Embodying The Educational Institution
Embodying The Educational Institution
Embodying The Educational Institution
4, December 2016
ABSTRACT
The way adults pursue their education through life is changing as the technology around us
relentlessly continues to enhance our quality of life and further enhances every aspect of the
different tasks we set out to perform. This exploratory paper looks into how every adult can
embody a comprehensive set of academic services, platforms and systems to assist every
individual in the educational goals that one sets. A combination of three distinct technologies
are presented together with how they not only come together but complement each other around
a person in what is usually referred to as a personal area network. The network in this case
incorporates an intelligent personal learning environment providing personalised content,
intelligent wearables closer to the user to provide additional contextual customisation, and a
surrounding ambient intelligent environment to close a trio of technologies around every
individual. Each of the three research domains will be presented to uncover how each
contributes to the personal network that embodies what one usually expects from an educational
institution. Three distinct prototype systems have been developed, tested and deployed within a
functional system that will be presented in this paper.
KEYWORDS
Higher Education, Personal Learning Environment, Intelligent Wearables, Ambient
Intelligence, Personal Area Network.
1. INTRODUCTION
Further and higher education institutions are struggling to keep afloat as financial and operational
difficulties have and continue to shift within a multi-facetted domain that is dealing with issues
related to competition, costs, human resources, accreditation, assessment, and quality assurance to
mention a few. Even the way and the quality of teaching and learning has shifted due to these
issues as educational institutions and academics had to adjust together with their epistemological
standing and pedagogical philosophy adopted. The high level group on the modernisation of
higher education[1] pointed out that the pedagogical models designed for small institutions
catering to an elite few are having to adapt, often under pressure, to the much morevaried needs
of the many, to greater diversification and specialisation within higher education, to new
technology-enabled forms of delivery of education programmes, as well as to massive changes in
science, technology, medicine, social and political sciences, the world of work, and to the onward
march of democracy and human and civil rights discourses. (pp.12) The impact of technology
DOI : 10.5121/ije.2016.4402
mentioned above is instrumental to a number of efforts to adapt to the changing times. Distance
education and eventually e-learning are different forms of delivery triggered by massive strides in
technology but which still carry their own challenges and limitations. The learner on the other
hand has to adapt to whatever the educational institution has to offer, which does not always
mean or result in an optimal outcome. As a matter of fact the demographics together with the
retention and students success rates tell a story of their own. Such matters within a competitive
and globalised higher education market have serious repercussions and damaging to an
educational institution [2]. The employment of technology within e-learning in an attempt to
alleviate these issues is a typical example of how some higher education institutions fail to focus
on the learner and instead assume that the technology will suffice. Macfadyen and Dawson [3]
point out that in reality those institutions that dominate their planning process by technical
concerns rather than what the learners need and focus on, fail to develop a clear vision for
learning technologies (pp. 161). Additionally, Beverly Park Woolf[4] specifically underlines this
issue when she attributes the shortcomings of instructional software to its inability to truly
respond to student needs and inflexibility to emulate teaching. In this paper a learner-centric
methodology will be presented as it embodies the basic elements of an academic institution. The
integrated e-learning platform attempts to provide all the required education needs of a student
with an essential personalisation element that tailors the content, delivery and feedback that a
learner expects from a one-to-one interaction with an educator.
The rest of the paper is organised as follows. The next section will cover the background
necessary to better understand the concepts that make up the integrated platform. This involves
the use of personal learning environments, the input from wearable and portable technologies, the
contribution of ambient intelligent surroundings, and the integrating medium of a personal area
network. The way these technologies and techniques converge together will be presented in
Section 3 where three distinct and autonomous prototypes are described in detail. In the next
section the test models are presented together with the documentation of the full set of results.
The future work and conclusions bring this paper to a close as they are presented in Section 5 and
Section 6 respectively.
