Evolving a Social Networking Platform into a Smart Personalised Learning Environment (PLE) or the Other Way Around: Your Choice?

Abstract

The future of learning environments lies with the merging of the better aspects of Learning Management Systems (LMS), with those popularised in Social Networking platforms, to personalise the individual learning experience in a PLE (Personal Learning Environment). After examining the details of a particularly flexible LMS, followed by the investigation of several key data structures behind the Facebook social networking platform, this paper then demonstrates how such a merging can be done at the conceptual schema level, and presents a list of novel features that it then enables.

Full text

TITLE: Evolving a Social Networking Platform into a Smart Personalised Learning
Environment (PLE) or the Other Way Around: Your Choice?
!
AUTHOR: Steve Goschnick, Swinburne University of Technology, Australia |
sgoschnick@swin.edu.au
!
THIS IS AN EARLY DRAFT COPY!
Abstract: The future of learning environments lies with the merging of the better aspects of
Learning Management Systems (LMS), with those popularised in Social Networking platforms, to
personalise the individual learning experience in a PLE (Personal Learning Environment). After
examining the details of a particularly flexible LMS, followed by the investigation of several key
data structures behind the Facebook social networking platform, this paper then demonstrates how
such a merging can be done at the conceptual schema level, and presents a list of novel features that
it then enables.
!
Keywords: Learning System, Personalised Learning Environment, PLE, Learning Management
System, LMS, Social Network, Facebook, Conceptual Design, Conceptual Schema
!
1. Introduction
!
This paper investigates the fusion of social networking capabilities with those from a fine-grained
Learning Management System (LMS), into a personalised learning environment (PLE), described
largely in terms of a design expressed as a conceptual schema. The social network platform
investigated is Facebook, which perhaps surprisingly, has several of the necessary generic structures
already in place, should that company choose to pursue the emerging personalised learning
environment marketplace. However, an LMS could equally be evolved, taking on aspects
popularised by social networking platforms, in new ways that personalise a learner's experience.
!
Some General Background
!
In situating the research in this paper it is first worth noting that learning systems have advanced in
parallel with similar strides in learning theory, and those in ICT (Information Communication
Technology) in general. Although they are interrelated, we concentrate here on the technology side,
with brief reference to some theory in learning, only when it is useful to the main research direction.
We are primarily concerned with personalised learning environments (PLEs) – which are predicated
on putting the student into a more central and active role in their own learning, both with regard to
the design of learning materials, and the learning path or trajectory they take in using them. Such
an underlying basis of PLEs makes them both more personalised and more adaptive than the typical
LMS (Learning Management Systems – such as WebCT, Blackboard and Moodle, and many lesser-
known products) that thousands of universities and schools have used for up to a decade now.
!
Tsolis et al (2011) state that 'A traditional LMS offers to all its users the same services and content,
meaning that all learners taking an LMS-based course, regardless of their knowledge, goals and
interests, receive access to the same educational content and the same set of tools, with no further
personalized support.' Some envisage PLEs as an approach that will replace (or at least speed-up
Page ! of !1 27

the evolution of) the LMS (Downes, 2010; Trolis et al, 2011). According to Downes: 'the typical
LMS is static, declarative, authority-based' whereas 'a personal learning environment is learning
through community; a PLE is a technological tool that allows us to do (that)'. According to Trolis
et al: they are 'extending the capabilities of a traditional open source LMS (Moodle), into (their)
proposed OWLearn system … an adaptive, personalized and open source e-Learning system' – yet
that paper is little more than a wish-list of the features they want to see realised in it, with little in
the way of design or specification toward achieving it. Others see a merging of PLEs and LMS into
something new beyond either (Mott, 2010). According to Mott: 'the LMS has become a symbol of
the status quo that supports administrative functions more effectively than teaching and learning
activities'; and that 'PLEs offer an alternative ...but with their own limitations' including security
and reliability. Mott believes the two approaches should and can be mashed up into 'open learning
networks'. Mott presents a good comparison of the strengths and weaknesses of both the LMS and
the PLE, as he sees them.
!
The concept of a PLE preceded the arrival of well endowed tools to demonstrate it, and as such
there is some variation around the edges of what a PLE is, or should be. Attwell (2007) believed
that the tools were already available. He believed that the software applications many people use
regularly, together with the services they call upon, could be aggregated to enact a PLE just by
changing the mindset of both students and teachers toward personalised learning. His example
application tools included: the word processor, Keynote (for presentations), Net Newsreader,
Garage Band (for podcasts) and iMovie (for video editing). While his example range of existing
services that could be folded in included: Skype, Delicio-us (for sharing bookmarks), Flickr (for
sharing photos), and Joomla (for creating web sites). 'It was not a software application, instead it
was more of a new approach to using technologies for learning... by the net-generation'.
!
Downes (2005) was happy with the blogging tools of the day – e.g. Blogger and WordPress – but
he saw a further need for some automation in the way that these page islands of output by creative,
active learners could be interrelated. Initially, the RSS aggregator (of newsfeeds from blogs in RSS
format) was Downes' early answer to that automation, in what Mott called 'the PLE (is) the
educational manifestation of the webs “small pieces loosely joined”' - incorporating a quote from
Dave Winer, the inventor of RSS.
!
Some researchers took that early optimism about available tools and services, and put it upon their
'net-gen' students to 'put students in a more central position in the learning process by allowing
them to design their own learning environment' (Valtonen et al, 2012). They embedded this in the
running of actual vocational courses in Finland, across a range of disciplines. Valtonen et al asked
themselves 'what kinds of personal learning environments would (the) students produce'. The
feedback was mixed, with many students expressing the negative impact of the learning of the tools
needed, against what they actually needed to learn in their respective subjects (i.e. catering services,
international business, health care, computer science, engineering, massage therapy). The
researchers cite their experience in retrospect as 'romantic constructionism'. One of their
conclusions is that instead of focusing on technology 'the emphasis ought to be on the pedagogical
demands of PLEs in education', and interestingly, that 'with adequate pedagogical support from
teachers, students can potentially make use of PLEs for learning and thus develop their
metacognitive and self-regulative skills'.
!
Page ! of !2 27

