Design and implementation of an exploratory search system with cloud computing techniques

Chapter 1
Introduction
1.1 Thesis objectives
Medical academic institutions and medical professional societies alike are in-
creasingly required to invest in external overspecialised expertise in order to enrich
their curricula and/or their educational material by developing overspecialized
courses and corresponding educational content. Such content must be easily dis-
covered, retrieved, re-used and thus shared across institutions and among medical
teachers and students (undergraduate and professionals alike). mEducator, a Eu-
ropean project funded under the eContentplus programme, and of which the Uni-
versity of Catania is partner, wants to enable specialized educational content to
be discovered, retrieved, shared and re-used. Anyway, to nd interesting stu that
could be shared and re-used, it’s possible to adopt some techniques which allow
to improve the search system, for example the implementation of an exploratory
search system. Therefore the aim of this M.S. thesis is the design and implemen-
tation of an exploratory search system in mEducator, which could allow to exploit
the base system of \focused search" and to expand then the search towards re-
sults that aren’t completely inherent to the specied criteria, but that could be
of interest, because they are tied through deeper connections (re-purposing, same
2
CHAPTER 1. INTRODUCTION
tags inserted by users, same words in the descriptions, etc.).
1.2 Thesis structure
Chapter 2 deals with all the main and introductory topics, useful to understand
what we will talk about in next chapters: the topics will be mEducator, the
learning object/resource, the exploratory search, the Self Organizing Maps and
the Cloud computing. In chapter 3 I will analyze what was the current state
of mEducator, the website MetaMorphosis and the work that was possible to
do on it in order to add new functionalities concerning exploratory search. The
following chapter, 4, will explain how the work theorized in chapter 3 has been
implemented. There was also space for improvements and adjustments, described
in chapter 5. The two appendices will give further information respectively on the
structure of our internal cloud and on how to install the developed functionalities
on MetaMorphosis.
3
Chapter 2
Main Topics
2.1 mEducator
The new opportunities oered by the Internet and the expansion of informa-
tion and communication technology (ICT) have enabled the explosion of online
education and e-learning. The eContentplus programme has the aim to make
digital content in Europe more accessible, usable and exploitable and the mEdu-
cator project (www.meducator.net), that is funded under it, pursues the scope to
enable specialized medical educational content to be discovered, retrieved, shared
and re-used across European higher academic institutions. This can be obtained
through the implementation and the critical evaluation of existing standards and
reference models in the eld of e-learning.
2.1.1 mEducator Objectives
The principal reason which led to the creation of mEducator is that there is
an abundance of medical educational content available in individual EU academic
institutions, but this is not widely available or easy to discover, retrieve and thus
share across institutions and among medical teachers and students, because of the
lack of standardized content sharing mechanisms. So mEducator wants to:
4
CHAPTER 2. MAIN TOPICS
 identify and collect a critical mass of dierent types of health educational
material such as conventional educational content types (lecture notes, books,
lecture presentations, exam questions, practices, scientic papers, graphs,
images/videos, algorithms and simulators), educational content types unique
in medical education (teaching les, virtual patients, evidence based medicine
forms, objective standard clinical examinations, clinical guidelines, anatom-
ical atlases, electronic traces of images, etc), alternative educational con-
tent types (active learning techniques and problem/case based learning ses-
sions via web 2.0 technologies, serious games (2D/3D), web traces, wikis,
blogs/discussion forums, etc.).
 analyze how existing standards for description of educational material (like
Healthcare LOM) and other standards (like SCORM) can be adequate, the
rst ones to address all types of health educational material listed above,
the second ones to support the packaging and seamless delivery of all these
types of educational material.
 examine possible extensions of existing ontological schemata, which describe
the semantics of Learning Objects (e.g. s-LOM ontology) and which enable
the contents to be shared and re-purposed.
 cooperate with standardization bodies like MedBiquitous Europe, IEEE,
IMS, CEN, Health On the Net, HL7, etc to ensure adoption of recommen-
dations/extensions into standards.
In order to derive best practices for medical educational content re-use and shar-
ing, mEducator needs a critical mass of medical educational content types (rather
than items), representing various educational approaches (e.g. conventional teach-
5
CHAPTER 2. MAIN TOPICS
ing, active learning, e-learning and blended learning, etc), various audiences, var-
ious languages and various cultures. mEducator is going to seek the best practice
by comparing two solutions:
1. exchanging content via \mash-up" technology and Web 2.0 tools for loosely
coupled isolated LCMSs
2. exchanging content via Semantic Web Services technologies for federated
LCMSs
The target audience of mEducator is formed by medical educators (clinical/non
clinical, in academia), medical students (under- and post-graduates), residents
and specialized doctors (continuing medical education) and one of the aims is to
invite all this audience to evaluate the outcomes.
2.1.2 mEducator current implementation
mEducator is surely a very ambitious project, so it’s not easy to implement
all its parts in a brief time. However, work is in progress and things are going on
in the best way. Right now, a particular eort is spent on the metadata standard,
in order to evaluate which elds are useful and which extensions could be added
to assure the achievement of the prexed objectives of accessibility, usability and
re-purposing.
