed, and a suggested re-
search and development path is offered. Similarly in [59],
author describes the process to translate a “traditional” re-
mote Web lab into a set of smart devices which are able to
work together in the cloud and IoT. Further in [60], authors
have highlighted IoT technologies including actuators,
web cameras, sensors, RFID, and Arduino which are used
to provide WBRL facilities particularly in South African
environment.
2.2.5 Activity Recognition & Behaviour Analysis
Understanding of leaners’ or students’ remote environ-
ment can provide key insights about when and how to de-
liver learning material, lessons and instructions. Further
DTL systems develop learner dependent cognitive. These
cognitive can later be utilized for planning student specific
lessons. One of the such technology is of activity recogni-
tion facilitated by using IoT’s infrastructure [61]. Activity
recognition assists in developing better cognitive about
learner’s behavior and daily routines. Once the under-
standing is achieved, DTL system can deliver lessons in an
adaptive manner rather than fixed schedule to the stu-
dents. For example with the help of activity recognition,
we identified student is free and not busy in any sort of
work and not resting even. This prompt the DTL system to
deliver some important short lesson on his mobile device,
this is only done at the time when student is free. This sort
of learning will be very useful for the people who are
working as well as registered in DTL courses. Other exam-
ple is by activity recognition, DTL application can record
student body language and nervous monuments. Whereas
in the other scenario, where the student solving a online
exams smart watches can be used to analyze his heart beat
which is connected to learning application [62]. Further
heart beat records corresponding to answering time can be
studied to see on which type of question student is nervous
and stressed while answering.
2.2.6 Artificial Intelligence in Distance eLearning
AI gained by machine learning (ML) and Deep learn-
ing (DL) methods, nowadays applied in almost every field
of our environment where we are living today. The use of
ML methods for data mining (DM) is quite old in DTL
field. However, their initial implementations addressed
simple applications in DTL [63]–[65], like web access pat-
tern mining to customize DTL systems as per individual
requirements and On-Line Analytical Processing (OLAP)
of student data. Further in [66], [67], ML classifiers are used
to identify factors which influence the performance of the
students in DL. This assists to upgrade DTL systems for
better learning outcomes. One of the major problem in DTL
courses is higher student dropout rates as compared to tra-
ditional classroom learning which is addresses using vari-
ous classifiers such as Naive Bayesian (NB), k-Nearest
Neighbor (KNN), Decision Tree (DT), Support Vector Ma-
chines (SVM) and Artificial Neural Network (ANN) [68]–
[70]. Literatures above are purely based on providing data
mining functionalities based on student historic records.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES 7

However, modern day DTL systems requirements are far 3 UTILEARN: PROPOSED FRAMEWORK
more beyond data mining. One issue with in past DTL sys-
We introduce in this section our UTiLearn framework, give
tems is the absence of auto guide which can instruct teach-
its overview, and present its algorithmic refinement.
ers to point out problems and possible solutions. An artifi-
UTiLearn is a distance eTeaching and eLearning (DTL)
cial advisor is used in to solve this problem [71] which uti-
framework designed for enhanced development, manage-
lizes fuzzy logic.
ment and delivery of teaching and learning in smart city
It is known that around the world students face hard-
and smart society settings. The framework is depicted in
ship in learning mathematical concepts and find it difficult
Fig. 2 and is further elaborated in Fig. 3. Moreover, we
to use them for problem solving. One of such system which
have developed a system, called the UTiLearn system,
addresses this by using AI power is ARTIMAT [72]. It is
based on the proposed framework. A prototype imple-
based on the idea to enhance mathematical problem solv-
mentation of the UTiLearn system is described and evalu-
ing abilities of the students. Further in [73], ARTIMAT is
ated in the next five sections (Sections 4 to 8).
evaluated and finding suggest that, it ease mathematical
problem solving procedure for the students and under-
standability of the mathematical concepts among the stu-
dents. Present day DTL systems are far more capable, user
friendly and automate. We believe that integrating new
technologies like IoT and AI can be useful for DTL systems.

Fig. 2. The proposed UTiLearn framework.

2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

8 IEEE ACCESS, MANUSCRIPT ID

Fig. 3. The Proposed UTiLearn framework.

The UTiLearn framework leverages on advanced and
emerging technologies (big data, HPC, AI, IoT, mobile
computing, and cloud) to provide an environment and
ecosystem for societies to enrich their knowledge, anytime,
anywhere, any age, through improved processes for collec-
tion, development, accessibility, pedagogy and cognition
of knowledge. Its users could be teachers, students, gov-
ernments, businesses, third parties, and virtual organisa-
tions or groups formed for specific purposes. The roles are
not exclusive; for example, an individual or organization
could be both teacher and learner based on her expertise,
contributions, and requirements. A learner might need to
learn a specific skill so she registers its interests, goals,
learning levels, and timelines with the system. The system
then suggest the learner the available courses or programs
and monitors her through to the completion of the learning
task. The system would maintain a profile for each user
over time and use the profile to understand the user pref-
erences and other cognitive abilities. The individuals could
also contribute teaching material in the areas of their ex-
pertise. A rating system could be maintained for individu-
als and their contributed material based on user satisfac-
tion. Certain groups or a government department can de-
velop specific material to reach out to a group of people or
the whole population to create awareness of a topic for so-
cial intervention purposes. An example is the change of
people behavior towards better driving practices, cycling,
or healthy lifestyles. Third parties could also provide their
services in bridging the service gaps that the framework is
not able to provide directly.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
The UTiLearn framework uses IoT as part of the dig-
ital infrastructure needed to connect with the user for op-
timisation of teaching and learning. The IoT infrastructure
could include smart phone sensors, smart watch sensors,
virtual reality technologies to enable physical-like experi-
ences in distance eLearning environments, virtual labs,
and other sensors. Mobile computing technologies are
used to connect to the users, sense their continuous activi-
ties, and provide them anytime anywhere opportunity for
DTL. Big data technologies are used to deal with the chal-
lenges related to the Vs of DTL data and benefit from the
opportunities these technologies offer. Computational In-
telligence and AI technologies are used for analysis, pre-
diction, optimisation and decision making. HPC technolo-
gies are needed to deal with high computational require-
ments of AI and other algorithms. A continuing trend is to
move businesses in the cloud environments using infra-
structure as a service (IaaS) or other models. The UTiLearn
system could be implemented on top of hybrid IaaS and
can be offered distance teaching and learning as a Service
(UTLaaS). Smart city technologies could be used to benefit
from converged infrastructure for integrated, enhanced
and optimized pedagogies, content delivery, and cognition
of the users. The UTiLearn system hence would contribute
to the development of smart society; a digitally-enabled,
knowledge-based society, aware of and working towards
social, environmental and economic sustainability.
The framework (and the system) comprises five main
components:
1. Sentiment Analyser
2. User Activity Recognition & User Identification
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES 9