2. BACKGROUND
The techniques and technologies presented in this section will help set the scene for their
employment in the next section. These distinct areas are research domain in their own right and
yet when combined together they promise to deliver more than they individually can as each adds
value and complement each other. The first to be presented is the Personal Learning Environment
or PLE that is not something new or innovative but which still plays an important role when
academic personalisation needs to be achieved. Secondly, every intelligent system requires input
from the user or from sensors like wearable and mobile devices as they are located in close
vicinity and part of the same user. As a matter of fact the second closest distance to the user is his
or her environment, and this can also serve as a sensory input but also as an output medium, and
this will be the third to be presented. Finally, the personal area network or PAN is simply a
concept and a technology that allows the previous three presented areas to coexist, interact, and
provide combined added value to the user.
10
set of tools and activities required for a personal learning network require one or more social
networking accounts to link up and communicate with other social networkers who have similar
interests and needs; follow, contribute and distribute content discovered or generated over a blog,
a wiki or any other social bookmarking online tool; join and participate in discussion groups, fora
and other social gatherings to acquire new information while at the same time sharing personal
knowledge with others. Much of these online tools have been made available and are possible
through the advent of Web 2.0 technologies [5] that characteristically present dynamic rather than
static websites displaying user-generated content.
The personal learning network element will be put into practice through the use of crowdsourcing
and the generic use of Web 2.0 technologies. Crowdsourcing is a simple practice of making use
of the collective knowledge of potential contributors to solve a common problem or to complete a
specific task. Such a practice is commonly applied and functionally proven within industry [9] as
overtly proven by the Open Source success within the ICT arena. There have been numerous
other situations where huge numbers of contributors have factored into the final overall success of
what they collectively had a vested interest in. Some familiar examples include Wikipedia,
Linux, Yahoo! Answers, and Mechanical Turk [10]. The concept of harnessing the power of the
masses has also been applied within a variety of domains. Bernstein, et al.,[11] report how
crowdsourcing was applied to a word processing application, while in another related project [12]
investigates how interfaces can be controlled by crowd input. In another project Christian,
Lintott, Smith, Fortson, &Bamford, (2012) apply crowdsourcing to the domain of astronomy.
Other examples can be found in journalism [13], art [14], medicine [15], government [16], and
finance [17]. Typically in industry, such a practice is employed to disseminate a problem a
company has encountered to all their collaborators and partners. These in turn send back potential
solutions and improvements that the same company collates and adopts to upgrade or enrich their
product. Literat[18] summed up the essence of crowdsourcing when she identified four
cornerstones of crowdsourcing, namely, connecting, communicating, collaborating, and learning
collectively. Within the context of this paper crowdsourcing refers to publically available
resources, as well as, content and contributions made available online. In particular, this research
evaluates the utilisation of crowdsourcing within the higher education domain, as one of the
practices employed to add value to e-learning. My interpretation of crowdsourcing in this respect
is simply the application and practical use of data provided over social networks to the domain of
education. The concept of employing crowdsourcing to higher education is a relatively novel
research area, but not a new one. It can be evidenced in a framework presented by [19] whereby
they aimed to evaluate active learning in conjunction with crowdsourcing. The improvement in
classification performance was compared to state-of-the-art methodologies and concluded that
crowdsourcing is a promising practice when applied to education, but noted that it could be
fallible in accomplishing specific tasks. They recommend that a thorough crowd evaluation is
required before applying the collective intelligence to work on a specific issue. Weld, et al., [20]
offer a direct association between online education and crowdsourcing, as they argue that such a
mechanism will likely be essential in order for e-learning to reach its full potential. The use of
social networks and the fact that crowdsourcing is considered to be leveraging available online
platforms is only part of its advantages. Another advantage that needs to be kept in mind is the
fact that it is a low cost and scalable method to source and mediate such rich and diverse
information sources [9]. The diversity of socially contributed opinions can actually diminish bias
especially in collective decision making within small teams as reported by Bonabeau[21]. This is
particularly true when one considers the financial repercussions of adopting a virtual
crowdsourcing panel of experts in contrast to a team of real professionals. Crowdsourcing comes
with its own issues and challenges especially due to the heterogeneous and unstructured nature of
12
the content itself. Web-based content especially that over social networks is predominately not
reviewed or humanly moderated, thereby creating a massive challenge to parse and compose
information that is valid and adequate for academic consumption. Some solutions lend
themselves very well already as they are easily separable into well-defined manageable sources.