Beyond the use of mundane/everyday software applications and existing services by net-gen
learners to spontaneously create or mashup effective PLEs, the two other approaches advanced in
the research thus far to achieve a PLE are: to evolve an existing LMS towards a PLE; or thirdly, to
create a new PLE unrelated to any founding LMS. A fourth approach considered in this paper: is to
examine what would be necessary to evolve an existing social network platform into a PLE.
!
So what features set a PLE apart from an LMS? Personalised learning includes both personalisation
to suit the individual learner – this often requires a much more detailed profile than in a typical
LMS, with tagging (i.e. assigning keywords) to highlight an individual's interests and also a similar
set of tags to represent their existing knowledge. Personalisation includes adaptability of the
learning materials presented, and also allows or empowers the student to be more pro-active in
creating learning material themselves, often in a collaborative way with participating peers. The
adaptability achieved through techniques such as using tagging, can be manual or automated in
some manner. The 'smart' part in the title of this paper, refers to some automation of the
personalisation and any automation of the adjustable learning content, as compared to PLEs that
only allow for such adjustment in a more manual manner, either by the student or the teacher.
!
Within learning technologies, the now well-used traditional LMS are generally characterised with
features that do not lend themselves well to personalisation and adaptability to individual learners,
such that those taking that route are on a difficult path to achieve their goal - unless the LMS has
more flexibility than most, or the necessary changes are coming via the LMS vendor. The open
source LMS called Moodle gains some such flexibility by way of the open source approach to its
licensing, allowing others to add plugins to it, a route that a number of the researchers of papers
cited in this paper, have taken or are in the process of taking. There are other LMSs that have
considerable flexibility in their initial design, such as the one described in this paper in Section 2.
!
In a significant contrast to the Valtonen et al (2012) approach of putting the onus on net-gen
students to come-up with their own novel PLEs, Limongelli et al (2011) take the third approach to
having a PLE: creating a new one from technologies unrelated to a traditional LMS. Adding to the
contrast, they put emphasis on helping the teachers role (by adding some smarts via some
automation) in providing a PLE experience to students. The basic technology they do call upon in
the absence of an LMS, is the use of planner technology based on Linear Temporal Logic (Mayer et
al, 2007) from AI, and the way they use it serves as a useful illustration of applying a formulaic
approach to adaptive learning:
!
From the adaptive learning perspective, Lecomps5 (their system) is for creating learning paths
through a sequence of Learning Objects (LO) that have metadata added to each, which they then
call Learning Components (LC). Their system allows for two types of such sequences: one where
the sequencing is performed at the beginning (by the planning software), and the second, an
adaptive step-by-step sequencing, following the student during their path through the graph of
Learning Components. Both types use quizzes to fuel the decision points in the planning software.
The metadata is added to the Learning Objects by the teacher and includes two sets of tags (either
from formal ontologies, or else any other terms): concept prerequisites are set as a series of
Required Knowledge (RK) tags – just identifiers representing a set of concepts; and the concepts-to-
be-acquired as a series of Acquired Knowledge (AK) tags. Each of those two series of tags represent
a set of like knowledge items (ki).
Page ! of !3 27

!
As a learner moves through a course (a sequence of LCs), their Cognitive State (CS) is represented
by a subset of ki from the pool of LCs (plus any ki's they had previously acquired as prerequisites
relevant to the current course). The Lecomps5 system also provides for different learning styles
(LS) of students (initially four styles: intuitive-visual, intuitive-verbal, sensing-visual, sensing-
verbal – adapted from Felder & Silverman, 1988), by having four different versions of the LCs
(specifically, variations of the instructional content within them). They then consider they have a
relatively simple student model (SM): Cognitive State (CS) and the LS measure of the learner i.e.
SM = CS + LS (Note: they allow the teacher to put weights against the four learning styles for the
one student, as a vector that can be adjusted over time). A traditional LMS requires a significant
amount of work on the part of the teacher to manually select and sequence the learning content and
activities, so the Lecomps5 system is focused on lightening that load on the teacher considerably.
!
Their approach to adding smart technology into the learning process does put focus back on
learning theory (which as already stated, is not the focus of this paper). However, a learning
framework by Luckin (2008) is specifically aimed at using technologies and other resources beyond
the desktop, as a scaffold (meaning tutorial assistance) for learning. It is called The Learner Centric
Ecology of Resources. It helps design learning experiences that matches up available resources to
the needs of the learner. The underlying premise is that the learner needs to be in a collaborative
environment, where a more abled participant can step in when need be, to occasionally guide the
learner, who is otherwise advancing at their own pace. The aim of most smart learning technology
is usually to be a part of that assistance, and/or facilitate it from other human participants. Luckin’s
framework sets about making that possible: “For technology to provide software scaffolding the
system needs a model of its learners”. The PLE developed by Limongelli et al (2011), as we have
seen, has a simple Learner model (SM), one that includes their knowledge state and learning style.
However Luckin’s approach is much more detailed and based on well-accepted learning theory.
!
The PLE developed by Limongelli et al (2011) is described by them as providing 'automated
production and adaption of personalized courses', but it has no features that enhance collaboration
or that enables learners to author their own content, nor any use of social software to make up for
those gaps in a PLE feature set. Nonetheless, they inform the reader that in another research effort
they are integrating their approach with Moodle – presumably for its administrative and other
functions, including the feature that allows it to have plugins from the community around that
particular open source LMS.
!
In a strong sense, the focus and tool development on automating teacher tasks by Limongelli et al,
and the focus of Valtonen et al (2012) upon the learner tasks with the tools at hand, are near the two
extremes of the envelope of features and focus that a PLE can have. Our research is looking toward
a good balance somewhere in the middle.
!
What Follows in This Paper
!
Section 2 describes in detail a pioneering online education content and delivery system developed at
the University of Melbourne (by a company then wholely owned by the University) in 1997-98
which foreshadowed the presentation of learning objects we now see in iBooks on the Apple iPad. It
was more flexible and adaptive than most mainstream LMSs then and now. As well as describing
Page ! of !4 27

the Creator LMS, we look at key structures in that system which lend themselves to similar usage in
a PLE. It was also early in employing 'smart' technology in an LMS, in the form of intelligent
software agents, to seek out a certain form of specific data across the web – again, a technique and
tool that could be much more broadly applied in a PLE future.
In Section 3 we look at structures describing the Facebook social network platform, from research
that inferred a conceptual data model from the textual programming API (Graph API V2.0) that
Facebook makes available to third party developers. We do this with the aim of uncovering aspects
of a social network platform that will add user-friendly proactive content creation features to a
smart PLE, where a student may create, curate, participate and share such content with peers and
friends, regardless of context, mobile or otherwise.
!
In Section 4 we draw from the models and experience presented in Sections 2 and 3, with a view to
putting some combination of them together in the one PLE system, in particular with an emphasis
on what would be needed to turn the existing Facebook conceptual model into a Personal Learning
Environment (PLE), as one such future possibility.
!
2.1 Technical Parallel between an Old and a New Learning System
What first sparked this research was an observation of a direct parallel in the technologies used to
deliver online course material to the browser within an earlier pioneering educational content
creation and management system, and a much more recent need to deliver personalised educational
content to tablets and large-format smartphones such as Apple iOS devices. This new need requires
high fidelity presentation and fine-grained delivery of educational content down to the page level,
and even multiple learning objects on such a tablet displayed page. The earlier system was called
Melbourne IT Creator and was briefly introduced in Goschnick (1998).
!
The authoring tools in the Creator platform were delivered in a technology open to most teachers,
academics and other content creators, including students using the premier cross-platform language
of that time (the Java language on the desktop). We chose HTML V4, CSS V1 and the JavaScript
language to deliver the content to students using almost any computers available (all those with a
web browser). The HTML mark-up code allowed broad student access via all the major web-
browsers, while the then lesser known CSS V1 was used to place learning objects (any
presentation/interactive technology that could be presented on the screen upon a rectangular sub-
area) within the HTML rendered page, at an absolute location (x,y, width, height) in the browser
window. This choice of underlying technology gave content developers WYSIWYG (what-you-
see-is-what-you-get) delivery, which was deemed particularly important by our designers and
multimedia artists. The JavaScript language was used to retrieve and manipulate server-side stored
content, and create the needed CSS and HTML code to render it as intended.
!
When investigating the iPad much more recently as the target device for personalised, dynamically
configurable content, the Page was once again seen as the fundamental unit of both design and
delivery to the learner. The iBook format – itself based on EPUB3 (V3) – was identified as an
attractive format for the purpose. For example, Adobe InDesign, a page mark-up tool used by many
page design professionals, can output in the EPUB3 format, as can a number of open-source tools,
which can in turn be directly imported into iBookstore on the iPad. A significant goal in the later
Page ! of !5 27