(UARI) Component text) of content delivery to suit the user preferences and
3. Adaptive Content Delivery Mode Adviser (ACDMA) cognitive abilities can be advised by ACDMA. The AC-
4. System and Network Provisioning & Demand Predic- DMA Component could restrict and enable access to sys-
tion (SNPDP) Component tem resources based on the user identification and activi-
5. Performance Analysis, Prediction and Intervention ties. Moreover, ACDMA computes and predicts the num-
(P2AI) Component ber of applications and clients that are needed at different
times and geographical locations.
Algorithm
Input:
UTiLearn main
D ⃪ Tweets and online news articles
The System and Network Provisioning & Demand Pre-
𝐴 ⃪ 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐷𝐵 diction (SNPDP) Component uses the information pro-
Output: 𝐷𝐶𝐶 𝑎𝑛𝑑 𝑃𝑂𝑅 duced by ACDMA to compute and predict the system (e.g.
1. Go to Sentiment Analyser
2. Get D core-hours) and network (e.g. bandwidth) resources re-
3. Clean content of D quired by the UTiLearn system. The Performance Analy-
4. Perform SA on D
5. If Sentiments == positive sis, Prediction and Intervention (P2AI) Component stores
6.
7.
Then
Motivation exists.
and analyses information related to the interaction of the
8. Develop/Update the DTL system user with the UTiLearn system continuously in the back-
9. Move to line 12
10. Else ground to inform UTiLearn users about their performance.
11.
12.
Terminate / continue to monitor
Recognize activities and identify users
Moreover, it suggests interventions to the users to help
13. Go to UARI them address specific shortcomings and improve their out-
14. Apply DLANN to predict activities and
identify users comes.
15.
16.
Update 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐷𝐵
Prepare course content corresponding to
All five components of the system are designed to op-
identified activity timise personalisation for individuals as well as perfor-
17. Go to ACDMA
18. Look in to Activity-to-Content table to identify mance for the UTiLearn he system.
which course content matches predicted activity
19.
20.
Network Provisioning & Demand Prediction
Go to SNPDP
3.1 UTiLearn: An Algorithmic Refinement
21. Analyzing students temporal activity patterns The framework presented in Fig. 3 has been refined into
from historic activity data from 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐷𝐵
22. Apply DLANN for Temporal activity pattern the algorithm presented in Fig. 4. We explain the algorithm
23. Calculate and predict the short-term and
long-term network requirements
as follows. The Main UTiLearn Algorithm takes Tweets,
(bandwidth, delay, etc.) online news articles and activity data as input and feeds to
24. Feed into the system for real-time, planning
and provisioning purposes the Sentiment Analyser component which performs senti-
25.
26.
If course content match == available bandwidth
Then
ment analysis in order to transform the social media data
27. DCC to student into opinions of users, businesses and governments. The
28. Record student and system interaction
29. Else UARI Component then uses DLANNs for classifying user
30. Secure additional bandwidth if possible activities and identifying users. The ACDMA Component
31. If successful
32. Go to line 21 suggests the course content corresponding to user activity
Else
33.
34. Look for available content quality which type and computes the number of clients and applications
suites available bandwidth in content quality table to be served by the system. The SNPDP Component com-
35. Record student and system interaction
36. Predict Students performance with respect to his putes and predicts the UTiLearn system and network re-
37.
interaction with system
Go to P2AI
sources. The P2AI component periodically, and as needed,
38. Apply DLANN to predict performance predicts user performance based on their interaction and
39. Relate performance of the student with the mode of
delivery, individually and for all the students. the interaction of other users with the UTiLearn system. It
40. Send performance report and predicted final
course outcome. Intervene and give suggestion for improvements
warns and intervenes users as necessary. The reader may
also find Fig. 2 and Fig. 3 useful in understanding the main
Fig. 4. The UTiLearn main algorithm. algorithm. Moreover, the main algorithm will become
clearer as we indulge into explaining each component of
Fig. 2 lists these five components and summarises their the UTiLearn system, one by one, in the upcoming sec-
purpose within the UTiLearn framework. The framework tions.
is further elaborated in Fig. 3 with additional details in a
workflow style. The purpose of the Sentiment Analyser
Component is to analyse social media for forming the busi- 4 THE SENTIMENT ANALYSER
ness case of the decisions taken by the system. The senti- Considering the increasing importance and benefits of so-
ments for business case can be explored on strategic, tacti- cial media sentiment analysis in all spheres of our life, as
cal, operational and real-time levels. The User Activity highlighted in the literature review (see Section 2.2.1 ), we
Recognition & User Identification (UARI) Component have added a sentiment analysis function in our UTiLearn
analyses user activity data acquired from smart phones, system, called the Sentiment Analyser Component. The
smart watches and other IoT units to understand and pre- task of this component is to analyse social media for form-
dict spatio-temporal activities and other patterns of the ing the business case for the decisions taken by the system.
user. The Adaptive Content Delivery Mode Adviser (AC- The sentiments for business case can be explored on mul-
DMA) uses the user activity and identification related in- tiple levels, at strategic, tactical, operational and real-time
formation produced by the UARI component to optimize levels. For example, monitoring and analyzing sentiments
the DTL material delivery, and the system and network to understand the need for UTiLearn system presence in a
performance. For example, a better mode (video, voice,
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
10
particular region, or the need for adding a particular
course, programme, or topic in the UTiLearn teaching
portfolio.
Algorithm Sentiment Analyser
Input: D ⃪ Tweets and news articles
X ⃪ {Positive, Negative, Neutral}
Output S ⃪ Sentiments
1. Prepare D
2. Filter 𝐷 for special character, white spaces and symbols
3. Apply Naïve Bayes (NB) classier Є P(X|Y) = 𝑃(𝑌|𝑋)𝑃(𝑋)/𝑃(𝑌)
4. Where 𝑃(𝑋|𝑌) is P(𝑋) class | Y ⃪ 𝐷(1,2,3,4….𝑛)
5. 𝑃(𝑌|𝑋) is P(Y) | X the class
6. P(X) and P(Y) independent of all other P
7. S ⃪ Aggregate(Predicted X)
8. Communicate S to Next Modules
Fig. 5. The Sentiment Analyser algorithm.
Take another example; suppose we wish to introduce
an UTiLearn system in Saudi Arabia. It is wise to make a
business case before such as decision is made. The social
media in Saudi Arabia, as well as the relevant international
social media can be analysed and questions can be asked,
such as: do we need distance eLearning in Saudi Arabia, a
particular region of the country, city, or rural areas? Is
there a particular course, topic or programme the teaching
thereof is highly desirable? Are there any particular inter-
ventions the government desires in the society that can be
achieved by using such a system? This data may come
from the government or private organization looking to
create awareness or knowledge for a particular subject or
for a particular intervention. Moreover, the data can be an-
alyzed and monitored for a near real time application such
as to deliver learning for a particular topic for a particular
user (a decision to personalise learning based on user pref-
erences). Consider another example; a user is learning a
course on data mining in the UTiLearn system. The system
suggests her to learn another course on data mining based
on her preferences, her goals, and some relevant changes
in her local and global context. This advice from the
UTiLearn system may even be for a user to learn a partic-
ular subject replacing the one she is currently taking, based
on the changes in job market.
In the rest of the section, we provide the methodology
used for the design and implementation of the Sentiment
Analyser Component. An algorithm is provided for the
component.
4.1 Methodology
It has been mentioned in Section 1 that we have used the R
statistical platform for modelling, analysis and decision
making of all the data and processes in our UTiLearn sys-
tem which includes the five components that we would
discuss from here on. We have gathered data from Twitter
and a total of 4221 tweets are downloaded for SA. We have
also used 50 published online news articles of well-known
newspapers and magazines including Guardian, Fortune,
BBC, Wired, and Times.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
IEEE ACCESS, MANUSCRIPT ID
Sentiments
100
80
Number (%)
60
40
20
0
Positive Negative Neutral
Opinion
Tweets Online News
Fig. 6. Sentiments of Twitteraties and online news media for distance
eLearning
For collecting the twitter and newspaper data, we used
search keywords including ‘Online distance learning +
eLearning’, ‘eLearning investment + finance’, ‘Distance +
eLearning drawbacks’, ‘Flexible learning’, ‘eLearning
growth’, ‘eLearning accreditation’, ‘eLearning + career +
jobs’ and ‘motivation + eLearning’. For sentiment analysis
of the tweets, we first downloaded data based on the above
mentioned keywords into the R software tool by using
twitteR package [74]. Data access is provided with the
combination of consumer key, consumer secret, access to-
ken and access secret code generated with the help of Twit-
ter API. Similarly for online news media, we used Google
to search for news articles related to the above mentioned
keywords. We then downloaded the data and imported
into R using RCurl package [75]. The RCurl package filters
out the useful text data from the html pages.
We have used the Naïve Bayes algorithm for the pur-
pose of sentiment analysis. We classify sentiments into
three categories, namely: positive, negative and neutral.
The Sentiment Analyser algorithm is given in Fig. 5 and,
given the explanation above, is self-explanatory.
4.2 Results
The results obtained from the Sentiment Analyser Compo-
nent show highly positive opinion about distance learning
among individuals, businesses and governments. The re-
sults include opinion of following subjects:
 Individuals’ opinions as to how they rate distance
eLearning and its future prospects for gaining
knowledge, building career and finding jobs.
 Businesses’ opinions as to how they see distance
eLearning from revenue perspective.
 Government opinion regarding the value of accredita-
tion of distance eLearning.
The results are visualized in Fig. 6 for the Twitter data
and the news article data. The SA results from the twitter
data show that 78% Twitteraties have positive sentiments
for distance and eLearning. The data from online newspa-
per and magazine articles show an 89% positive senti-
ments towards distance eLearning. These results show that
individuals are interested in taking benefits from distance
eLearning. The businesses are also interested in exploiting
this surge in popularity among adults and youth by
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES 11

providing new DTL systems with advanced learning fea-
tures. The approach presented in this section can be used
to collect and analyse real time data regarding specific
course or programs, certain pedagogies, or modes of deliv-
ery to make informed decisions about the users and sys-
tems. The data can be analysed on spatio-temporal basis to
understand the emerging behavior of people across times,
cities, countries and regions.
Table 1
Benchmark datasets for activity recognition
No Dataset Dataset Details
Name
1 Ugulino [76] The dataset is recorded using
accelerometers for 8 hours by 4
participants.
2 Casale [77] The dataset is recorded using
accelerometers by 15 partici-
pants.
3 Reiss [78] The dataset is recorded using
inertial measurement units and
a heart rate monitor by 9 partic-
ipants.
4 Stisen [79] The dataset is recorded using
smart watch sensors (accel-
erometers and gyroscope) and
smart phone sensors (accel-
erometers and gyroscope) by 9
participants.
2. To identify students based on the spatio-temporal ac-
tivity data.
3. To use the obtained temporal activity patterns for
computing and predicting the number and type of cli-
ents and applications.
4. To compute and predict the bandwidth, latency and
other characteristics of the networks required by the
UTiLearn system by using the data obtained in Point
(3) above.
5. To compute and predict computational and other sys-
tem characteristics required by the UTiLearn system
by using the data obtained in Point (3) above.
Activities Year
5 activity classes: standing-up, 2012 Table 2
standing, walking, sitting-down,
and sitting.
HAR datasets and their respective input parameters for DLANN
7 activity classes: working at 2012
computer, standing, standing up,
No Dataset Epochs Two Layers
walking going up-down stairs,
walking and going up-down 1 Ugulino [76] 800 400 Neurons each
stairs, talking while standing,
walking and talking with some- 2 Casale [77] 800 400 Neurons each
one.
3 Reiss [78] 60 200 Neurons each
18 activity classes: cycling, run- 2012
ning, walking, Nordic walking, 4 Stisen [79] 120 60 Neurons each
standing, sitting, lying, watching
TV, working on computer, driv-
Smart Watch accelerometer
ing car, climbing stairs, coming 5 Stisen [79] 80 60 Neurons each
down the stairs, vacuum clean-
ing, house cleaning, ironing, fold- Smart Watch gyroscope
ing laundry, rope jumping, play-
ing soccer, rope jumping and 6 Stisen [79] 60 180 Neurons each
other (transient activities).
Smart Phone accelerometer
6 activity classes: standing, sit- 2015
ting, walking, biking, stair-up
7 Stisen [79] 100 50 Neurons each
and Stair-down. Smart Phone gyroscope

5 Anguita [80] The dataset is recorded using
smartphones by 30 partici-
pants.
6 Roggen [81] The dataset is recorded using
ambient and on-body sensors
by 4 participants.
7 Banos [82] The dataset is recorded using
accelerometers and gyroscope.
8 Reyes-Ortiz The dataset is recorded using
[83] smartphone sensors such as ac-
celerometers and gyroscope by
30 participants.
6 activity classes: walking Up-
stairs, walking downstairs, walk-
ing, standing, sitting and laying.
4 activity classes: walk, stand, sit
and lie.
33 sports related activities
6 activity classes: walking, walk-
ing upstairs, walking downstairs,
standing, sitting, lying
2013 It is possible to compute the application characteristics
that would be needed to support the various teaching ma-
2010
terial via different modes of content delivery over time.
This in turn would allow planning the networks that are
needed to support the system generated traffic. The net-
2014
work planning, both long-term and short-term, could be
2016 done by analyzing the historical and real-time activities
and other data.