Tools like ConsiderIt1 , SuggestBot[22], and Soylent2 , further assist to facilitate and alleviate this
classic yet fundamental crowdsourcing concern.
Another issue worth keeping in mind when considering social networks in conjunction with
education is the element of student engagement in relation to the connectivism learning theory
that will be discussed later in the chapter. Studies have clearly showed that there exists a direct
correlation between social networking and engagement. [23]have statistically confirmed, through
analyses of Twitter communications, that students and faculty were both highly engaged in the
learning process in ways that transcended traditional classroom activities (p.1). Their study
provided experimental evidence that Twitter can be used as an educational tool to help engage
students and to mobilize faculty into a more active and participatory role (p.1). Similarly,
Rutherford [24] has shown that there is a positive correlation between student use of social media
and the quality of their educational experience. The study gave positive insights into the impact
the use of social media can have on the level of pre-service student engagement. Other
studies[23] have also shown that leveraging social networks during the educational process
enhances student engagement.
They provide the required connections between users thereby facilitating communication,
collaboration, and collective learning at the same time. These are the reasons that led to the
adoption of social networks within the context of crowdsourcing as part of the personal learning
network component. The combination of education and crowdsourcing research is being argued
to be a natural blend that potentially offers fruitful outcomes. At classroom level, an educator is
more than willing, excited, and eager, to share knowledge and convey information. The same
educator employs the most appropriate medium to optimise the communication transaction, while
ensuring the most efficient educational process. This is not always the case unfortunately, but it
simply demonstrates that even though the educator might have the best of intentions to teach, and
the student is willing to learn, communication problems are possible. Creating the most
favourable environment to reduce the possibility of communication break downs is the least that
could be done. In the case of crowdsourcing, a favourable environment could be supported by
providing adequate tools and applications to encourage and engage providers and consumers. The
same consumers, in a recursive manner, become contributors and providers, thereby creating a
cycle within a collaborative healthy eco-system that implicitly spawn educational processes. The
research project assumes an idealistic scenario whereby educators enthusiastically contribute and
facilitate the educational process through their collaboration, passion and vocation.
Crowdsourcing provides the opportunity to harness the power of the crowd and deal with the vast
amounts of potentially useful data provided by massive numbers of users, contributors, academics
and experts. The challenge here that this research study is attempting to address is to process the
massive amounts of data available, which is otherwise humanly laborious to do, and use it
fruitfully. The outcome will be a freely available resource that provides content as part of the
learners personal learning network.
_________________________
1
2
http://consider.it/
http://projects.csail.mit.edu/soylent/
13
participants, but also personal priority issues together with re/scheduling habits and patterns.
Within the academic domain the profile generated encapsulates as much as possible the
comprehensive learner characteristics that deal with knowledge, interests, and educational needs.
In this respect a learner profile is considered a collection of inferences about information
concerning a student that one is not able to observe [32]. The main use of the learner profile is to
adapt and personalise the learning process as well as the content and the delivery of the
educational material. An automated learner profile can be generated using Computer Science
techniques that go beyond the scope of this paper but for completeness sake the most commonly
employed profiling techniques will be highlighted shortly. Adding value to services and
personalising the information presented to the user is a very well researched domain within the
ICT arena [33]. Numerous methods are available to generate a specific user profile that will
eventually be employed to tailor the information intended for user consumption [34]. In this
context the students individual profile can be employed to identify and adapt the educational
content for their focused use. Such practices have been successfully employed to generate user
profiles and personalise the service provided to the same users.[35]employ user profiles to
personalise information access to control the information overload problem experienced by web
users. Similarly, the same concept was applied to healthcare consumers [36], e-commerce
applications [37], and mobile guides [38]. Within the educational domain there have been several
research projects and initiatives that employed personalisation to deliver educational content [39].