project was to find a good balance between: professionally designed and built learning objects, and
professionally designed page renditions that house those objects; or alternatively, end-user
developer pages and media content, and a smart personalised system that can dynamically select the
particularly learning-objects most applicable for the individual learners' (and their peers') current
circumstance. EPUB3 is basically HTML5 (with a formal index-page structure) – which is a term
generally used interchangeably with three closely integrated standards used in the web browser,
namely: HTML V5 mark-up, CSS V3, and JavaScript (W3C, 2014) – i.e. these are simply later
versions of the same three technologies used to deliver Creator content in the web-browser, 15+
years earlier. That being the case, the design and conceptual data model behind Creator lent itself to
further investigation as an eminently suitable one for storage and retrieval of learning-objects in a
newly envisaged personalised learning environment (PLE).
!
2.2 Description and Conceptual Model of Creator
!
The earlier online education system presented in this section was called Melbourne IT Creator
(Goschnick, 1998). It was the culmination of a research and development project from mid-1997
until mid-1998 in the company Melbourne IT Pty Ltd, then a wholely-owned subsidiary of the
University of Melbourne. Note: The company was floated – the IPO initial public offering - on the
ASX (Australian Stock Exchange) in the year 1999, raising $90 million for the university.
!
Creator was an integrated system that provided an end-to-end solution to authoring, management
and presentation of web-based online learning materials, in the genre of what now is the LMS. It
had three primary functions:
•Content Authoring
•Content Storage and Management
•Content Presentation and Delivery
!
The central concept introduced in Creator was the learning object, of which there are two
categories: simple and complex (composite in nature). Creator also defined a number of views of the
overall system, as seen by users with distinctly different roles, to help users navigate what had
become a large and sophisticated system.
!
Simple Learning Objects
Simple learning objects can be one of the following:
•Plain Text (paragraphs); Tables (HTML <table>); Images; Audio; Video; Interactive
Animation (e.g. Flash, Shockwave); Programmed Component (Javabeans, Java Applet,
Microsoft COM objects); Question (Simple True/False; Multiple Choice; Mutually-
exclusive Choices; Matching (2D matrix of choices); or, Free text entry).
Composite Learning Objects
Composite (also called complex) learning objects can be one of the following:
•Page – a HTML page that may contain numerous and any combination of the simple
learning objects above.
•Reference Work – a hierarchy of sequentially linked Pages.
Page ! of !6 27

•Learning Activity – a complex web of Pages linked in a non-linear sequence (a network
graph), with conditional paths based on student choices and progress.
•Document – an externally authored document in one of various formats including MS/
WORD, Adobe Acrobat (PDF), and RTF.
Figure 1 is a conceptual data model of the learning objects, and how they are inter-related within the
unifying model. This logical model can be easily detailed further into the physical model of the
server-side database, which stores the current state of a Creator-hosted online learning system in a
Relational DBMS server, a type of system that later became known as a Web 2.0 approach (2005).
Fig. 1 – Part 1 of the Conceptual Data Model that stores the Learning Objects in Creator.
!
Note 1: Each of the main entities shaded in yellow in Figure 1 (REFERENCE_WORK, PAGE,
LEARNING_ACTIVITY, QUESTION_OBJECT, KEYWORK/KEYWORK_LIST and
RESOURCE_PALETTE) had a dedicated authoring tool written in the Java language. Figure 4a is
an early mock-up of the interfaces of the authoring tools and the various menus regarding
functionality within them. Figure 4b is an example of the interface of one of the authoring tools: the
Learning Activity Tool.
Note 2: The entities shaded in grey are associative entities (which represent many-to-many
relationships); those in light-blue are less-significant objects, such as icons and images, which do
not need custom authoring tools; while those in white are simply less significant entities in the
model, included for completeness.
Page ! of !7 27

2.3 The Page as Fundamental Unit of Course Delivery
!
Early in the life of the project, a one-week discussion between many stakeholders took place,
regarding the prominent place of objects in the design versus the prominent place of pages in the
design – with effectively two sides of roughly equal numbers of stakeholders, leaning each way. (In
a sense, it was parallel to the self-dialogue in Downes (2005), where he effectively dismisses the
'atoms' that are learning-objects in an LMS, structured together into activities, subjects and courses;
but then he outlines his alternative vision of the web-page (in blogs, etc), similarly aggregated
together into a loose structure, of more personally suited content via RSS aggregation). The
resulting conceptual schema in Creator places both Pages and Objects in the design, with each
holding a significant foundational position in the underlying structure: the OBJECT entity is a so-
called super-type to the simple learning objects, such as IMAGE, VIDEO, HTML_TXT,
HTML_TABLE, and others (shaded in aqua in Figure 1); while PAGE can appear in both
REFERENCE_WORKs and LEARNING_ACTIVITYs.
!
Figure 2 is a sub-section of the model in Figure 1, showing the central structure and the
relationships between OBJECTs and PAGEs. All of the large variation of types of learning-objects
listed above have the common visual quality (apart from Audio) that they may each be represented
on a user's screen in a rectangular sub-area of a larger section of screen - the PAGE. This is a more
obvious choice in the age of the iPad, than it was in 1997/98 as HTML V4 was just becoming
available.
Fig. 2 – Detail from Figure 1 showing the relationship between Objects, Pages and
Reference_Works.
While a PAGE is itself one of the OBJECT_TYPEs, the structure named ON_PAGE_OBJECT (rhs
in Figure 2) represents the flexible relationship between a given PAGE and a number of OBJECTs
that appear on that PAGE. It allows a given OBJECT to appear on any number of authored PAGEs
Page ! of !8 27