5.1 Methodology
A number for human activity recognition datasets are
5 ACTIVITY RECOGNITION & USER IDENTIFICATION available publicly [76]–[83]. Table 1 lists the most widely
used eight human activity recognition datasets. The da-
(UARI) COMPONENT tasets have not been given a specific name by their origina-
Integration of Human Activity Recognition (HAR) and IoT tors except the PAM (Reiss [78]) and Opportunity (Roggen
has gained significant importance in recent times, particu- [81]) datasets. For presentation and brevity reasons, we
larly in context-aware systems. Accurate and reliable HAR have named each dataset according to its first author’s sur-
has been used in several applications such as improving name (see Column 2). The third column gives brief details
quality-of-care for infants, elderly and people with poor of each dataset and Column 4 lists the name of the activity
health conditions [61]. It has been found helpful in opti- classes. Column 5 lists the year of publication of the respec-
mizing energy consumption of IoT based smart homes tive datasets. The accuracy of the location reported in the
[84]. It can also help to predict various human behaviors datasets can be found based on the devices used in the ex-
such as rash driving styles [85] and emergency situations periments.
in advance like aggressive actions which can lead to vio- We have used Random Forest (RF) generated feature
lence [86] using smartphone based sensors. In the current importance (relevance) to select a subset of attributes from
prototype implementation of the UTiLearn system, we the datasets based on their relevance [87]. We then fed da-
have used human activity recognition for the following taset to H2O’s deep learning artificial neural networks
goals: (DLANNs) algorithm [88], which is based on a multi-layer
1. To provide adaptive educational content delivery to feed forward ANNs that are trained using back-propaga-
the students who are registered for a course. tion (BP) with stochastic-gradient descent. We divided da-
taset into two parts, 60% for the training phase and 40% for
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

12 IEEE ACCESS, MANUSCRIPT ID

the testing phase. We have used a different number of iter-
ations (for different datasets) with two hidden layers each
containing a certain number of neurons. The number of it-
erations (epochs) and the number of neurons used can be
found in Columns 3 and 4 of Table 2, respectively.
The Algorithm for the Activity Recognition & User
Identification Component of the UTiLearn system as de-
scribed above is given in Fig. 7. Lines 1 and 2 in the algo-
rithm set response variables to the activity and user class.
The algorithm is deliberately written descriptively and is
self-explanatory.
Algo- Activity Recognition & User Identification (ARUI)
rithm
Input: 𝑀 ⃪ 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐷𝐵
readable. The y-axis in the figure gives the list of activities
that have been identified from the datasets using our algo-
rithm. The x-axis gives the time dimensions of the human
activity (the time is relative, not absolute). The space di-
mension is represented by box sizes (z space dimension),
box shape outline colours (x space dimension), and shape
fill colours (y space dimension). The largest box represents
the highest Z value of space dimension; red colour in the
box outline represents the largest X value of the space di-
mension; and the light blue colour represents the highest Y
value of the space dimension. The figure shows, for exam-
ple, that the sitting activity was only carried out between 5
minutes to 12.5 minutes, and the biking activity was rec-
orded between 22 to 33 minutes. The time for other activi-

𝑆𝑡𝑢𝑑𝑒𝑛𝑡𝑠𝑝𝑎𝑡𝑖𝑎𝑙 ⃪ 𝑆𝑒𝑛𝑠𝑒𝑑 𝑠𝑡𝑢𝑑𝑒𝑛𝑡′𝑠 𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠

Output: 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑐𝑙𝑎𝑠𝑠 ⃪ Predicted activity class
1. Set response variable → activity class (For human activity recog-
nition)
2. Set response variable → user name (For User identification)
3. Impute missing values using Multiple Imputation by Chained
Equations (MICE)
4. If Data matrix 𝑀 contain NAs
5. Then
𝑥1 𝑥2 𝑥3
Let 𝑀 = 𝑥4 𝑁𝐴 𝑥6
6. 𝑥7 𝑥8 𝑁𝐴
Imputation function = 𝐼𝑚𝑝𝑢𝑡𝑒𝑀𝐼𝐶𝐸 (𝑀) ties can be found similarly.
𝑥1 𝑥2 𝑥3
𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑 = 𝑥4 𝑥5 𝑥6
7. 𝑥7 𝑥8 𝑥9 Fig. 8. Spatio-temporal human activity patterns
8. Go to line 11
9. Else
Fig. 9 gives the accuracy of the human activity
recognition DLANNs algorithm. The results depicted are
10. Go to line 11
the mean square error (MSE), the root mean square error
11. 𝑓1, 𝑓2, 𝑓3 , 𝑓4 … . 𝑓𝑛 ⃪ feature space of 𝑀 or 𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑
(RMSE), and the percent classification accuracy. The figure
12. Compute weight 𝑊(𝑥1 ,𝑥2,….𝑥𝑛) using Random Forest shows that these three metrics are dependent on the
feature selection specific datasets. Morever, we note that fairly low errors
13. Select subset of 𝑊𝑚𝑎𝑥 , | new data matrix 𝑀𝑠𝑢𝑏𝑠𝑒𝑡 = 𝑊𝑚𝑎𝑥 (M) and high accuracies can be achieved using the DLANN
14. Set parameters of DLANN algorithm. However, these high accuracies, while
15. Divide 𝑀 or 𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑 into 𝑀𝑡𝑟𝑎𝑖𝑛 ⃪ 60% 𝑜𝑓 𝑛𝑟𝑜𝑤(𝑀)
dependent on the quality of datasets, are computationaly
expensive to obtain (see [10], where a comparison of eight
and 𝑀𝑡𝑒𝑠𝑡 ⃪ 𝑀 − 𝑀𝑡𝑟𝑎𝑖𝑛
data mining algorithms, including DLANNs, is given in
16. Training and testing using DLANN
terms of accuracy, computational complexity and
17. Parameter tuning if required, go to line 14
applicability for IoT data).
18. Feed 𝑆𝑡𝑢𝑑𝑒𝑛𝑡𝑠𝑝𝑎𝑡𝑖𝑎𝑙 to DLANN
19. Get 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑐𝑙𝑎𝑠𝑠
20. Update 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐷𝐵 with{ 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑐𝑙𝑎𝑠𝑠 , 𝑆𝑡𝑢𝑑𝑒𝑛𝑡𝑠𝑝𝑎𝑡𝑖𝑎𝑙 }

Fig. 7. The Algorithm for the Activity Recognition & User Identifica-
tion (ARUI) Component of the UTiLearn system.