The use of online resources and web services have also been investigated by researchers like[40]
to enrich e-learning in terms of personalised learning systems, where the system itself was able to
trace learner needs and track progress. This would fit very well with the proposed research
whereby large amounts of users could potentially possess individual and unique academic
profiles. As a matter of fact [41] document how research in e-learning started taking form in a
way that also involved the tailoring of tools, terminals, and communications, to the needs of the
individual people as captured through their profiles, needs and experiences. In this context, a
number of authors started describing the use of developing the right technologies to be able to
provide a more effective e-learning service [42]. Most often this type of research was rather
detached from the complexity of the human learning involved, and this is why such systems were
morphed into personalised learning networks more than simple systems [43]. What becomes
obvious from such setups is the convergence to an idealistic learning scenario. A textbook model
would be a one-to-one student-teacher set-up whereby the content provided is perfectly tailored to
the students needs and interests. Such personalisation is what this research study is aiming to
investigate within a wider context and with the help of additional online tools. Three of the most
widely employed and artificial intelligent techniques employed to generate a user profile will be
briefly presented and brought into context.
Association rules are simple directives that imply specific associations or correlated relationships
between groups of items within a specific domain. Such rules are generally employed to discover
patterns from data collected. Given a set of facts about the users academic achievements, and an
associated list of tasks and topics, the rules formalise the connotations between them. For the sake
of simplicity, a classic example involves shopping trends of consumers. It is common knowledge
that whoever purchases wine and cheese, the chances of purchasing grapes as well is about 75%.
Such a rule, with an associated numeric value, can work out what future consumers will likely to
go for after purchasing cheese and grapes. Association rules have been applied to different
domains like large databases, eCommerce, and education. If such rules were to be employed, then
some previous knowledge of students trends and interest will be required to create the required
association rules.
15
Another artificially intelligent technique used to generate a user profile is called case-based
reasoning that makes use of previous cases that are similar to the current problem at hand and
applies or adopts the solution to the situation. So, given a student who has a problem with
understanding a particular academic topic, the case-based reasoner retrieves relevant cases that
match such a request and adapts solutions that were effective to solve the similar problem. The
difficult part for the classification task arises when the system is required to identify a target class
for a case that has no classification. In such instances the solution to this dilemma is simply fitting
the class that is most similar. Case-based reasoning has been employed as a learner profile
generator in various customisation scenarios like web information searching, topical filtering of
data, and document clustering.
The final computer science technique that is commonly used to create a user profile automatically
is through the use of Bayesian networks. A network is simply a set of points interconnected to
each other with links or lines. This system makes use of lines to link items of interest that are
related to each other. This means that there is no relationship between disconnected topics of
interest to a student. On the other hand a subject that is connected to another topical matter, that
in turn is connected to a third theme, automatically infers a transitional association, thereby
proposing the third item as a potential topic of interest to the student. [44]make extensive use of
such networks to capture students behaviour within an e-learning system and model their learning
styles. Other uses include web browsing personalisation, intelligent help systems, and expert
systems.
What emerges from the above brief descriptions is that any of these basic techniques can be
employed to generate a student profile and personalise the content that is presented to the same
student. This does not mean that they all perform equally but that they do so differently. The
ultimate goal, as far as this paper is concerned, is that an artificial intelligent piece of software is
applied as part of the personal learning portfolio component, which together with the personal
learning network make up the e-learning personal learning environment under investigation.
Whereas the functionality of the personal learning network component is achieved through the
use of crowdsourced social networks, the personal learning portfolio component is implemented
through a simple process of user profile generation that is sourced through the combination of
explicit interest declaration and the eventual interactions with the environment. The learners
feedback is used to refine the generated profile to better personalise the educational content that
follows.