in the system, and ultimately allows for personalised pages in a learning trajectory. In data
modelling terminology it is of type associative entity (these are generally shaded in light-grey in
Figures 1,2 and 3). The data attributes within it named: x_origin,y_origin, width, height – hold the
few necessary values that the CSS code needs to place an object on the page, just so, as the PAGE
author intended it (WYSIWYG – what-you-see-is-what-you-get). This placing of objects on a page
by absolute positioning (allowed by CSS) ran contrary to the flowing nature of HTML mark-up,
which in turn tended to upset many page designers using HTML. That dichotomy appears again in
the two types of ebooks allowed in Apple's iBookstore today: the flowing text format, and the fixed-
page format - added in a subsequent version.
!
Note 1: The OBJECT attributes version_no, version_approval_date and editorial_group_id – which
allowed for a formal process regarding the creation and subsequent evolution of all types of
learning-objects in the system. While copyright_owner allowed for attributions right down at the
individual learning object level.
Note 2: The OBJECT_AUTHOR entity allowed for the authorship of learning objects to be either
by an individual or a group of authors.
!
Figure 2 also shows the arrangement of a set of PAGEs into REFERENCE_WORKs. Clearly from
its name, a Reference Work can be seen as an online ebook with interactivity, but what is less
obvious is that the Reference Work may actually be customised down to the Page level, for
individual learners, as we will see further down. That is, the PAGE was designed as the fundamental
unit of course/subject delivery in the Creator system. The entity REF_WORK_OUTLINE is the
structure that holds the data to build the Contents of a Reference Work. The authorship of the Page
(and the complex learning objects) is not restricted to teachers and tutors – it is entirely in the hands
of the administration rights in the Creator system (which are non-hierarchical), such that a PLE of a
sort is possible with little change to the tools.
2.4 The Learning Activity Learning Object
The most complex learning-object in Creator is the Learning Activity. It also interacts with the
Question Object – which appears on the Page as a question that a student answers. A Learning
Activity here is a pre-designed series of PAGEs that a learner encounters, the order of which is
dynamic and generally non-linear, depending on the learner's interaction with it; as opposed to the
sequential movement through a Reference_Work.
!
As with a REFERENCE_WORK the PAGE is the fundamental unit of delivery for a
LEARNING_ACTIVITY. The overall structure of a Learning Activity is that of a network graph
(see Figure 4b for an example of a network graph of Pages within the authoring tool), as opposed to
the hierarchical structure of a Reference Work or an ebook. The entities directly involved in a
Learning Activity are portrayed in Figure 3. A NODE represents a Page in the network graph which
includes a milestone manually set by the author of a particular Learning Activity.
!
The LINK entity represents the arc between two NODEs, the IDs of which are stored in attributes
node1_ID and node2_ID. LINKs can be either one or two-way, at the authors discretion. The
learner can back-step if it is designated a two-way link. The transition of the learner from one
NODE/PAGE to the next, depends on an event stored in the LINK_TRIGGER_EVENT entity. One
such triggering event is a particular QUESTION_CHOICE that a learner has selected within a
Page ! of !9 27

multiple-choice Question, as they negotiate the Learning Activity. This allows for a dynamic path
through the network of PAGEs, based on the learning outcomes of an individual learner. E.g. if
they pick the right answer it goes in one direction, if not, then it goes a different way, possibly a
remedial learning path to make sure they achieve a particular learning milestone. Another trigger
event type is the simple pressing of a particular button with button_ID. In addition to the trigger
event, a number of filtering Conditions can be placed on the Link, for example, that a particular
video was watched in-full by the learner.
Fig. 3 – Part 2 of the Conceptual Data Model that stores the Learning Objects in Creator, but also
student responses as they move through a Learning Activity.
!
An individual student's personal learning trajectory is recorded in the entity
STUDENT_CHOICE_PATH. We take a deeper look at it in Section 4 where we press the concept
and technique into finer-grained personalised learning trajectories, taking the student's current
context into consideration with respect to the types of events that can trigger a particular path
through a Learning Activity.
!
What is not clear in Figure 3 is that the concatenated primary-keys (the attributes above the
dividing-line in an entity box) in the two entities STUDENT_SUBJECT_ACTIVITY and
subsequently, STUDENT_CHOICE_PATH, include several foreign-keys (FK) from entities that
don't appear in Figure 3, namely, person_ID, subject_ID and subject_stream_ID. Those identifiers
come from entities left out of Figure 3 for clarity sake. We revisit them down in Section 4. Enough
to say here that there is a whole other Creator sub-model that deals with the administration of
Page ! of !10 27

subjects, students and subject-streams, some of which appear down in Figure 9 (the white entities).
Fig. 4a – Initial mock-up of the authoring tool screens and menus.
Fig. 4b – The interface of (one of) the actual authoring tool for creating Learning Activities -
clearly showing the network graph structure of such an activity.
Page ! of !11 27

!
What can be gleaned from the details in Figure 3 is that the entity
STUDENT_SUBJECT_ACTIVITY records an individual student's progress within a particular
Learning Activity, via their enrolment in a particular SUBJECT_STREAM, storing the state of that
progress in the appropriate data attributes: start_datetime, finish_datetime, milestone_reached,
activity_mark_tally and comment.
!
Finer-grained detail of a student's navigation through a Learning Activity, node-by-node, is
recorded in the STUDENT_CHOICE_PATH entity, in the data attributes: arrive_time, arrive_event,
arrive_link_ID, arrive_event, leave_time, leave_link_ID, leave_event_ID, and node_mark_tally.
Such data could be data-mined to identify general problems in a given Learning Activity.
!
2.5 User Views and Roles, and the Tools that Service Them
!
User Views and User Roles
!
There are five User Views that facilitate the chunking of functionality in the Creator system, with
respect to locating and using the right tool for the right task. Very briefly, these views are as
follows:
1. Authoring (roles: Teacher/Author/Senior Tutor) – this view brings together the Authoring
Tools, the Search Tools and Version Controls Services to facilitate the creation of subject
and course content, down to the page level.
2. Learning (roles: Student/ Tutor) – this view brings together the Session Control interface,
the Search Tools and the Communication Tools (see immediately below).
3. Discussion (roles: Teacher/Author/Tutor) – The Communication Tools represent the totality
of this view. Note: Melbourne IT Creator used off-the-shelf tools for email, newsgroups and
conferencing – namely Microsoft NetMeeting and Exchange Server, for expediency.
4. Library (roles: Content Administrator/IP Manager) - This view included the Reference
Works accumulated in the system, external documents included in various Learning
Activities, the Search Tools, the Keyword Manager tool, and Version Control and IP
Management services.
5. Administration (roles: Course Administrator/ System Administrator) – Course, Subject and
Student administration tools formed one side of this view, with System Administration and
Database Administration from the OS vendor (Windows Server and Microsoft SQL Server)
forming the other part of this view.
!
As can be glimpsed from these five user views the breadth of application of Melbourne IT Creator
was designed to facilitate up to the level of the content creation and teaching for a whole online
University or School, and hence is a full-blown LMS. However, the granularity of detail goes right
down to individual paragraphs and questions in the content, and down to an individual student's
progress through Learning Activities, including their responses to questions, and page-by-page
trajectories taken through such an activity – suggesting the applicability of those aspects to a
personalised learning system (a PLE). Furthermore, the authoring tools were highly usable, and
available to anyone given the appropriate rights to do so (rights were not hierarchical in their
Page ! of !12 27