5.2 Results
We used the datasets given in Table 1 and the algorithm
given in Fig. 7 to calculate spatio-temporal activities of the
users. These are depicted in Fig. 8 for the Stisen smart
phone accelerometer dataset [79]. The dataset contains ap-
proximately one million records. However, in the figure,
we have used 5% sample data from the dataset to make it
Fig.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. 9. Visualization
Personal of thebuthuman
use is also permitted, activity recognition
republication/redistribution results.
requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES
6 ADAPTIVE CONTENT DELIVERY MODE ADVISER
(ACDMA)
The Activity Recognition & User Identification Compo-
nent described in the previous section is able to recognize
human activities and identify them based on their activi-
ties. The purpose of the Adaptive Content Delivery Mode
Adviser (ACDMA) that we will discuss in this section is to
use the user activity and automatic identification infor-
mation to optimize the DTL material delivery, and the sys-
tem and network performance. We have already men-
tioned the potential uses of the Activity Recognition &
User Identification Component as a list at the beginning of
Section on . The T&L material delivery can optimized, for
example, for suggesting a better mode (video, voice, text)
of content delivery to suit the preferences, learning charac-
teristics, pedagogical requirements, and cognitive abilities
of the users. The system performance can also be opti-
mized by using the user activity and identification data for
system planning or system performance prediction; for
motivation see [89], where we have studied system perfor-
mance for healthcare systems in computational Grid envi-
ronments. The user activity and identification data can also
be used for network planning, location based services, pre-
serving user privacy and security, including user authen-
tication and authorization [90].
Algorithm Adaptive Content Delivery Mode Adviser (AC-DMA)
Input: user_activity, user_preferences, other_factors
Output: DCC and receive acknowledgement
1. Relate course content delivery media to
2. Activity Type
3. Application type
4. Network type
5. Device type
6. DCC appropriately select content for student
7. Received delivery acknowledgement from student
8. Update course content database
9. Compute the appropriate mix of applications and devices
Figure 10. Algorithm for Adaptive Content Delivery Mode Adviser
(ACDMA).
Our earlier work has looked into creating geographic
firewalls based on user location data allowing us to restrict
and enhance services for individual users within a specific
geographic area such as a lecture or examination room
[91]. The control of distant physical or virtual environ-
ments, and the users within them, is typically a difficult
task given the remote nature of distance eLearning envi-
ronments. This work has been extended to policy based
control of pervasive mobile learning environments using
detection and resolution of feature interaction [92]. The ap-
proaches and works mentioned above could alleviate these
problems.
To explain the function of ACDMA further, consider
Fig. 8 which plots human activities across time and space.
The course material is provided to the user based on its
preferences and other contextual information such as the
user current and future activity, light or heavy study
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
13
mode, availability of learning material and mode of deliv-
ery, availability of network bandwidth and latency, etc.
For example, due to the safety reasons, it may not be safe
to deliver any course content to a user who is biking, driv-
ing or climbing stairs; it may be ok, arguably, to deliver
voice-based course material to this user. The user, how-
ever, can override anytime the system suggested mode of
delivery and teaching material subject to legal and proce-
dural constraint. The algorithm for ACDMA is given in Fig-
ure 10 .
We now look at a potential scenario that we have de-
veloped as a prototype implementation of the Adaptive
Content Delivery Mode Adviser Component. The scenario
is motivated by our earlier work on a healthcare system
[11]. Fig. 11 depicts a potential scenario where UTiLearn
subsystems in four countries are connected with each other
through IP communications (this is for demonstration pur-
poses and we would expect the UTiLearn system to have
global presence). Fig. 12 gives a more detailed networked
system architecture by narrowing down the system be-
tween two cities, Munich in Germany and Jeddah in Saudi
Arabia. The infrastructure does not need to be owned by
the UTiLearn system, it could be acquired as an infrastruc-
ture or platform as a service.
Table 3 gives the number of client applications distrib-
uted across three types of distance eLearning places (see
Fig. 12: remote classrooms, homes, and mobile devices) be-
tween Munich and Jeddah. We have kept the number of
devices and eLearning places symmetric across the cities in
accordance with the original article [11] in order to train
our model with the networking related data and validate
the networking results. Further details of networking
training and validation are provided in Section 7.
7 SYSTEM AND NETWORK PROVISIONING &
DEMAND PREDICTION (SNPDP) COMPONENT
By now, we have described and evaluated three of the five
components of the UTiLearn system. The first component
of the UTiLearn prototype system, the Sentiment Analyser,
provided us the results about the opinions of individuals,
governments and businesses on distance eLearning. This
component could advise on making decisions regarding
specific courses, pedagogies, or modes of content delivery.
The second component, Activity Recognition & User Iden-
tification, provided us the information on the users, their
locations and activities. This in turn enabled us, using the
third UTiLearn Component, Adaptive Content Delivery
Mode Adviser (see the previous section) to compute and
predict the number of applications, clients that are needed
across time and space (see Fig. 13 and Table 3). In this sec-
tion, we describe the System and Network Provisioning &
Demand Prediction Component and present its results.
The SNPDP Component uses the (applications and clients)
results obtained from the ACDMA Component to compute
and predict the bandwidth, latency and other characteris-
tics of the networks required by the UTiLearn system.
Moreover, it is also able to use the results from the AC-
DMA Component to compute and predict computational
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

14 IEEE ACCESS, MANUSCRIPT ID

and other system characteristics (e.g. how much computa-
tional power and memory would be needed) required by
the UTiLearn system. It could even be possible to compute
and predict energy requirement of the UTiLearn system
based on the computational system and network load.
7.1 Methodology
Our methodology to design and implement the SNPDP
Component is to use the application traffic (see Table 3) and
network characteristics (Fig. 12) from the ACDMA Com-
ponent to compute and predict the future network and sys-
tem requirements of the UTiLearn system. This could be
done on long-term, short-term and near real time levels.
We have used DLANN in this paper. However, we wish to
compare and augment it with other computational intelli-
gence methods such as Markov chains. Some of our earlier
work on modelling (including large Markov models) of
communication networks and other applications could be
found in [93]–[98].

Fig. 11. UTiLearn System with networked communications between four countries.

Fig. 12. UTiLearn Prototype Network Architecture for two cities.

2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES 15

Algorithm System and Network Provisioning & Demand Prediction (SNPDP) vided the data into 60/40: 60% of the data is used for train-
Input: 𝑁𝑐 ⃪ {𝑑, 𝑁𝑏 , 𝑁𝑡𝑦𝑝𝑒 } ing and 40% for testing.
𝐴𝑝𝑝𝑐 ⃪ {𝐴𝑝𝑝𝑛 , 𝐷𝑞 } The data used to train the networks in this prototype
Output 𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑡𝑟𝑎𝑓𝑓𝑖𝑐 ⃪ {𝑇𝑟 , 𝑇𝑠, 𝑇𝑙 } implementation of the SNPDP component is taken from
1. Measure 𝑁𝑐 [11] (see also Table 3 and Fig. 13). A detailed network mod-
2. Aggregate 𝐴𝑝𝑝𝑐 elling related work has been carried out in [11] to explore
3. Feed to DLANNs ⃪ {𝑁𝑐 , 𝐴𝑝𝑝𝑐 } QoS performance specifically for intracity, intercity, and
4. Set DLANNs P
global environments for healthcare systems. They have
5. Train and Test DLANNs
used a variety of applications with different network and
6. Tune P
system sizes. Future work will focus on gathering more
data, both in terms of quantity and diversity, for network
7. If (P== best results)
training purposes. The SNPDP algorithm provides the
8. Then
Network planning functionality; the function to calculate
9. M ⃪ 𝐷𝐿𝐴𝑁𝑁𝑠
and predict computational characteristics of the UTiLearn
10. Else
system could be implemented in a similar manner except
11. Tune P Further
that the attributes used in the function would be system
12. Go to Line 4
related such as computational time, memory etc (see [89]
13. 𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑡𝑟𝑎𝑓𝑓𝑖𝑐 ⃪ 𝑀(𝑁𝑐 , 𝐴𝑝𝑝𝑐 ) for motivation). The load balancing of UTiLearn servers
14. Secure bandwidth for 𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑡𝑟𝑎𝑓𝑓𝑖𝑐 would also be required to create high availability, we plan
to rely on our earlier work in [99] for this purpose.
Fig. 13. System and Network Provisioning & Demand Prediction
(SNPDP).
Video Performance Comparison: IntraCity
Table 3
4
Client applications and distance eLearning places
3.5
3
Munich, Germany Jeddah, Saudi Arabia
Delay (ms)

2.5
2
1.5
Place Client Applications Place Client Applications 1
0.5
FTP DB Video FTP DB Video 0
M96
L96

H96

VH192

VH384

VH768

M1248
H1248
L1248
UH192

UH384

UH768
VH96

M192

M384

M768
L192

H192

L384

H384

L768

H768
UH96

VH1248
UH1248
Clients Clients Clients Clients Clients Clients
78 78 52 78 78 52
Remote Remote

Classroom Classroom
Actual Predicted

78 78 52 78 78 52
Home Home
Fig. 14. IntraCity Video Delay for Jeddah and Munich (symmetric):
Actual Data vs Predicted by DLANN.
78 78 52 78 78 52
Anywhere Anywhere

234 234 156 234 234 156

Applications Video Performance Comparison: InterCity
Total
114
624 624 112
City Total
110
108
Delay (ms)

106
Total 1248 Clients
104
102
100
98
The SNPDP Algorithm of the UTiLearn system for ap- 96
94
plications bandwidth prediction and network provision- 92