2.1.3. PLE and Personalisation
The combination of a personal learning network (PLN) and a portfolio (PLP) help in establishing
an environment that is not only personal, but even more effective due to its customised and
tailored content that fits even closer to the users needs and interests. Siemens [44] explored
aspects of personalised learning with a focus on how to connect all the information provided
online in a way which makes sense in context; using networks to help amalgamate all the
information acquired in a meaningful way. [45]report how these technologies connect knowledge
workers to their online personal networks for information exchange, informal learning, and social
support, thereby supporting the notion of a personal learning network that has value-added
advantages due to the use of social media. The personal learning environment brings together the
two components (PLN and PLP) in a conceptual way within an integrated e-learning user
interface. [46]have teamed up in an attempt to integrate personalisation in the online courses they
offer. This partnership that started in 2013 has launched a full scale initiative earlier this year by
16
offering four hundred thousand first year university students the possibility to make use of
personalised educational services. Tailored feedback and customised academic advice was
delivered based on information that was extracted from the same students success and failure
patterns while going through the educational material. Another interesting partnership was struck
between University of Edinburgh and CogBooks who developed an online tool that personalises
the students graphic user interface as they progress through the different course activities [47].
The academics in return have used this same information to fine-tune their material and teaching
in general. Two other similar partnership between CogBooks and Arizona State University and
University of Colorado Boulder have also been using personalisation techniques to provide
formative feedback to individual students based on analysis of learner-generated data [48].
CogBooks are encouraged by the result obtained and claim that they are successfully achieving
their goal of educating everyone uniquely [49]. Similar results were reported by the University
of Wisconsin-Milwaukee [50] who developed a self-paced programme called U-Pace that
incorporates personalised feedback, individualised progress reports, and motivational
notifications to make students aware of their strengths and weaknesses. The outcome showed that
students performed sixteen percent higher on assessments over those who did not follow the UPace programme. This trend is also evidenced in courses being offered by MITx[51], the MIT
wing that gives away free online courses, with the premise that specific students, as a result of
their declared needs and interests, might be presented with variations in the academic content
presentation. Two other related initiatives worth mentioning are those established by the Bill &
Melinda Gates Foundation, and the IMS Global Learning Consortium. The Gates foundation set
up a grant program called ALMAP (Adaptive Learning Market Acceleration Program) that
promoted personalized learning research, while setting up also a ground-breaking learning
program called Enlearn whose purpose was to assist and encourage the development of adaptive
learning material that can enable a more personalized teaching and learning experience thereby
transforming the entire classroom ecosystem into an adaptive environment suitable to the learning
needs of each student. On the other hand the IMS initiative brought together a consortium of over
three hundred universities, higher-education institutions and vendors in an effort to standardize
and establish a shared vocabulary for recording students academic data. The protocol of metrics,
called Caliper[52], was intended to make it easier to describe a learners profile across institutions
and learning environments.
It is obvious and natural that a human educator is much more effective when a personalized
methodology is employed. Within an e-learning environment such recognition is also being
confirmed as institutions across the world agree that a single invariable and inflexible style,
method or approach is not possible for all learners [53]. Lonn et al.,[54] define personalized
learning in a way that higher education institutions can take technological advantage through the
measurement, collection, analysis and report of data about learners and their contexts, for
purposes of understanding and optimizing learning and the environments in which it occurs (p.
4). The on-going research in this area is on the rise as access to data tools and techniques are
easier to use and highly accessible together with the availability of large sets of data that will
assist in the customisation of the learning process.
18
3. CONVERGENCE OF TECHNOLOGIES
The technologies highlighted in the previous section have been implemented within functional
prototypes as a proof-of-concept to show that a personal educational network (PEN) is possible
following best practices. The PLE developed has already been widely accepted in academia [56]
and even though it has been through extensive testing, the intelligent PLE is now being employed
in conjunction with other technologies as well as evolving to accommodate latest developments.
The intelligent wearables have also been developed as a proof-of-concept within another domain
but have now been adapted to serve as input sensors to the PEN. Finally, the ambient intelligent
system is the encompassing system that homes both prototypes and ensures that the delivery and
the communication with the user is consistent and effective.
together with the educational material employed, the methodology engaged, sequence, medium
through which its presented and finally the overall success or not. Tomei[57] defines pedagogy
as the art and science of teaching children (p. 1), and describes how the evolution of learning
theories has transformed the pedagogical model from a state of submissive or receptive child and
teacher knows-it-all, to a learner-centered and academic facilitator. Pedagogy must not be an ad
hoc concept that is left to chance or not given enough thought and planning, but requires sound
theoretical foundations especially within the area of technology-enhanced education. McKenzie
[58] points out that its because of a pedagogical model was not followed that numerous
academic institutions had a low return on their technological investments. McKenzie was reacting
to a statement by the US secretary of education, Dr Roderick Paige, who side-lined the
importance of pedagogy and imposed changes that were not grounded in any learning theories.