application, but were set on a 128-bit access key) – meaning students could themselves be given
access to these tools, on any desktop computer that runs the Java language.
2.6 Authoring Tools
There were five distinct Java-language based authoring applications, that made up the one
integrated software tool, which were:
!
1. Page Designer – an innovative (for the time) WYSIWYG editor for creating pages in
HTML with absolute positioning of learning-objects, using the CSS standard. A Page thus
created, could be used in either: a Reference Work, and/or a Learning Activity.
2. Page Linker – used to create the innovative Learning Activities as seen in Figure 4b and
described above in Section 2.4.
3. Resource Palette – used to amass a collection of learning-objects, both simple and complex,
that may then be easily accessed and drag-and-dropped onto a Page within the Page
Designer.
4. The Outliner – used to create a Reference Work from pre-authored Pages.
5. Keyword Manager – used to populate the keyword-related entities in Figure 1, namely:
KEYWORD, SYNONYM, OBJECT_KEYWORD and KEYWORD_LIST – in turn used in
keyword searches across the internal store of learning-objects. I.e. Within the other
authoring tools, keyword searches can be done by an author to locate appropriate objects in
the overall Creator system, during an authoring session.
2.7 Search and the FAME Agent
!
While the Keyword Manager built indexes against the learning-objects held within a Creator
system, there was a need for external Internet-wide search, particularly with regard to research-
paper citations, often needed by the authors of subject-based learning materials. To cater for this
gap in Creator a research project was undertaken by researchers from the Intelligent Agent Lab
within the Department of Computer Science & Software Engineering at the University of
Melbourne, who had expertise in precisely this area (Cassin & Sterling, 1997; Loke, Davison, &
Sterling, 1996; Goschnick & Cassin, 1997).
!
The intelligent software agent was thereafter known as the Finder and Minder Agent Engine
(FAME Agents). The snippets of information from content authors lead to URLs that were then
used to seed deep web-crawling searches of the site and its links on a regular basis. I won't expand
any further in this paper on the FAME Agent. However, the current field of recommendation
systems use either software agents or other AI techniques in a similar way to FAME usage in
Creator, suggesting many potential applications within PLEs with regard to the personalisation of
learning content, and the alignment of collaborative tasks across individual learners.
!
3 Conceptual Models from a Social Networking Platform (Facebook)
!
While Downes (2010), Mott (2010) and others saw benefits of social networking software within
PLEs, they mainly saw them as linked informally by the student themselves in their daily learning,
Page ! of !13 27

in the form of light-weight mashups. Functionally, that could be as simple as a learner having
multiple windows from multiple apps on the one screen, or using an RSS aggregator to pull
information in from such sites on the web. In more recent times there has been a growing body of
research that directly investigates the use of social software (the terms used vary but have similar or
overlapping meanings: social media, social semantic web, social networks, social software) with
respect to learning, particularly personalised learning: Halimi et al (2014) investigate personalised
recommendations based on similarity using the social semantic web; Leung (2015) investigates the
use of social media against academic performance and perceived social support; Dabbagh &
Kitsantas (2012) investigate the combination of social media and self-regulated learning with
personalised learning environments (PLEs); while Archee (2012) suggests that the lure of Facebook
as an external tool being embraced by educational institutions is 'mistaken' and that the proponents
thereof (those that 'regard Facebook as a pedagogical model' at least) are 'technological
determinists'. None of those studies looks at a specific social network platform from an information
system design point-of-view, to investigate either: how the key features that attract people to social
networking platforms such as Facebook, LinkedIn, Twitter and Google+, might be put into a PLE
technology; or alternatively, how easily the vendor of such a platform could turn it into a PLE,
should the vendor decide to point their respective considerable commercial focus toward the PLE
marketplace.
!
We investigate these two scenarios in Section 4, but before doing so we need to outline the why and
the what of social network platforms, that are of such interest and potential in the creation of PLEs.
!
3.1 The Why of Social Software in a PLE
!
Looking at the functionality of current-day social networking platforms, they generally eclipse the
earlier use of email and newsgroups in the Creator LMS, regarding social functionality. Beyond
communication and collaboration possibilities, the sheer ease that some simple content can be
created and the speed that it can be uploaded and shared in the social media platforms, has set the
precedence regarding usability and immediacy for a generation or two of what learners now expect
in any current or forthcoming PLE.
3.2 The What of Social Software in an advanced Smart PLE
!
While Facebook has a programming API available to third-party developers, they do not publish a
formal or even a conceptual data model overview of what is clearly a structured data system behind
the scenes. Recently, the author inferred a complete conceptual model of the Facebook platform
(Goschnick, 2014), one that fits the data textual descriptions in the publicly available Facebook
Graph API V2.0 (Facebook, 2014). The inferred conceptual model has 85 entities.
!
The first point of note about the Facebook model is the number of entities – the Facebook platform
is far from the light-weight service that the earlier cited PLE enthusiasts (e.g. Mott, 2010; Attwell,
2007) were considering for inclusion in their mashups. Despite its relatively easy to use desktop
interface - and discarding the sheer scale and logistics of its physical model and infrastructure to
cope with upwards of 1.3 billion users - underlying the Facebook platform is a sophisticated and
extensive conceptual data model. As pointed out in (Goschnick, 2014), Facebook has gone to
Page ! of !14 27

considerable lengths to deny an underlying logical structure, instead saying it is simply a realtime
Graph, they call the ‘People Graph, akin to Google’s Search Graph’. And yet they gather
information that fits strict, limited pigeon holes that they themselves define and provide to users,
unlike Google's search engine which must deal effectively with the unstructured or differently
structured data from hundreds of millions of independent content creators. That is, the Facebook
platform has a definite structure, albeit an extensive one of enterprise system proportions.
!
However, Facebook has shown considerable innovation recently in making their enterprise-scale
data model highly usable on smartphones and tablets, demonstrated with their recently released
apps: Facebook Groups (November, 2014) and Messenger (August, 2011 on Android; July, 2014 on
iPad). Each app provides limited but focused slices of functionality from the enterprise-scale model,
in a highly user-friendly manner on every-day devices. This is a usability strategy that many
enterprise-scale platforms could learn from.
!
In this paper, in the next 4 sub-sections we will look in detail at four of the sub-models from the
overall Facebook model - those we perceive as having some value to our cross PLE/Social network
software comparison, in our goal for an improved smart PLE. They are: the part of the model that
deals with Posts (which can give context to and about the learner’s situation); the part that deals
with Groups (which can replicate the sorts of groups represented in learning environments but also
the wider social groups that a learner is a part of); the part that deals with Events (which can chart
future events such as lectures, project meetings, reading group meetings, etc. - and facilitate all sorts
of contextual dialog around them); and the part that deals with Pages - given the central place that
Pages has in both LMS and envisaged PLEs, as discussed in Sections 1 and 2.
3.3 The Facebook Posts Sub-model
!
Goschnick (2014) begins with a general description of social networks, with particular reference to
Facebook, saying that they “are about connecting people and collecting data; they are
communication services upon which further services can be added and customised. The 'people' part
leads to groups (of friends, family and associates), sub-groups and events. The 'communication' part
leads to posts, messages, comments and questions. The 'collecting data' part leads to links, photos,
albums, videos, photo-tags, video-tags and check-ins to particular places (including locations in the
real world)”.
Page ! of !15 27