ing is given in Fig. 13. The algorithm uses 𝑁𝑐 for the net-
M96
L96

H96

VH192

VH384

VH768

M1248
H1248
UH192

UH384

UH768
L1248
VH96

M192

M384

M768
L192

H192

L384

H384

L768

H768
UH96

VH1248
UH1248

work characteristics and 𝐴𝑝𝑝𝑐 for the application charac-

teristics. The network characteristics include 𝑑 for delay, Actual Predicted
𝑁𝑏 for bandwidth, 𝑁𝑡𝑦𝑝𝑒 for network type. The application
characteristics include 𝐷𝑞 for the quality/size of video Fig. 15. InterCity Video Delay between Jeddah and Munich: Actual
streams, voice streams, database queries, FTP workloads, Data vs Predicted by DLANN.
Files, 𝐴𝑝𝑝𝑛 for the number of applications such as database
queries, file upload and download, and videos. DLANNs 7.2 Results
parameters are represented by P including /DLAANs to
The results obtained from the DLANN modelling (The
acquire M (model) to 𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑡𝑟𝑎𝑓𝑓𝑖𝑐 (predict traffic). The
SNPDP Algorithm in Fig. 13) are depicted in Fig. 14 and
data used is a synthesized data containing 100,000 records
Fig. 15. We have predicted a large number of scenarios for
produced using the network related data from [11]. We di-
network traffic including database and FTP traffic but for
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
16
brevity reasons we will present the video results alone. The
predicted results using the DLANN models are compared
with actual response times taken from [11]. Fig. 14 plots the
video delay values encountered by the clients for 25 differ-
ent traffic profiles in intracity environment. The term in-
tracity means that all the clients and servers are within the
same city; Jeddah or Munich. Both the cities will have same
performance because the architecture and scenarios are
same. The traffic profile L96 in the figure means low net-
work traffic for 96 nodes and UH1248 profile means ultra
high traffic with 1248 nodes. The details of all these traffic
profiles, the bitrates etc. can be found in [11]. Similarly, Fig.
15 plots the video delay values for intercity traffic between
Jeddah and Munich. Note that the difference between the
actual and predicted values is mostly minimal. Fig. 16 de-
picts the root mean square error (RMSE) between the pre-
dicted and actual values for video. This RMSE value is
computed for all the network traffic including video, FTP
and database queries. Note that the prediction error is very
small.
Fig. 16. RSME shows the differences between predicted and actual
values obtained by DLANNs model.
8 THE PERFORMANCE ANALYSIS, PREDICTION
AND INTERVENTION (P2AI) COMPONENT
The Performance Analysis, Prediction and Intervention
(P2AI) Component is the fifth and last component of the
UTiLearn system. Its need and design is motivated by the
fact that the users (students, teachers and others) would
like to know whether the objectives they had set for their
teaching and learning are being met, the extent of their suc-
cess or failure, and the detailed reasons behind their
achievements and shortcomings. The P2AI Component
stores and analyses the information continuously in the
background to inform UTiLearn users about their perfor-
mance. The historical and current information about a
user’s activities is analysed continuously to inform the user
about their current performance. The information related
to the individual users, as well as the collective infor-
mation, is also used to predict course outcomes and other
user outputs. For example, a student can be informed
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
IEEE ACCESS, MANUSCRIPT ID
about her worsening performance due to a pattern in her
course related activities and inactivity. A general perfor-
mance pattern can be identified for the user population
based on their activities and inactivity and this can be used
to anticipate, predicted and inform individuals and groups
about their performance. Moreover, the P2AI Component
can suggest interventions (strategies, actions, courses,
modes of delivery, pedagogy related suggestions) to the
user to help her address specific shortcomings and im-
prove her outcomes. Such awareness and capabilities of
the UTiLearn system about its users would allow users to
get real-time, short-term and long-term detailed feedback
and take necessary actions to improve their performance.
The system will report performance to the users by differ-
ent means including email, SMS, multimedia message and
browser pop-ups. These functionalities of the P2AI Com-
ponent are captured in the P2AI process cycle, depicted in
Fig. 17.
Fig. 17. Performance Analysis, Prediction and Intervention (P2AI)
Component process cycle.
8.1 Methodology
Fig. 18 gives the performance prediction and intervention
algorithm for a prototype implementation of the P2AI
Component. The core of the P2AI component relies on the
H2O DLANNs algorithm [88], which is based on a multi-
layer feed forward ANNs that are trained using back-prop-
agation (BP) with stochastic-gradient descent. We divided
dataset into two parts, 60% for training and 40% for testing.
We performed 400 iterations with two hidden layers each
containing 800 neurons.
For testing the P2AI Component, we have used a real
dataset [7] available online by edX published in 2014. The
major feature of this data [7] is that, it describes degree of
interaction between the student and edX. The edX online
education entity is a massive open online course (MOOC)
provider, founded in year 2012 by the collaborative efforts
of Harvard and MIT. It offers high quality online courses
that are delivered and designed by some of the best minds
around the world.
The edX data that we used contained a large number
of missing values. We therefore discarded rows which had
more than four missing columns; i.e. the columns which
were related to our analysis, containing students’ event-
based records describing student and system interaction
(Data Cleansing: Fig. 17). Further we imputed missing val-
ues using MICE [100]. Subsequently, we selected a subset
of attributes which are most relevant by using RF gener-
ated feature importance [87]. A subset of these selected fea-
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES 17

tures, only those that describe student and system interac- population interacted with the edX. The student
tion, are listed in Table 4. Using the selected features data, population is separated into two groups: the students who
we performed predictions based on the “certified” attrib- gained upto 50% marks (depicted by the yellow color bars)
ute (not mentioned in the table) with class label of 1 for and those who gained between 51% to 100% marks
certified (successful) students and 0 for unsuccessful stu- (depicted by the blue color bars). The average number of
dents. times the student population interacted with edX are
Table 4 plotted in four different categories: Average videos played,
Features that describes student and system interaction
average chapters read, average activities, and average
events (see Fig. 19. A clear message from the figure is that
No Feature Description
the students who received higher marks (blue bars)
1 explored Student who accessed minimum the half chapter of
interacted with the system significantly higher number of
a course
times (3239 vs 1051) than those who received lower marks
2 last_event_DI Last date of interaction between student and DTL
(yellow bars). We could evidently imply that a student’s
3 nevents Total interactions of the student with DTL system
marks are directly proportional to the number of times the
for the particular course
student interacts (uses the course content) with the system.
4 ndays_act Total days student interacted with the DTL system
Fig. 20. The classification error plot of the P2AI algorithm.Fig. 20
5 nplay_video Total videos played by student for the particular
depicts the accuracy of the DLANN prediction algorithm
course
(see Fig. 18) in terms of the classification error. It shows a
6 nchapters Total chapters from which student interacted
very small error (the y-axis scale is very small). The algo-
rithm achieved a 99.83% classification accuracy, a 0.0018
Algorithm P2AI: Performance Prediction and Intervention
mean square error (MSE), and a 0.0427 root mean square
Input: 𝑀 ⃪ 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝐷𝐵 (historic data of student who finished
error (RMSE). These are relatively high accuracies
the course) achieved by the use of DLANNs algorithm though, as
𝐶 ⃪ 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝐷𝐵 (on going student) mentioned earlier, it incurs high computational costs (see
𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑟𝑒𝑝𝑜𝑟𝑡 ⃪ 𝑆𝑡𝑢𝑑𝑒𝑛𝑡 ′ 𝑠 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑤𝑖𝑡ℎ 𝐷𝐿𝑆 [10]).
Output: Predicted 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑜𝑢𝑡𝑐𝑜𝑚𝑒
1. Set response variable → Certified class name
2. Impute missing values using Multiple Imputation by Chained Equations
(MICE)
3. If Data matrix 𝑀 contain NAs
4. Then
𝑥1 𝑥2 𝑥3
Let 𝑀 = 𝑥4 𝑁𝐴 𝑥6
5. 𝑥7 𝑥8 𝑁𝐴
Imputation function = 𝐼𝑚𝑝𝑢𝑡𝑒𝑀𝐼𝐶𝐸 (𝑀)
𝑥1 𝑥2 𝑥3
𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑 = 𝑥4 𝑥5 𝑥6
6. 𝑥7 𝑥8 𝑥9
7. Go to line 10
8. Else
9. Go to line 10
10. 𝑓1 , 𝑓2 , 𝑓3 , 𝑓4 … . 𝑓𝑛 ⃪ feature space of 𝑀 or 𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑
11. Compute weight 𝑊(𝑥1,𝑥2 ,𝑥3….𝑥𝑛) using Random Forest feature selection Fig. 19. Students' scores against different type and number of inter-
12. Select subset of 𝑊𝑚𝑎𝑥 , | new data matrix 𝑀𝑠𝑢𝑏𝑠𝑒𝑡 = 𝑊𝑚𝑎𝑥 (M) actions with UTiLearn.
13. Set parameters of DLANN
14. Divide 𝑀 or 𝑀𝑖𝑚𝑝𝑢𝑡𝑒𝑑 into 𝑀𝑡𝑟𝑎𝑖𝑛 ⃪ 60% 𝑜𝑓 𝑛𝑟𝑜𝑤(𝑀) and 𝑀𝑡𝑒𝑠𝑡 ⃪ 𝑀 −
𝑀𝑡𝑟𝑎𝑖𝑛
15. Training and testing using DLANN
16. Parameter tuning if required, go to line 13
17. Feed 𝐶 to DLANN
18. Get 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑜𝑢𝑡𝑐𝑜𝑚𝑒
19. Send 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑜𝑢𝑡𝑐𝑜𝑚𝑒 report to students from time to time
20. Warn and intervene student by Email, Multimedia, text messages
and web browser pop-ups
21. Suggest possible way to improve 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑜𝑢𝑡𝑐𝑜𝑚𝑒

Fig. 18. The performance prediction and intervention algorithm for a

prototype implementation of the P2AI Component.