Based on these factors, this documented research study together with the methodologies adopted
are structured around a predominant learning theory, Connectivism. Connectivism[59] is
considered by numerous researchers (Downes, 2008; Kop & Hill, 2008; Duke, Harper, &
Johnston, 2013) as the leading learning theory in the digital age as social networks and learners
online presence is considered influential on their academic work and personal lives [60]. The
authors argue that according to George Siemens theory online social network contacts represent
a potential and valuable source of information (p.142). This source of information is not enough
and definitely does not constitute a complete learning environment. As a matter of fact Ng [61],
amongst others (Hung, 2014; Duke, Harper, & Johnston, 2013), asserts that learning theories that
support online learning like connectivism need also take into consideration those teaching
contexts that are not in real time (asynchronous) as these situations have a major impact on the
learning outcome. In this respect Mayes & De Freitas[62] actually argue against the adoption or
need of new learning theories to accommodate the digital age and assert that all that is required
for effective learning is the knowledge of how the underlying processes and theoretical constructs
enable learning, be it face-to-face or over e-learning. The point being made here is that a learning
theory adequate for learning within the digital area and applied to this e-learning research is not
enough or complete in isolation. This is especially true when a combination of methodologies is
being proposed to enhance the effectiveness of e-learning. As a matter of fact Ng (2015)
subscribes to this same notion when he states that It is inevitable that the blending of more than
one learning theory in the design of a sequence of pedagogically sound learning activities would
be required (p.93).
The connectivism learning theory has been associated with the use of social media in education,
and coined as a learning theory for the digital age [63]. This theory puts into context the online
reality of learners making use of social networks as it dismisses the three dominant learning
theories, behaviourism, cognitivism, and constructivism, according to Wheeler (2012). The
educational process is envisaged external to the learner within a personal network of
technologies, communities and social media. Closely related to this definition also lies the socialconstructivism theory that according to Vygotsky[64] learning occurs as a result of interactions
between individuals influenced by the cultural and societal environment. Whereas this learning
theory takes into consideration the role of others within the learning process as mediators to
acquire novel information and knowledge, connectivism takes it a step further and highlights the
importance of the networked information whereby the learner and the mediators contribute and
receive in a mutual beneficial learning community. I particularly argue that the connectivism
learning theory significantly contributes to this research project as it highlights the importance of
learners identifying the source and the content itself of what interests them and what they need to
learn. This places the responsibility directly on the learner who is required to bring together a
20
cohesive set of personal learning tools within an environment that is socially networked and
academically healthy within which learners can store their knowledge. Such an ideology
subscribes to my own post-positivistic epistemological point of view whereby the contextual
reality of an online experience determines and distinguishes our overall interaction and the
amount and quality of what we intellectually extract. The medium employed is clearly an
imperative factor in the facilitation of the learning process. The extent and capacity of the
mediums influence is also dependent on the student at the receiving end of this interaction. A
number of educational studies have been reported that directly refer to the learning theory of
connectivism. Loureiro and Bettencourt [65] investigated how to enhance the educational process
by focussing on optimising such process within higher education by integrating Web 2.0 tools
and subscribing to connectivism. Robson [66] took a step further to investigate the next
generation of online courses by scrutinizing the content and processes of initial generations of elearning courses. He draws the conclusion that e-learning content is experiencing a shift in its
underlying pedagogical theories from cognitive, instructive, and behaviourist to social,
constructivist and connectivist. Even Duke, Harper, & Johnston, [67] argue that connectivisms
diversity through different networks is ideal to assist learners in the new generation to learn.