!
Fig. 5 – Users, Friends and their Posts, taken from (Goschnick, 2014 – Figure 3.1).
!
A User has a set of Friends who each receives the user's Posts and vice-versa. The Posts may be a
simple text message (status update), but it may be accompanied with a Photo, a Video or a Link (to
some web page – often taking a graphic and a headline from that www page). Posts appear in a
user's Newsfeed – which is effectively a default home page, of chronologically listed posts for one’s
social network (depending on various privacy settings). There is a timeline version of the newsfeed,
which is generally just the full list of a User's own Posts, each entry timestamped.
!
This simple mechanism can be supported by the conceptual data model in Figure 5. Note: the
entities Link, Video, and Photo are sub-types of Post - itself a so-called super-type in ER modelling
terminology. The Profile entity is a generalised way of representing a profile in Facebook of more
than just people, which is discussed in the next section. The With_Tags entity (an associative entity)
is a simple mechanism to store the presence of various friends or peers, with the user when they
posted a Post – this is separate from the people 'tagged' in an actual photo which are stored down
the bottom in the Photo_Tags entity. The Media and Interest entities are a simple method of storing
Page ! of !16 27

the various books, music, movies, bands and other things from popular culture, that a user has
flagged as of particular interest to them (Note: there is a more extensive way of flagging interests
via Pages, covered further down in section 3.4). Facebook stores a variety of resolutions of a single
image in the Image entity to deliver a version that is most appropriate (efficiency-wise) to the
current device of the post-receiving user.
!
The Users, Friends and Posts are foundational to Facebook and go back to its early history.
FriendList allows some basic grouping of one’s Friends. The free-flowing newsfeed of content
(which is conceptually and procedurally similar to a dynamically updated report coming from a
database system of underlying data), by way of text, photos, video and links, provides a simple and
quick content publishing system using a basic stylesheet, broadcast to one's social network. People
in that social network may show their support via a Like (a one-button click/touch response), and/or
a Comment, which can expand into an ongoing conversation, involving any one who has access to
that item in the newsfeed.
!
This simple, authoring and communication mechanism, with the immediacy of affirmation and
feedback from friends and peers, is now the norm of a generation or two. The smartphone has
placed access to it wherever the mobile user currently is. Images from the smartphone can be
posted to one's social networks almost instantly and via a user-interface with indiscernible friction.
(Note: the Instagram app is also now owned by Facebook).
!
Video requires somewhat more bandwidth, but the evolving telecommunication infrastructure and
data plans, means that uploading video will soon be nearly as painless as uploading images for a
large percentage of users.
!
That usability and the responsiveness of the communication is now what a PLE needs to use or to
replicate, to meet learner expectations with regard to personalised authoring of content,
broadcasting it, and communicating around it with peers, friends and others. Simple in its use,
complex but efficient in its delivery infrastructure, and powerful regarding the consequences upon
the emotional aspects of learning.
3.4 The Facebook Groups Sub-model
Groups in Facebook are like a series of generalised FriendLists (see entity top-left in Figure 5), but
with extra capabilities and features. For example the link attribute in the Group entity in Figure 6,
holds a url to the Home Page website of the group. The members of a particular Group are all
represented in the associative entity Member. A user may join an existing group or start a new one.
The Group has a record in the Profile entity (together with the user_id of the User who first set-up
or 'owns' the Group). This is a clever mechanism which allows a Group to have a newsfeed of Posts
by members, just like an individual User has a newsfeed of Posts. All posts emanating from the
group via its members, are referenced by records in the associative entity Group_Feed. The users
get alerts when new Posts emanate from another group member.
!
Page ! of !17 27

!
Fig. 6 – Facebook Groups reproduced here from (Goschnick, 2014; Figure 4.1).
External files created by all sorts of external applications can be uploaded to the Group, as recorded
in the Files entity. In addition to those, internal Group_Doc documents can be created within
Facebook (using HTML formatting), and these can be jointly edited by the members of the group.
!
The attribute in the Group entity named parent represents the id of its parent (if it has one), while
parent_profile_type can be either: a Page, an App or another Group. I.e. Groups can recursively
own other Groups, creating a hierarchical structure of groups that is entirely similar to the
newsgroups from earlier Internet times, and to forums in various wikis and most traditional LMS.
!
The Groups feature was only added to Facebook in 2010 and when it was introduced, it quickly
became their most popular feature (Goschnick, 2014). A free smartphone app (named: Facebook
Page ! of !18 27

Groups) dedicated to just those features in Groups was released by Facebook in 2014 - currently
available from both: the Apple App Store for iOS devices, and the Google Play Store for Android
OS devices. The features available through this app replicate most of those available through the
web browser interface for Groups.
3.5 The Facebook Events Sub-model
Events in either the virtual or real world can be organised, tracked and recorded in Facebook. The
What? When? Where? Who's invited? Who is coming? And who can't make it? - are all captured and
broadcast via the features of the Event part of the Facebook platform.
!
Fig. 7 – Conceptual model of Facebook Events taken from (Goschnick, 2014; Figure 8.1).
Furthermore, as with Groups (via the same Profile mechanism), Events have their own newsfeed of
Posts coming from those users posting to the event. The Event entity has a privacy attribute (see
Figure 7) which can be set to either: Invite-Only, Public, or Friend-of-Guests – which are self-
descriptive in terms of who can see and join an Event. As noted earlier, a timetabled lecture, a
project meeting, or a tutorial are each events forward in time - all ideal candidates for Facebook
Events.
Page ! of !19 27

!
Another attribute of the Event entity of interest here is venue_id, which is the id of a Page
(described in the next section), of either the: place, location, organisation or some other entity
hosting the event.
3.6 The Facebook Pages Sub-model
The Page, the fundamental unit of the www was introduced into Facebook as an entity separate
from a User, in April 2010. It became the way that Facebook allowed corporate and other identities
(e.g. celebrities by stage-name), into what had previously been a people-identity-only system –
where everybody was supposedly going by their real names and identities. (Note: that has been a
point of contention over the years, where a number of effectively anonymous trolls have seemed to
continue on in Facebook with impunity).
!
Fig. 8 – Pages, Places and Locations, from (Goschnick, 2014; Figure 9.1).
The Page entity in the conceptual model is a super-type (see Figure 8) with numerous specialised
sub-types across the top of the Figure: Company_Organization, Brand_Product, Local_Business,
Websites_Blogs, Other. Each of the sub-types have a few extra attributes beyond what they have in
common in the Page entity. They also each have a distinct stylesheet, evident when the different
types are displayed in the browser. Facebook has scores of categories and sub-categories of the
Page types, with little or no current variation across many of them wrt the attributes available in the
Graph API, so only five sub-types are currently necessary in the conceptual model. This aspect of
Facebook renders it much like a content management system (CMS), where the default presentation
of a page is provided by carefully designed stylesheets. That also means that authoring such pages
Page ! of !20 27