8.2 Results
The results related to the student performance obtained
through the P2AI Component prototype are depicted in
Fig. 19. The figure plots the number of times the student
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
18
Fig. 20. The classification error plot of the P2AI algorithm.
9 DISCUSSION, CONCLUSION AND OUTLOOK
The education industry around the globe is undergoing
major transformations. MOOC organisations such as
Coursera Khan Academy are advancing new business
models for education. Several major industries have
dropped university degrees from the requirements for job
applicants. Industry is increasingly focusing on nurturing
specific skills in the workforce as opposed to, or in addition
to, a university degree. The economics of higher education
institutions are under threat in a continuing gloomy global
economy, while digital and lifelong learners are increas-
ingly demanding new T&L paradigms. There is a clear and
urgent need to transform teaching and learning landscape
in order to drive global economic growth. The use of dis-
tance eTeaching and eLearning (DTL) is on the rise among
digital natives alongside our evolution towards smart so-
cieties. However, the DTL systems today lack the neces-
sary sophistication due to several challenges including
data analysis and management, learner-system interactiv-
ity, system cognition, resource planning, agility, and scala-
bility.
We also discussed the developments around future cit-
ies and smart societies and highlighted the fundamental
need of developing knowledge, and human and social cap-
ital for their enablement. It was discussed that smart soci-
ety developments would extend the role of learning from
mainstream “education for employment” scope to the no-
tion of social and collaborative governance where the soci-
ety will collaborate to train each other in maintaining its
knowledge, moral fiber, good practice, resilience, compet-
itiveness, and for bringing innovation, and becoming a
knowledge-based economy. The key to such efforts would
be a converged ubiquitous infrastructure comprising
many systems that share information among them and col-
laborate with each other as necessary.
We felt motivated by the challenges and opportunities
surrounding the education industry and digital learners,
as well as the developments in urbanisation, and pro-
posed, in this paper, a Ubiquitous eTeaching & eLearning
(UTiLearn) Framework. UTiLearn leveraged IoT, big data,
supercomputing, deep learning, mobile computing and
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
IEEE ACCESS, MANUSCRIPT ID
cloud computing to provide enhanced development, man-
agement and delivery of teaching and learning in future
city and smart society settings. UTiLearn framework pro-
vides an environment and ecosystem for societies to enrich
their knowledge, anytime, anywhere, any age, through im-
proved processes for collection, development, accessibil-
ity, pedagogy and cognition of knowledge.
Its users could be teachers, students, governments
(such as for cloud-based healthcare systems [11], [101]),
businesses, third parties, and virtual organisations or
groups formed for specific purposes. A learner may regis-
ter her preferences and the system advises her through to
the completion of the learning task. In addition to the usual
functions of an MOOC, a government department can de-
velop specific material to reach out to a group of people to
create awareness of a topic for social intervention pur-
poses. The UTiLearn framework can run in a cloud and can
be offered as a service, UTLaaS.
A proof of concept implementation of the UTiLearn
framework, called the UTiLearn system, comprising five
components was developed in the paper. The five compo-
nents were described in details along with their functions,
methodologies, algorithms, implementation, as well as
their evaluation. We used 11 widely used datasets for sys-
tem evaluation. The R statistical platform was used for the
data analysis, data quality management, prediction and
decision making. The methods used included Naïve Bayes
classifier, Random Forest, and DLANNs. The experiments
were carried out on the supercomputer Aziz.
Future work will focus on building additional capacity
of the UTiLearn system in terms of the quality and func-
tionality of the processes used in the various components
of the system, data quality management, extension of big
data related functionalities, and IoT related functionalities.
A major issue in designing a real life UTiLearn system
would be its data and information management. We could
benefit by the research and practice on designing sustain-
able enterprise information systems. Our earlier work in
this respect [102], [103] would form the basis of our future
direction on extending information management aspects
of the UTiLearn system. Another major issue in a scalable
design of the UTiLearn system would be to deal with big
data and computational load related issues; Our research
direction in this respect will rely on modelling, analysis
and design methods that leverage HPC technologies and
are able to deal with big data, such as our earlier work in
[104], [105]. The work on the UARI component in this pa-
per has revealed that the human activity datasets available
online are limited in their quality and extent. We believe
that in the future many more detailed and high quality da-
tasets will be available to allow a better understanding of
human activities across space and time. Moreover, systems
like UTiLearn can record the activity data of the users and
buildup its own datasets over time to have a better under-
standing of human activities to allow better optimisation
of pedagogy, human cognition, delivery mode selections,
and required application traffic and network characteris-
tics. Additional work is also needed in finding and incor-
porating networking, applications and system workload
related datasets to be used in the SNPDP Component.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES
These are required to enhance the precision of the training,
computation and prediction of networking characteristics,
application traffic models, and system models. In this re-
spect, our earlier work on city transportation modelling,
evacuation, and telematics, such as in [106], [107], [108],
[109], would be useful both from activity recognition and
learning point of view. That is, information, learning and
entertainment would be integrated, and system optimiza-
tion would benefit from all kinds of (big) data.
Finally, the use of computational and artificial
intelligence and human-machine interaction demands a
careful look at the design of UTiLearn and similar systems.
There is need to consider the theory and practice of
sociotechnical systems [110]–[113].
ACKNOWLEDGEMENT
The authors would like to acknowledge the technical and
financial support from the Deanship of Scientific Research [17]
(DSR) at the King Abdulaziz University (KAU), Jeddah,
Saudi Arabia, under the grant number 1437/611/715. The
experiments reported in this paper were performed on the
Aziz supercomputer at KAU. We are also thankful to UCI [18]
data repository for making datasets available.
The corresponding author for this paper, Rashid
Mehmood, could be contacted via email at
RMehmood@kau.edu.sa.
REFERENCES
[1] “OpenCourseWare,” Wikipedia. [Online].
https://en.wikipedia.org/wiki/OpenCourseWare.
[2] L. Burke, “University degrees irrelevant to big employers,”
new.com.au, 2016.
[3] L. Sherriff, “Ernst & Young Removes Degree Classification
From Entry Criteria As There’s ‘No Evidence’ University
Equals Success,” The Huffington Post UK, 18-Jan-2016.
[4] L. A. Tawalbeh, A. Basalamah, R. Mehmood, and H.
Tawalbeh, “Greener and Smarter Phones for Future Cities:
Characterizing the Impact of GPS Signal Strength on Power
Consumption,” IEEE Access, vol. 4, no. c, pp. 858–868, 2016.
[5] A. Caragliu, C. Del Bo, and P. Nijkamp, “Smart Cities in
Europe,” 3rd Cent. Eur. Conf. Reg. Sci., pp. 45–59, 2009.
[6] N. Komninos, “Intelligent cities: Variable geometries of
spatial intelligence,” Intell. Build. Int., vol. 3, no. 3, pp. 172–
188, 2011.
[7] A. D. Ho, J. Reich, S. O. Nesterko, D. T. Seaton, T. Mullaney,
J. Waldo, and I. Chuang, “HarvardX and MITx: The First
Year of Open Online Courses, Fall 2012-Summer 2013,”
SSRN Electron. J., no. 1, pp. 1–33, 2014.
[8] ITU, “Recommendation ITU-T Y.2060 (06/2012),” 2012.
[9] D. Evans, “The Internet of Things - How the Next Evolution
of the Internet is Changing Everything,” CISCO white Pap.,
no. April, pp. 1–11, 2011.
[10] F. Alam, R. Mehmood, I. Katib, and A. Albeshri, “Analysis
of Eight Data Mining Algorithms for Smarter Internet of
Things ( IoT ),” Int. Work. Data Min. IoT Syst. (DaMIS 2016),
vol. 98, no. DaMIS, pp. 437–442, 2016.
[11] R. Mehmood, M. A. Faisal, and S. Altowaijri, “Future
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
19
Networked Healthcare Systems: A Review and Case Study,”
Big Data Concepts, Methodol. Tools, Appl., pp. 2429–2457, 2016.
[12] A. Chianese, F. Marulli, F. Piccialli, P. Benedusi, and J. E.
Jung, “An associative engines based approach supporting
collaborative analytics in the Internet of cultural things,”
Futur. Gener. Comput. Syst., vol. 66, pp. 187–198, 2017.
[13] I. Arel, D. C. Rose, and T. P. Karnowski, “Deep Machine
Learning - A New Frontier in Artificial Intelligence Research
[Research Frontier],” IEEE Comput. Intell. Mag., vol. 5, no. 4,
pp. 13–18, 2010.
[14] X. Du, Y. Cai, and S. Wang, “Overview of deep learning,”
Youth Acad. Annu. Conf. Chinese Assoc. Autom., 2016.
[15] “What is high performance computing?,” Inside HPC.
[Online]. Available: http://insidehpc.com/hpc-basic-
training/what-is-hpc/.
[16] Ryann K. Ellis, “A Field Guide to Learning management
systems,” 2009.
B. Khatri, P. Chouskey, and M. Singh, “Comparative analysis
study of e-learning and traditional learning in technical
institution,” Proc. - 2013 Int. Conf. Commun. Syst. Netw.
Technol. CSNT 2013, pp. 770–773, 2013.
C. Sorgenfrei and S. Smolnik, “The Effectiveness of E-
Learning Systems : A Review of the Empirical Literature on
Learner Control,” Decis. Sci. J. Innov. Educ., vol. 14, no. 2,
2016.
[19] P. H. Winne, “A metacogntive view of individual differences
in self-regulated learning,” Learn. Individ. Differ., vol. 8, no. 4,
pp. 327–353, 1996.
[20] B. J. Zimmerman, “Investigating Self-Regulation and
Available:
Motivation: Historical Background, Methodological
Developments, and Future Prospects,” Am. Educ. Res. J., vol.
45, no. 1, pp. 166–183, 2008.
[21] E. Danili and N. Reid, “Cognitive factors that can potentially
affect pupils’ test performance,” Chem. Educ. Res. Pract., vol.
7, no. 2, p. 64, 2006.
[22] M. S. Rosli, N. S. Saleh, B. Aris, M. H. Ahmad, A. A. Sejzi,
and N. A. Shamsudin, “E-Learning and Social Media
Motivation Factor Model,” Int. Educ. Stud., vol. 9, no. 1, p. 20,
2015.
[23] S. Essam and J. Al-Ammary, “The Impact of Motivation and
Social Interaction on the E-Learning at Arab Open
University, Kingdom of Bahrain,” Creat. Educ., vol. 4, no. 10,
pp. 21–28, 2013.
[24] B. Cheng, M. Wang, J. Moormann, B. A. Olaniran, and N. S.
Chen, “The effects of organizational learning environment
factors on e-learning acceptance,” Comput. Educ., vol. 58, no.
3, pp. 885–899, 2012.
[25] Y. Park and K. Lim, “Effects of Environmental and Human
Constructs on e – learning Effectiveness in Online University
Settinings,” Indian J. Sci. Technol., vol. 8, no. January, pp. 103–
109, 2015.
[26] “E-Learning market trends and forecast: 2014 - 2016 Report
by Docebo,” 2014.
[27] I. E. Allen and J. Seaman, “Tracking Online Education in the
United States,” 2015.
[28] K. Leach and S. Walker, “Internet-Based Distance
Education : Barriers , Models , and New Research,” Proc.
WebNet World Conf. WWW Internet, 2000.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
20
[29] S. Mayes, T. and de Freitas, “JISC e-Learning Models Desk
Study Stage 2 : Review of e-learning theories , frameworks
and models,” Learning, vol. 202, no. September, p. 43, 2004.
[30] M. Aydemir, E. E. Özkeskin, and A. A. Akkurt, “A
Theoretical Framework on Open and Distance Learning,” [49]