They encourage educators to continually evaluate how connectivism in conjunction with other
learning theories can be used in the online learning process. Furthermore, Hung [68], makes
extensive use of ideas from this same learning theory to design new models in an effort to
optimise the movement of connected knowledge, expanding learning spaces and structures, and
employing open technology to connect people.
Based on this learning theory and implemented as an online course the intelligent personal
environment (Figure 3) personalised the delivery of the educational material while at the same
time employed the users interests and interactions to personalise and further tailor the interface
and the content alike. Machine learning techniques were employed to develop and evolve a
learner profile while crowdsourcing and social networks, namely Facebook and Twitter, were
employed to complement the content.
21
to be the most frequently used words by the user after analysis of the gathered data to the groups
that correspond best to the interests categories. Thus if the list of most commonly occurring
words contains some word that is found in the list of identified interests, then that particular
interest category is marked as relevant. Although this is one form of classifying the user, it was
deemed too trivial and too risky when considering the result accuracy. As a second measure of
profiling the word ontology results were introduced by which the system compares the synonyms
of the most frequently used words to the stereotype categories. To further complement this, in the
final stage of categorization, the system also looked at the synonyms of the landmark types and
performed one final check in order to categorize the user into the most representative categories
based on his interests. That ensures that if the list of most commonly occurring words does not
contain the exact name of an interest field, then more checks are carried out to increase the
chances of obtaining a hit. At the end of this cycle the result would be a user profile consisting of
the interest fields that are deemed to be of interest to the user.
The intelligent wearables were used as both input and output media to help integrate learners with
the personal learning environment and also together. Recommendations generated through the
identification of interest fields were communicated through a wrist band and hyperlinked to
educational content within the correct context. Tests using iWatch3 and Fitbit4 have shown that
seamless learner notification together with customised content according to the specific interest
adds value and enhances student understanding especially if depicted within an ideal context [69].
Input wearable devices are more limited but the use of the wristband served also to pair up
learners with common interests. The different user profiles were coded onto the wristband from
the backend server and once one wristband identifies the vicinity of another matching wristband a
vibrating notification alerts both learners. Such methodologies were already found to be effective
[70] and their use highly recommended at all level of education. In this particular context their
combination with other technologies within a higher education scenario they tend to be more
effective and inspirational.
http://www.apple.com/watch/
https://www.fitbit.com/eu
23
4. TESTING
Each individual component was tested in isolation with the latter ambient intelligent technology
making use of the former two. The intelligent personal learning environment was tested with a
hundred and twenty higher education participants with positive results [71] as it was compared to
the traditional face-to-face medium and the classical static e-learning environment. The study
showed that by combining three techniques, crowdsourcing, learner profiling and personalisation,
into a functional system it was possible to enhance effectivenessby delivering a tailored
environment that adapts to the learner. It also showed that it is possible to make use of social
networks to crowdsource additional information to supplement the available academic content.
The intelligent wearables were also tested with a smaller group of forty-five university students
who were first asked to explicitly mark which of the interest fields they thought best described
their interests, and later asked to make use of the system together with purposely purchased
wristbands. All the students owned a smartphone and tested the PLE over their mobile device
while networked with the wristband and the backend application. The system extracted the
participants social media profiles and generated a learner profile accordingly, and eventually
compared to the interest fields marked by the learners. The results showed an average precision
rate of 81% when the system classified the learner into one correct category and return relevant
suggestions. The recall values returned by the system ranged from 60 to 100% when relevant
interest field were found in the learner profile. Students were also asked about the general
deployment of the wearable devices and the use of their smartphone as part of their academic
experience. The feedback was astounding as the use of such devices tended to blend much more
with their everyday interaction outside academia and thereby resulted in a much more natural
interaction rather than a forced one.
Finally the use of simulated ambient intelligence was evaluated as the students were asked to
comment about the use of the labs.