is inflexible, compared to an LMS or a CMS with WYSIWYG editing tools available to the content
creator, such as Creator covered in Section 2.
!
In 2010 Facebook also added Places (see the Place and Location entities in Figure 8) to their
platform, which have a one-to-one relationship with Page, but only with pages that represent a
place in the real world – such as a store/shop, the head office of an organisation, or a park, and so
on. This was Facebook's foray into location-based services (e.g. attributes in the Place entity
include parking and the opening hours), whereby Facebook users can check-in and check-out of
actual places in the real world (see the Checkin entity between Page and User), either by themselves
or via their friends. Note that Location has a self-relationship such that hierarchies of locations (e.g.
a store in a street, in a suburb, in a city, in a state, in a country) can be represented, in a GIS-fashion
(Geographical Information System). That is, the Page and Place entities can be used to tie in
location-based educational or other facilities related to one’s learning activities.
!
A second location-services inspired feature was added to the Facebook platform in the recent
rendition of the Graph API in April 2014: it is represented in Figure 8 by the three-way Place_Tag
associative entity between Place, Profile and Post. It allows for people other than the user sending
the post, to also be tagged at a Place, at the time the user posted the Post. I.e. Though currently
enacted by people, it technically allows for automatic recorded check-ins by GPS, or via other
sensors as the Internet of Things (IoT) progresses, of people from a user's friendlist, placing them at
the scene, at the time of the post. The privacy issues associated with such an automation are
substantial.
!
Other entities from this model of specific interest to us here, include:
!
Ratings – Ratings and user-stories about a Page can be accumulated via this associative entity.
E.g. a Subject can have a Page, as can a finer-grained Learning Activity.
Banded_Users – People who have offended the owner of a page in some non-resolved way, can be
banded from viewing and interacting with a given page, by the page owner.
Milestone – Milestones there-in are a special type of post that appears in the timeline of a Page, so
that it is visibly flagged in the chronological history of the Page.
Tabs - Is an associative entity between Page and Apps (i.e. Facebook has an App store too),
such that user-selected apps can be installed-on/accessible-from a specific page (via a
tab), making them directly clickable from that Page. Note: this is a bit analogous to the
complex Learning Objects that can be placed on a Page in Creator – though less flexible
in its placement on the page, it is more powerful regarding the broad selection of third-
party apps that might be installed on a Page (although most apps currently in the
Facebook App Center are games, of which there is no educational-games category as
yet).
!
All these features and others could be put to very good use within an advanced PLE.
4 A Merging of Models
There is some recent research regarding how students use existing social networking platforms with
respect to their working patterns within current study routines, alongside the traditional learning
technologies (Arndt & Guercio, 2012), but none we could find regarding how the social networking
Page ! of !21 27

platforms could be enhanced to incorporate such working routines; or conversely, how an LMS or
PLE platform could be modified likewise. There is a paper outlining a planned set of features
researchers would like in a conventional LMS, mirroring something of the features in social
networking platforms, but when it comes to the specifics, there are vague ideas about how such
features could be used to make learning more personalised and adaptive. For example, in Tsolis et al
(2011), in which they propose modifying the Moodle LMS given its adaptability, they outline a
brief plan to: expand the user profile page allowing the user to add tags (keywords) indicating their
interests; have shared bookmarks; have shared galleries of digital content (this last proposal is like
the shared Resource Palette in Creator covered in Section 2, though they make no mention of
learning-objects).
!
In Section 3 we investigated several aspects of the Conceptual Data Model behind the Facebook
platform with an open mind regarding two possible future directions that PLEs and social
networking platforms might advance, regarding personalised learning, namely: what can be gained
from such a case study with respect to folding similar social functionality into a PLE; and
conversely, to gauge how accommodating the current Facebook structural design is, should that
vendor choose to move their market focus in the personalised education direction. Interestingly, we
can examine both those scenarios in the one design exercise: by folding appropriately selected
aspects of the two models (those from Section 2 and those from Section 3) into the one model.
!
Fig. 9 – Merging models from Figures 3, 6 and 7.
Page ! of !22 27

While the concept of posts to one's friendlist was a central early strength of social networking
software, when looking across the models in Figures 6, 7 and 8, four generic concepts stand out:
Groups; Events; Pages; and Places – all of which were added much later to the Facebook platform
(mainly in 2010). The generic nature of the four makes them highly adaptable to other application
areas; not just applications in the small-feature app sense, but applications at the platform level,
such as a personalised PLE, or an adaptive LMS.
!
Looking at Figure 8, the way that a number of Page sub-entities are derived from the generic Page
super-type, is indicative of what can also be done by sub-typing GROUP in service of our aim, e.g.:
Class, Tutorial-Group, Study Group, Year-level, Reading Group, Project Team, Lab Group, and so
on. Similarly, sub-types of EVENT could include: Lecture, Lab Session, Workshop, Demonstration,
Tutorial, Seminar, Research Seminar, and so on. In other words, the current Facebook platform is
already well endowed with generic structures and capabilities that could be specialised into an LMS
of some capability, by the vendor should they choose such a path (note: not an infeasible decision
given the Facebook company founder's large donations to public education in the US: $100M in
New York in 2010, and $150M in the San Francisco Bay area public school system in 2014).
!
Furthermore, if we take a finer-grained view of learning down at a single Learning Activity level as
demonstrated in Creator in Section 2 (i.e. where a Learning Activity which in that system may be as
small as a single page with learning-objects on that page, that in turn may or may not also be a sub-
section of a more traditional Lab Session, a Workshop, a Tutorial and so on), it can be very
personalised in a PLE that incorporates social networking capabilities. As a small illustration,
Figure 9 combines the Learning Activity concepts represented in Figure 3, with the generic entities
from the Facebook Platform model in Figures 6 (Groups) and 7 (Events) – the rest of Figures 6 and
7 are assumed to be part of the merged model, without the need to re-present them and clutter-up
the figure. Alternatively, Groups and Events are also generic entities (they may have different
names) in many LMS and indeed other systems too, such as multi-agent systems, e.g. the ShaMAN
multi-agent meta-model has them as SocialWorlds and Events (Goschnick, Sonenberg & Balbo,
2008).
!
Events are fine-grained via a possible relationship between Event and Student_Subject_Activity. We
have also introduced four sub-entities of Group, namely - Tute_Group, Project_Team, Lab_Group,
Reading_Group – each of which may have extra specific attributes. E.g. tutor_id, mobile_no,
health_n_safety, and current_book respectively as simple examples. Furthermore, Events and
Groups are related via a many-to-many relationship, as one is often related to the other.
The combined structure would enable new features such as:
!
•An automatic posting of the achievement of reaching a milestone in a single learning
activity, to one’s friends and family (those that have the right privacy setting);
•The many-to-many relationship means that some events can involve many groups, and
some groups will be involved in many events – and which-are-which can easily be set and/
or discovered;
•Those people reading the same current_book within and across different reading groups can
be alerted to the possibility of fruitful discussions, if so sort out;
•Those learners that are up to the same milestone can be alerted to their common situation;
Page ! of !23 27

•A course administrator can quickly see if there are any lab groups that have no one that has
yet completed the pre-requisite health-and-safety course – and post them on their mobiles
about the observation (A 'smart' Lab, could check that all by itself – see next paragraph);
•And more ;
!
to be implemented with straight-forward/light-weight procedural coding. And furthermore, if
entities like many of those in Figure 8 was also merged in the combined structure:
!
•A location-based service across a university campus could likewise be interfaced model-
wise, at the Location entity in Figure 8.
•An app from a library of subscribed-to or purchased educational apps, could be placed by
the author (may be a student) at the individual Page level in the PLE, via the Tab entity in
Figure 8.
•Both students and teachers could provide significant feedback at the individual Page level
via the Review entity in Figure 8, any time rather than just at Quality-of-Teaching/Student-
Experience survey time; and so on.
5 Conclusion
!
In Section 4 we highlighted some of the key features evident in a social network platform like
Facebook, that could be included in the blueprint of an advanced PLE platform. We also sketched
out just a few of the new features that could be built upon such a merged model.
!
Furthermore, in demonstrating aspects of that merged model, we have also shown how 'education-
ready' the current Facebook platform is, given the number of generic structures such as Groups and
Events, that have been folded into the Facebook platform in recent times. With relatively superficial
technical modifications at the foundational structures level, and a changed or additional direction in
the marketing focus of that company, Facebook Inc. could enter an emerging PLE marketplace.
!
Either way, an advanced PLE based on such a merged conceptual model would be capable of a
significant degree of personalisation, adaptability and socialisation.
!
Another aspect of the Facebook platform touched upon in Section 3 is their strategic slicing up of
functionality into discrete apps (each simpler than their desktop/web interface), as a way to keep the
usability high and the individual user focused on specific sub-sections or views (i.e. the Facebook
Groups and Messenger apps) of their system. This works, despite the enterprise-system scale of
their overall platform. Both LMSs and future PLEs could also benefit from such an approach,
particularly the LMS given the typical enterprise nature of them.
!
Future Directions
!
What we have not yet investigated is the Apps part of the Facebook conceptual data model, with
regard to a personalised learning focus. That part of the whole platform model in Goschnick (2014)
induced from the app-centric aspects of the Graph API, amounts to 24 app-specific entities and
associative entities – a significant proportion of the overall total of 85. How educational apps are
brought into Personalised Learning Environments is a key interest in our future research with
Page ! of !24 27