A new website for the project has been launched, it’s called MetaMorphosis
and it’s available at http://metamorphosis.med.duth.gr with the aim (already
partially reached) to create a social environment to share educational resources
in health sciences. In chapter 3 there is a wide description of MetaMorphosis.
6
CHAPTER 2. MAIN TOPICS
2.2 What is a learning object?
We are dealing with learning objects and re-purposing, so an explanation of
these terms is necessary.
The learning object (LO) is in few words a digital resource which can be
reused to support learning. Many denitions have been given but, as David Wi-
ley observes, \the proliferation of denitions for the term ’learning object’ makes
communication confusing and dicult" [4], so I limit myself to explain what it’s
used for. To explain what a learning object is we have to think to the littlest
resource used to achieve a formative process, like a slide content, a video, a test,
etc. All these parts can be reused in dierent contexts and aggregated to create
on-line courses.
The idea generates from a suggestion by Reigeluth and Nelson (1997), who
take as example teachers, that often, when they use for the rst time learning
resources, they have to decompose them in many pieces and then recompose and
adapt them to their teaching needs; if teachers received learning resources as single
components, this initial phase of \decomposition" could be avoided, augmenting
so speed and eciency.
So the LO’s principal features are:
 self-contained: each LO can be taken independently, in that it represents a
minimal unit formed by one or more assets (an image, a video, etc.)
 reusable: it can be used in dierent elds
 modular: LOs can be grouped into larger collections of content
 tagged with metadata: each LO is tagged with metadata, that are descrip-
tive information allowing the LOs to be easily retrieved by a search
7
CHAPTER 2. MAIN TOPICS
 interoperable: they can work on dierent platforms that erogate learning
resources (LMS), thanks to the attention to standards (SCORM) that dene
the rules to pack and use the LOs
 granular: to make the LO modular, its dimensions must be adequate. Right
now there aren’t strict rules on the dimensions, even though many indica-
tions have been given, for example by Cisco, which establishes in 7 (+/-2)
the number of concepts to put in each LO.
The expansion of new information and communication technologies led to the
growth of the e-learning approach and since one of its important aims is to save
time and money in the design and development of new learning resources, it
conjugates well with the LO approach.
2.2.1 Metadata standards
Where and how is it possible to nd learning objects? To answer to this
question, some on-line repositories containing them have been created (Merlot,
Unitexas, Celebrate, FreeLOms, etc.). However this approach is not sucient to
reach good results, since the search without the use of semantics could generate a
lot of documents in which we aren’t interested. So it’s necessary to dene common
rules to classify each LO and to make users nd the best ones for what they are
searching. So metadata are used to describe datas, classifying LOs by following
some rules, as a book in a library has a record in the card catalog. Right now
there isn’t a nal standard for metadata, but there are some proposals that go
in that direction. Perhaps the most important is IEEE LOM (Learning Object
Metadata), approved by IEEE (Institute of Electrical and Electronics Engineers)
in June 2002, which xes the minimal set of properties necessary to handle, al-
8
CHAPTER 2. MAIN TOPICS
locate and evaluate LOs. In particular, there are nine descriptive areas, which
contain group of attributes:
1. General - 11 elds: general information on learning objects;
2. Lifecycle - 6 elds: information on the life of the LO (version, creation
data, etc.);
3. Meta-metadata - 9 elds: information on the adopted metadata schema;
4. Technical - 12 elds: technical features of the LO;
5. Educational - 11 elds: educational features of the LO;
6. Rights - 3 elds: copyright and similar information;
7. Relation - 7 elds: indications on the relations between the LO and the
others;
8. Annotation - 3 elds: comments on the didactical use of the LO;
9. Classication - 8 elds: information on the subject of the LO.
Another important proposal to create a metadata standard has been given by
DCMI (Dublin Core Metadata Initiative), and there are also other minor pro-
posals, usually related to specic elds. Both of these standards have seen wide
uptake since their establishment, together with a growing body of both practical
implementation experience and applied research into their application. National
and international standards bodies involved in educational technology have main-
tained interest in supporting them, and various communities of interest and or-
ganizations have attempted to achieve some measure of interoperability through
9
CHAPTER 2. MAIN TOPICS
Figure 2.1: A schematic representation of the LOM data model - Source (Wikipedia)
development of a number of application proles (CanCore, UK LOM, etc.): these
are usually recommendations on how to interpret the often too many elds and
on which elds are mandatory and which ones are instead optional; they have
helped a lot a progress of these standards that is in the same time an approach
between them, but the risk in these cases is to complicate and confuse even more
the situation, so it’s necessary to pay attention. To help nding a way in this
maze, in the last years some documents came out (for example [6]), with the aim
to analyze all the standards elds and give guidelines on how a merge could be
reached.
In particular, mEducator decided at rst to use IEEE LOM as metadata stan-
dard, then it was decided to turn to a lightweight schema with reuse of ontologies
and vocabularies of the semantic web. Right now an application prole is under
study, because the theme \medicine", like other themes, requires that a right dis-
tinction is done among the elds and that the decision to ll or not a eld is well
weighted.
10