5. FUTURE WORK
More work could be done to improve the performance of the system with respect to its
personalisation capabilities, more specifically, to improve on the Ontology-based approach
adopted in this study. The first issue that should be tackled is the cold-start problem that the
system might encounter when working on some profiles. This problem could be tackled by
looking at alternative sources through which it could acquire data for personalisation, which may
be other social media platforms or through mild forms of explicit data gathering. Also, the use of
a hybrid approach to personalisation would perhaps be ideal. Boosting the personalisation
capabilities of the system could also be achieved through obtaining more information about the
user from other sources. There is a reluctance to move towards explicit data gathering but the
need for better input data is clear. This can be achieved from other sources such as a users
browser history. Future research should focus on how e-learning turns the tables on the learners in
relation to participation, behaviour and required effort. What methodology should be adopted to
effectively introduce the learners to the e-learning platform? Which learning theory subscribes to
such a philosophical undertaking? How can resistance to change be controlled and actually
reverse the learners mind-set? How will this be implemented and eventually performed in
reality?Other future research directions might include the customisability and control of the
environment by the learner, especially in the tertiary level. Participants pointed out that they
24
would have recommended the iPLE to their peers if they could switch it on and off at their will.
This interesting concept emulates the behaviour of online learners who are very focused when
they have a specific objective to reach, but willing and open to related suggestions and
recommendations when they are following a course, like a MOOC, out of interest and without
any assessment repercussions or time restrictions. Another research direction can potentially be
within the crowdsourcing domain where additional sources could be automatically included and
harvested to add richer and diverse content to the knowledge base of the iPLE. The current
version collates information from pre-stated sources on specific interest areas. An extensible and
fully automated system would be able to enhance the repertoire, or even better, refine the interests
categories thereby optimising further more the personalisation process of the e-learning
environment.
6. CONCLUSIONS
This paper has been a tiny step towards the right direction. An attempt into how to optimise elearning through a personal education network. The research that pursued following this goal
characterised the design, methodologies adopted and the actual development of the prototypes
presented and evaluated. The evaluation concluded that the goals and objectives of the research
have been successfully met or exceeded. This research study and this paper do not only
recapitulate all the hard work performed over the last four years, and nor do they characterise the
end of an exhilarating journey, but merely demarcate the beginning of a promising way forward
as new research avenues have been uncovered which potentially could characterise the future of
online education and intelligent e-learning platforms and personal educational networks. This is
all very promising and encouraging because the work presented helps to improve and enhance
peoples interaction and attitude towards e-learning and online education in general.
REFERENCES
[1]
M. McAleese et al., "Improving the quality of teaching and learning in Europes higher education
institutions," High Level Group on the Modernisation of Higher Education, Report to the European
Commission, Luxembourg, ISBN 978-92-79-30360-9, 2013.
[2] M Yorke and B. Longden, Retention and Student Success in Higher Education. Maidenhead, UK:
Open University Pres, 2004.
[3] L. P. Macfadyen and S. Dawson, "Numbers Are Not Enough. Why e-Learning Analytics Failed to
Inform an Institutional Strategic Plan," Educational Technology & Society, vol. 15, no. 3, pp. 149
163, 2012.
[4] B. Park-Woolf, Building Intelligent Interactive Tutors : Student-Centered Strategies for
Revolutionizing E-learning. Burlington, MA, USA: Morgan Kaufmann, 2010.
[5] N. Sclater, "Web 2.0, Personal Learning Environments, and the future of Learning Management
Systems," Colorado, 2008.
[6] S. Leone, Characterisation of a Personal Learning Environment as a Lifelong Learning Tool. Ancona:
Springer, 2013.
[7] S. Gooren-Sieber and A. Henrich, "Systems for Personalised Learning: Personal Learning
Environment vs. E-Portfolio?," in Hybrid Learning. Germany: Springer, 2012, pp. 294-305.
[8] D. Morrison. (2013, January) Online Learning Insights. [Online].
https://onlinelearninginsights.wordpress.com/2013/01/30/why-students-need-personal-learningportfolios-more-than-we-do/
[9] L.P. Cox, "Truth in Crowdsourcing," IEEE Journal on Security and Privacy, pp. 74-76, 2011.
[10] A.D. Ramakrishnan and A.Y. Halevy, "Crowdsourcing systems on the World-Wide Web,"
Communications of the ACM, vol. 54, no. 4, pp. 86-95, 2011.
25
28