Procedia - Soc. Behav. Sci., vol. 174, no. 2003, pp. 1750–1757,
2015.
[31] M. Leontidis, C. Halatsis, and M. Grigoriadou, “Using an
affective multimedia learning framework for distance
learning to motivate the learner effectively,” Int. J. Learn. [51]
Technol., vol. 6, no. 3, pp. 223–250, 2011.
[32] K. Passerini and M. J. Granger, “A developmental model for
distance learning using the Internet,” Comput. Educ., vol. 34,
no. 1, pp. 1–15, 2000.
[33] C. J. Bonk, K. Zhang, C. J. Bonk, and K. Zhang, “Introducing [52]
the R2D2 Model: Online learning for the diverse learners of
this world Introducing the R2D2 Model,” Distance Educ., vol.
27, no. 2, pp. 249–264, 2006.
[34] H. Cartner and J. Hallas, “Exploring the R2D2 model for [53]
online learning activities to teach academic language skills,”
Proc. ASCILITE Auckl., 2009.
[35] R. Shen, P. Han, F. Yang, Q. Yang, and J. Z. Huang, “An open
framework for smart and personalized distance learning,” [54]
Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif.
Intell. Lect. Notes Bioinformatics), vol. 2436, pp. 19–30, 2002.
[36] S. A. Khoja, F. Sana, A. Karim, and A. A. Rehman,
“Implementing constructivist pedagogical model in
dynamic distance learning framework,” Commun. Comput.
Inf. Sci., vol. 20 CCIS, pp. 191–201, 2008.
[37] F. Piccialli and J. E. Jung, “Understanding Customer
Experience Diffusion on Social Networking Services by Big
Data Analytics,” Mob. Networks Appl., pp. 1–8, Dec. 2016.
[38] F. Manjoo, “Social Media’s Globe-Shaking Power,” The New
York Times, 16-Nov-2016.
[39] V. Balakrishnan and H.-S. Loo, “Social Media in E-Learning:
An Empirical Analysis among Students and Academicians,”
GSTF J. Comput., vol. 2, no. 4, 2013.
[40] X. Li, K. Ganeshan, and G. Xu, “The role of Social
Networking Sites in e-learning,” 2012 Front. Educ. Conf. Proc., [58]
pp. 1–6, 2012.
[41] A. Stanciu, F. Mihai, and O. Aleca, “Social Networking As an
Alternative Environment for Education,” Account. Manag. [59]
Inf. Syst., vol. 11, no. 1, pp. 56–75, 2012.
[42] L. M. Blaschke, “Using social media to engage and develop
the online learner in self-determined learning,” Res. Learn.
Technol., vol. 22, no. 0, Nov. 2014.
[43] C. S. Fishpaw and M. Ketel, “Online social networking
practices and the implications for e-learning solutions,” Conf.
Proc. - IEEE SOUTHEASTCON, 2014.
[44] B. Liu and L. Zhang, “A Survey of Opinion Mining and
Sentiment Analysis,” Min. Text Data, pp. 415–463, 2013.
[45] M. Tsytsarau and T. Palpanas, “Survey on mining subjective [62]
data on the web,” Data Min. Knowl. Discov., vol. 24, no. 3, pp.
478–514, 2012.
[46] J. Bollen and A. Pepe, “Modeling Public Mood and Emotion :
Twitter Sentiment and Socio-Economic Phenomena,” Fifth
Int. AAAI Conf. Weblogs Soc. Media, pp. 450–453, 2011.
[47] P. Fornacciari, M. Mordonini, and M. Tomauiolo, “Social
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
IEEE ACCESS, MANUSCRIPT ID
Network and Sentiment Analysis on Twitter : Towards a
Combined Approach,” 2011.
[48] G. Paltoglou, “Sentiment Analysis in Social Media,” Online
Collect. Action, no. June, pp. 3–17, 2014.
P. D. Azar, “Sentiment Analysis in Financial News,” vol.
Bachelor o, p. 68, 2009.
[50] S. Asur and B. A. Huberman, “Predicting the future with
social media,” Proc. Int. Conf. Web Intell. Intell. Agent Technol.,
pp. 492–499, 2010.
V. K. Singh, R. Piryani, A. Uddin, and P. Waila, “Sentiment
analysis of movie reviews: A new feature-based heuristic for
aspect-level sentiment classification,” 2013 Int. Mutli-
Conference Autom. Comput. Commun. Control Compress. Sens.,
2013.
A. Ceron, L. Curini, and S. M. Iacus, “Using Sentiment
Analysis to Monitor Electoral Campaigns: Method Matters--
Evidence From the United States and Italy,” Soc. Sci. Comput.
Rev., vol. 33, no. 1, pp. 3–20, 2014.
L. C. Wood, T. Reiners, and H. S. Srivastava, “Expanding
sales and operations planning using sentiment analysis:
demand and sales clarity from social media,” ANZAM 2013,
Stream 15 Technol. Innov. Supply Chain Manag., pp. 1–17, 2013.
M. Zitnik, “Using sentiment analysis to improve business
operations,” XRDS Crossroads, ACM Mag. Students, vol. 18,
no. 4, p. 42, 2012.
[55] G. Adorni, M. Coccoli, and I. Torre, “Semantic web and
internet of things supporting enhanced learning,” J. E-
Learning Knowl. Soc., vol. 8, no. 2, pp. 23–32, 2012.
[56] I. Asseo, M. Johnson, B. Nilsson, C. Neti, and T. Costello,
“The Internet of Things: Riding the Wave in Higher
Education,” EDUCAUSE Review, 2016. [Online]. Available:
http://er.educause.edu/articles/2016/6/the-internet-of-
things-riding-the-wave-in-higher-education.
[57] D. Kurzweil and S. Baker, “The Internet of Things for
Educators and Learners,” EDUCAUSE Review, 2016.
[Online]. Available:
https://er.educause.edu/articles/2016/8/the-internet-of-
things-for-educators-and-learners.
M. F. Schulz and P. Long, “Possible Futures for Remote
Laboratories,” Internet Access. Remote Lab. Scalable E-Learning
Tools Eng. Sci. Discip., pp. 493–510, 2011.
M. L. García, G. C. Fernandez, E. S. Ruiz, A. P. Martín, and
M. C. Gil, “Rethinking remote laboratories: Widgets and
smart devices,” Proc. - Front. Educ. Conf. FIE, pp. 782–788,
2013.
[60] N. Dlodlo, “The Internet-of-things in remote-controlled
laboratories,” Proc. 13th Annu. Conf. World Wide Web Appl.,
no. September, pp. 14–16, 2011.
[61] A. Wilde and E. Zaluska, “Revisiting activity theory within
the Internet of Things. ,” 1st Int. Work. Internet Sci. Web Sci.
Synerg., pp. 3–6, 2013.
Ericsson, “Wearable technology AND Internet.” [Online].
Available:
https://www.ericsson.com/res/docs/2016/consumerlab/
wearable-technology-and-the-internet-of-things-ericsson-
consumerlab-2016.pdf.
[63] S. H. Ha, S. M. Bae, and S. C. Park, “Web mining for distance
education,” Proc. 2000 IEEE Int. Conf. Manag. Innov. Technol.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access
MEHMOOD ET AL.: UTILEARN: A PERSONALISED UBIQUITOUS TEACHING AND LEARNING SYSTEM FOR SMART SOCIETIES
, vol. 2, pp. 715–719, 2000. [80]
[64] W. Hämäläinen, J. Suhonen, S. Member, A. Hershkovitz, R.
Nachmias, E. Sutinen, and H. Toivonen, “Data mining in
personalizing distance education courses,” Proc. 1st Int. Conf.
Educ. Data Min., p. 16, 2004. [81]
[65] L. Wang, “Data mining in individuality long-distance
education system applied research,” 2nd Int. Symp. Inf. Sci.
Eng. ISISE 2009, pp. 99–103, 2010.
[66] N. K. M. Tech and P. Engineering, “Study And Analyze The
Factors Related To Distance Education Using Data Mining,”
Int. J. Eng. Res. Technol., vol. 2, no. 2, pp. 1–5, 2013.
[67] José Gerardo Moreno Salinas and C. R. Stephens, “Applying [82]
Data Mining Techniques to Identify Success Factors in
Students Enrolled in Distance Learning: A Case Study José,”
Adv. Artif. Intell. Its Appl., vol. 9414, pp. 208–219, 2015.
[68] S. Pal, “Mining Educational Data Using Classification to [83]
Decrease Dropout Rate of Students,” Int. J. Multidiscip. Sci.
Eng., vol. 3, pp. 35–39, 2012.
[69] E. Yukselturk, S. Ozekes, and Y. K. Türel, “Predicting
Dropout student: An Application of Data Mining Methods [84]
in an Online Education Program,” Eur. J. Open, Distance e-
Learning, vol. 17, no. 1, 2014.
[70] M. A. Santana, E. B. Costa, and J. B. A. Rego, “A predictive
model for identifying students with dropout profiles in [85]
online courses,” CEUR Workshop Proc., 2015.
[71] K. R. Roll, I., Baker, R. S., Aleven, V., McLaren, B. M., &
Koedinger, “Generating Computer-Based Advice in Web- [86]
Based,” User Model. Springer Berlin Heidelb., pp. 367–376,
2005.
[72] V. Nabiyev, H. Karal, S. Arslan, L. A. K. Erumit, and A. Cebi,
“An artificial intelligence-based distance education system: [87]
Artimat,” Turkish Online J. Distance Educ., vol. 14, no. 2, pp.
81–98, 2013. [88]
[73] H. Karal, V. Nabiyev, A. K. Erümit, S. Arslan, and A. Çebi,
“Students’ Opinions on Artificial Intelligence based Distance
Education System (Artimat),” Procedia - Soc. Behav. Sci., vol. [89]
136, pp. 549–553, 2014.
[74] J. Gentry, “R Based Twitter Client,” CRAN, 2015.
[75] D. T. Lang, “General Network (HTTP/FTP/...) Client
Interface for R,” CRAN, 2016. [90]
[76] W. Ugulino, D. Cardador, K. Vega, E. Velloso, R. Milidiú,
and Hugo Fuks, “Advances in artificial intelligence - SBIA
2012: 21th Brazilian symposium on artificial intelligence
Curitiba, Brazil, october 20-25, 2012 proceedings,” Lect. Notes [91]
Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect.
Notes Bioinformatics), vol. 7589, pp. 52–61, 2012.
[77] P. Casale, O. Pujol, and P. Radeva, “Personalization and user
verification in wearable systems using biometric walking
patterns,” Pers. Ubiquitous Comput., pp. 563–580, 2012. [92]
[78] A. Reiss and D. Stricker, “Introducing a new benchmarked
dataset for activity monitoring,” Proc. - Int. Symp. Wearable
Comput. ISWC, pp. 108–109, 2012. [93]
[79] A. Stisen, H. Blunck, S. Bhattacharya, T. S. Prentow, M. B.
Kjærgaard, A. Dey, T. Sonne, and M. M. Jensen, “Smart
Devices are Different: Assessing and MitigatingMobile [94]
Sensing Heterogeneities for Activity Recognition,” Proc. 13th
ACM Conf. Embed. Networked Sens. Syst. - SenSys ’15, pp. 127–
140, 2015.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
21
D. Anguita, A. Ghio, L. Oneto, X. Parra, and J. L. Reyes-Ortiz,
“A Public Domain Dataset for Human Activity Recognition
Using Smartphones,” ESANN 2013 proceedings, Eur. Symp.
Artif. Neural Networks, Comput. Intell. Mach. Learn., 2013.
D. Roggen, A. Calatroni, M. Rossi, T. Holleczek, K. Forster,
G. Troster, P. Lukowicz, D. Bannach, G. Pirkl, A. Ferscha, J.
Doppler, C. Holzmann, M. Kurz, G. Holl, R. Chavarriaga, H.
Sagha, H. Bayati, M. Creatura, and J. R. Millan, “Collecting
complex activity datasets in highly rich networked sensor
environments,” 2010 Seventh Int. Conf. Networked Sens. Syst.,
2010.
O. Banos, M. A. ttila Toth, M. Damas, H. Pomares, and I.
Rojas, “Dealing with the effects of sensor displacement in
wearable activity recognition,” Sensors (Basel)., vol. 14, no. 6,
pp. 9995–10023, 2014.
J.-L. Reyes-Ortiz, L. Oneto, A. Sama, X. Parra, and D. A. I.-O.
http://orcid. org/000.-0002-4943-3021 Anguita, “Transition-
Aware Human Activity Recognition using smartphones.,”
Neurocomputing An Int. J., vol. 171, pp. 754–767, 2016.
W. T. Cho, Y. X. Lai, C. F. Lai, and Y. M. Huang, “Appliance-
aware activity recognition mechanism for iot energy
management system,” Comput. J., vol. 56, no. 8, pp. 1020–
1033, 2013.
D. A. Johnson and M. M. Trivedi, “Driving style recognition
using a smartphone as a sensor platform,” IEEE Conf. Intell.
Transp. Syst. Proceedings, ITSC, pp. 1609–1615, 2011.
N. Banovic, T. Buzali, F. Chevalier, J. Mankoff, and A. K.
Dey, “Modeling and Understanding Human Routine
Behavior,” Proc. 2016 CHI Conf. Hum. Factors Comput. Syst. -
CHI ’16, pp. 248–260, 2016.
P. Romanski, L. Kotthoff, and M. L. Kotthoff, “Package
FSelector: Selecting Attributes,” CRAN, p. 18, 2016.
A. Candel and V. Parmar, “Deep Learning with H2O Deep
Learning with H2O,” 3rd ed., no. September, H2O.ai, Inc,
2015.
S. Altowaijri, R. Mehmood, and J. Williams, “A Quantitative
Model of Grid Systems Performance in Healthcare
Organisations,” 2010 Int. Conf. Intell. Syst. Model. Simul.,
2010.
G. Ayres and R. Mehmood, “LocPriS: A Security and Privacy
Preserving Location Based Services Development
Framework,” Knowledge-Based Intell. Inf. Eng. Syst. KES 2010.
Lect. Notes Comput. Sci., vol. 6279, pp. 566–575, 2010.
G. Ayres, R. Mehmood, K. Mitchell, and N. J. P. Race,
“Localization to Enhance Security and Services in Wi-Fi
Networks under Privacy Constraints,” Lect. Notes Inst.
Comput. Sci. Soc. Informatics Telecommun. Eng., vol. 16, pp.
175–188, 2009.
A. Alvi, Z. Nabi, D. Greaves, and R. Mehmood, Controlling
real world pervasive environments with knowledge bases, vol.
6279 LNAI, no. PART 4. 2010.
R. Mehmood, R. Alturki, and S. Zeadally, “Multimedia
applications over metropolitan area networks (MANs),” J.
Netw. Comput. Appl., vol. 34, no. 5, pp. 1518–1529, 2011.
R. Alturki, K. Nwizege, R. Mehmood, and M. Faisal, “End to
end wireless multimedia service modelling over a
metropolitan area network,” in 11th International Conference
on Computer Modelling and Simulation, UKSim 2009, 2009, pp.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2668840, IEEE Access