respect to learning systems.
!
The future of learning environments lies with the merging of the better aspects of flexible Learning
Management Systems with features popularised in Social Networking platforms, to personalise the
individual learning experience in Personal Learning Environments. I believe we have underlined
such a possibility in this paper by demonstrating how that can be done at the conceptual schema
level.
!
REFERENCES
Arndt, T., & Guercio, A. (2012). Social Networking and E-Learning: Towards a More Effective
Online Learning Experience. GSTF Journal on Computing, 2(2), pp89-94. doi:
10.5176/2010-2283&lowbar;2.2.173
!
Attwell, G. (2007). The Personal Learning Environments – the future of eLearning? eLearning
Papers, 2(1), ISSN 1887-1542.
!
Berners-Lee, T., Hendler, J. & Lassila, O. (2001). The Semantic Web. Scientific America, 284(5), pp
34-43, May 2001.
!
Cassin, A. & Goschnick, S.B. (1998). FAME Agent System Design and Specifications. Internal
Document, Melbourne IT Pty Ltd, University of Melbourne, 44 pages.
!
Cassin, A., & Sterling, L. (1997). IndiansWatcher: Single Purpose Software Agent. In Proceedings
of the Second International Conference on the Practical Application of Intelligent Agents and
Multi-Agent Technology (PAAM 97), London, UK, 21-23 April, pp. 529-538.
!
Chatti, M.A., Jarke, M. and Frosch-Wilke, D. (2007). The future of e-learning: a shift to knowledge
networking and social software. Int. Journal of Knowledge and Learning, 3(4/5), pp. 404-420.
!
Dabbagh, N. & Kitsantas, A. (2012). Personal Learning Environments, social media, and self-
regulated learning: A natural formula for connecting formal and informal learning. Internet and
Higher Education, 15(2012), pp. 3-8. doi:10.106/j.iheduc.2011.06.002
!
Downes, S. (2010). Elements of a Personal Learning Environment. ITK Conference, Keynote
address. Hameenlinna, Finland. Accessed here: http://www.downes.ca/files/Personal%20Learning
%20Environments.pdf
!
Downes, S. (2005). E-learning 2.0. eLearning Mag. October 2005: http://elearnmag.acm.org/
featured.cfm?aid=1104968
Facebook (2014). Facebook's Graph API 2.0. Accessed April 30, 2014 at: https://
developers.facebook.com/docs/reference/api/
!
Felder, R.M. & Silverman, I.K. (1988). Learning and teaching styles in engineering education.
Engineering Education, 78 (7).
Page ! of !25 27

!
Goschnick, S. (2014). Facebook from Five Thousand Feet: A Visual Mapping from Conceptual
Model to Ground-level Graph API Data. EbookDynasty.net, Melbourne, Australia, 128 pages,
ISBN: 978-0-9803595-5-8
!
Goschnick, S., Balbo, S. & Sonenberg, L. (2008). From Task Models to Agent-Oriented Meta-
models, and Back Again. In, Proceedings, TAMODIA-2008 the 7th Workshop on Task Models and
Diagrams, July, Pisa, Italy.
!
Goschnick, S. (1998). Design and Development of Melbourne IT Creator – a System for Authoring
and Management of Online Education. In, Melbourne, 23-27 November, 15 pages.
Goschnick, S.B. & Cassin, A. (1997). Requirements Document – Agents-in-cMILE V1.0, Internal
Document, Melbourne IT Pty Ltd, University of Melbourne.
!
Halimi, K., Seridi-Bouchelaghem, H., Faron-Zucker, C. (2014). An enhanced personal learning
environment using social semantic web technologies. Interactive Learning Environments, 22(2), p.
165-187. doi:10.1080/10494820.2013.788032
!
Leung, L. (2015). A Panel Study on the Effects of Social Media Use and Internet Connectedness on
Academic Performance and Social Support. International Journal of Cyber Behavior, Psychology
and Learning. 5(1), pp. 1-16, Jan-Mar 2015. doi:10.4018/ijcbpl.2015010101
!
Limongelli, C., Sciarrone, F., Temperini, M., & Vaste, G. (2011). The Lecomps5 framework for
personaliszed web-based learning: A teacher's satisfaction perspective. Computers in Human
Behavior 27 (2011), pp. 1301-1320. doi:10.1016/j.chb.2010.07.026
!
Loke, S.W., Davison, & Sterling, L. (1996). CiFi: An Intelligent Agent for Citation Finding on the
World-Wide Web. Technical Report 96/4, Department of Computer Science, the University of
Melbourne, 11 pages.
!
Luckin, R. (2008). The Learner Centric Ecology of Resources: a Framework for using Technology
to Scaffold Learning. Computers & Education 50, pp. 449-462.
!
Mayer C., Linongelli, C., Orlandini, A., & Poggioni, V. (2007). Linear temporal logic as an
executable semantics for planning languages. Journal of Logic, Language and Information, 1 (16).
!
Mott, J. (2010). Envisioning the Post-LMS Era: The Open Learning Network. Educause Review
Online. March 3, 2010. Accessed here: http://www.educause.edu/ero/article/envisioning-post-lms-
era-open-learning-network
Tsolis, D., Christia, P., Kampana, S. & Tsakalidis, A. (2011). OWLearn: An Open Source e-
Learning Platform Supporting Adaptability, Personalization and Mobile Learning. In Proceedings of
the International Conference on Information Technologies (InfoTech-2011), Bulgaria, 15-16 Sept.,
pp. 383-392. !
!
Page ! of !26 27

Valtonen, T., Hacklin, S., Dillon, S., Vesisenaho, M., Kukkonen, J. & Hietanen, A. . (2012).
Perspectives on personal learning environments held by vocational students. Computer &
Education 58 (2012), pp. 732-739. doi:10.1016/j.compedu.2011.09.025
!
W3C (2014). The web standards model – HTML CSS and JavaScript. Accessed here: http://
www.w3.org/wiki/The_web_standards_model_-_HTML_CSS_and_JavaScript
!
Appendix A:
Melbourne IT Creator Project Team at the V1 launch in January, 1998.
Page ! of !27 27