22 IEEE ACCESS, MANUSCRIPT ID

532–537. [111] J. Hendler and T. Berners-Lee, “From the Semantic Web to

[95] R. Mehmood and R. Alturki, “A scalable multimedia qos social machines: A research challenge for AI on the World
architecture for ad hoc networks,” Multimed. Tools Appl., vol. Wide Web,” Artif. Intell., vol. 174, no. 2, pp. 156–161, 2010.
54, no. 3, pp. 551–568, 2011. [112] N. R. Shadbolt, D. A. Smith, E. Simperl, M. Van Kleek, Y.
[96] R. Mehmood and J. A. Lu, “Computational Markovian Yang, and W. Hall, “Towards a classification framework for
analysis of large systems,” J. Manuf. Technol. Manag., vol. 22, social machines,” in Proceedings of the 22nd International
no. 6, pp. 804–817, 2011. Conference on World Wide Web - WWW ’13 Companion, 2013,
[97] R. Mehmood and J. Crowcroft, “Parallel iterative solution pp. 905–912.
method for large sparse linear equation systems,” Comput. [113] N. R. Jennings, L. Moreau, D. Nicholson, S. Ramchurn, S.
Lab. Univ. …, 2005. Roberts, T. Rodden, and A. Rogers, “Human-agent
[98] R. Mehmood, “Disk-based techniques for efficient solution collectives,” Commun. ACM, vol. 57, no. 12, pp. 80–88, Nov.
of large Markov chains,” University of Birmingham, 2004. 2014.
[99] F. Alam, V. Thayananthan, and I. Katib, “Analysis of Round-
robin Load-balancing Algorithm with Adaptive and
Predictive Approaches,” 11th Int. Conf. Control, 2016.
[100] S. Van Buuren and K. Groothuis-Oudshoorn, “MICE:
Multivariate Imputation by Chained Equations in R,” J. Stat.
Softw., 2011.
[101] L. A. Tawalbeh, R. Mehmood, E. Benkhlifa, and H. Song,
“Mobile Cloud Computing Model and Big Data Analysis for
Healthcare Applications,” IEEE Access, vol. 4, pp. 6171–6180,
2016.
[102] N. Ahmad and R. Mehmood, “Enterprise systems and
performance of future city logistics,” Prod. Plan. Control, vol.
27, no. 6, pp. 500–513, Apr. 2016.
[103] N. Ahmad and R. Mehmood, “Enterprise systems: Are we
ready for future sustainable cities,” Supply Chain Manag., vol.
20, no. 3, pp. 264–283, 2015.
[104] R. Mehmood, R. Meriton, G. Graham, P. Hennelly, and M.
Kumar, “Exploring the Influence of Big Data on City
Transport Operations: a Markovian Approach,” Int. J. Oper.
Prod. Manag., vol. Forthcomin, 2016.
[105] R. Mehmood and G. Graham, “Big Data Logistics: A health-
care Transport Capacity Sharing Model,” Procedia Comput.
Sci., vol. 64, pp. 1107–1114, 2015.
[106] J. M. H. Elmirghani, B. Badic, Y. Li, R. Liu, R. Mehmood, C.
Wang, W. Xing, I. J. Garcia Zuazola, and S. Jones, “IRIS: An
inteligent radio-fibre telematics system,” in Proceedings of the
13th ITS World Congress, London, 8-12 October, 2006.
[107] G. Ayres and R. Mehmood, “On discovering road traffic
information using virtual reality simulations,” in 11th
International Conference on Computer Modelling and Simulation,
UKSim 2009, 2009, pp. 411–416.
[108] R. Mehmood and M. Nekovee, “Vehicular AD HOC and grid
networks: Discussion, design and evaluation,” in 14th World
Congress on Intelligent Transport Systems, ITS 2007, 2007, vol.
2, pp. 1555–1562.
[109] Z. Alazawi, O. Alani, M. B. Abdljabar, S. Altowaijri, and R.
Mehmood, “A Smart Disaster Management System for
Future Cities,” WiMobCity ’14. Int. Work. Wirel. Mob. Technol.
Smart Cities, pp. 1–10, 2014.
[110] M. Büscher, P. Coulton, C. Efstratiou, H. Gellersen, D.
Hemment, R. Mehmood, and D. Sangiorgi, “Intelligent
mobility systems: Some socio-technical challenges and
opportunities,” in Communications Infrastructure. Systems and
Applications in Europe, Lecture Notes of the Institute for
Computer Sciences, Social-Informatics and Telecommunications
Engineering, 2009, vol. 16, pp. 140–152.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
View publication stats http://www.ieee.org/publications_standards/publications/rights/index.html for more information.