Poster Abstracts - LORNET

kettlecatelbowcornerΤεχνίτη Νοημοσύνη και Ρομποτική

7 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

90 εμφανίσεις

Poster Abstracts

56

Nedjma Meksoud, Monique Noirhomme
-
Fraiture

University of Namur


Approach of retro
-
engineering and engineering to create, share and reuse the educational contents


With the development of distance learning, we

nd more and more on Web and on platforms of

education
various kinds of educational resources ranging from simple traditional courses under electronic form to
different animations, simulations and evaluations. However, it is at present very dif

cult to reuse these
resources. In this poster, we propo
se a methodological approach to create educational contents designed in
order to be shared and reused by different teachers. So we propose a solution that separates a particular

eld’s ontology of concepts and the learning methods that are related to the p
rofessor’s choices..


57

Ty Mey Eap, Marek Hatala, Dragan Gaševic

Theme 1: Simon Fraser University


The Communication Middleware for Collaborative Learning Object Repository Network


Web Service Description Language (WSDL) is designed for well
-
de

ned servi
ces. This requirement is
essential for Web Services (WS) to ensure interoperability throughout a service lifecycle, support service
-

oriented architecture, and provide solid framework for business transactions. However, this constraint
makes WSDL unsuitabl
e for de

ning common interface functions for a collaborative system. This thesis
argues that XML
-
based messaging middleware infrastructure is more effective and suitable for providing
communication protocol for an evolving and collaborative system. The infrastruct
ure can bene

t from the
well
-
established WS framework without depending on the WS technologies. We underpin our claim
through the design and implementation of the eduSource Communication Layer (ECL). We show how
XML
-
based messaging middleware can expand WS

technologies to provide communication framework for
an ongoing developing environment of learning object repository networks. Finally, we con

rm the
advantage of our approach using comparative analysis and use our experiences to write guidelines and
recom
mendations for future development.


58

Nima Kaviani, Dragan Gaševic, Marek Hatala

Theme 1: Simon Fraser University


Uni

cation of Policies with Rules in Semantic Web Services


Simplifying the discovery of web services on one hand and protecting them from m
isuse on the other hand
has initiated several lines of research in the area of policy
-

aware semantic web services. However, the
diversity of approaches, ontologies and languages chosen for de

ning Semantic Web services and policies
has made the research a
rea cluttered. It is now ambiguous how different registries and agents with different
policy languages and Semantic Web service ontologies would share their information. In this paper we try
to solve the problem of exchanging information between the regist
ries by de

ning an interchange
framework to transform business rules and concepts from one language to another using a third
intermediary language called R2ML. The expressivity of the new framework exempts any service provider
or service requester from the dif

culti
es it may encounter during the process of transformation from one
business rule language to the other. It also guarantees that information loss during transformation would be
minimal.


Poster Abstracts

59

David Brokenshire

Theme :1: Simon Fraser University


Causal Network

Overlay of a Self
-
Regulated Learning Ontology


Causal Bayesian networks and Ontologies provide structurally similar representations with different
reasoning possibilities. Work has begun in the literature on integrated these two areas. In this work we
inv
estigate possibility of overlaying a causal Bayesian network on an existing ontology of Self
-
Regulated
Learning theory, from education. Using the structure of the ontology in addition to instance data, we
attempt to use machine learning techniques to learn

the strengths of links in the network or new links in the
network. Eventually we hope to use these techniques to provide educational systems with improved ability
to reason about students individually and in groups.


60

Victor Chen, Cheryl Qian, Robert Wo
odbury

Theme 1: Simon Fraser University


Interactive visual query system in AVIRE


We motivate, describe and demonstrate an interactive visual query system in a generic repository for visual
material (www.avire.ca).When novice users search information, que
rying with Boolean logic may be a
problem. Many English
-
speaking users assume everyday semantics are associated with Boolean operators
when expressed using “AND” and “OR”, rather than their logic equivalents. Visual query system can be a
solution in additi
on to the full text Boolean queries. Based on a compact interactive visualization panel
(NEAR) of AVIRE, we propose a visual interactive querying mechanism to allow people to visually
recognize the union, interaction and difference between two sets. Users
can iterate the interaction to
compose a composite Boolean query. In AVIRE, objects are organized as exhibitions, annotations and
resources. Relationships between objects could be inclusion, been
-
included, co
-
citation and bibliographic
coupling. NEAR panel

visualizes such relations and provides easy navigation path for people. Visual query
component in the NEAR panel provides Boolean search function on top of visualization and navigation.


61

David Botta, Robert Woodbury

Theme 1: Simon Fraser University


De
signing Learning Objects for Multiple Situations


Digital objects used by several post
-
secondary institutions for a 1st
-
year level course introducing design
need to represent at least the minimal requirements of all the stakeholders. Each institution needs

to agree
that the pedagogical plan, design concepts, digital objects, delivery scheme, and management scheme all
are robust and plastic enough to be moldable to the individual needs of its teachers. Our goals and expected
results are (1) to validate or re
fute propositions about design
-
process best practices, (2) discover properties
of group communication exhibited over decisions with the above qualities of interoperability and metadata.
Our approach is to collaboratively design and build such a course, whi
le interjecting some design
-
level
information into the process; conduct a longitudinal survey during the process; and record and analyze the
design protocol.


Poster Abstracts

63

Dario Correal

Theme 2: LICEF TELUQ/UQAM


UNIANDES


Extending e
-
learning Process Models with
Knowledge Management Viewpoints



Process models are fundamental because they guide the activities executed in any organization and
constitute an entrance point to de

ne measurement, control and improvements politics. Nevertheless, to
de

ne and maintain processes models are dif

cult and expensive tasks because it is vital to preserve the
coherence and correctness of the model after each modi

cation. In this work, we pr
opose a mechanism
called process viewpoints to express crosscutting concerns that modi

es, in a modular and non
-
intrusive
way, process models, preventing the pollution of the original process model with external elements. Our
proposal is based on the Aspec
t Oriented Modeling technology and permits to weave the process model
with the viewpoints models to obtain a new executable model. This mechanism is being tested with e
-

learning process models de

ned according to the IMS
-
LD speci

cation, in order to de

ne

viewpoints
related with knowledge management concerns in the pedagogical process.


64

Delia Rogozan

Theme 2: LICEF TELUQ/UQAM


Managing Semantic Annotations and Evolving Ontologies


Our research is part of the LORNET program (Theme 2), which aims to devel
op an e
-
learning and
knowledge management system to support users in searching and using resources that are semantically
annotated based on ontologies that evolve over time. Ontology evolution may have far
-
reaching side effects
since changes such as deleti
ng or merging concepts may hinder the access to or the interpretation of
ontology
-
based annotated resources. However, if there are resources conforming to the “old” ontology, we
must often be able to access these resources and interpret them correctly via
the “new” ontology also. We
propose therefore the SemanticAnnotationModi

er system that supports users to maintain the integrity of
inter
-
linkage between resources and evolving ontologies. In this poster, we will

rstly discuss the effects of
different types of ontology changes. Secondly, we will illustrate how the SemanticAnn
otationModi

er
system may help users to modify the semantic annotation of resources in order to preserve their access and
interpretation via the “new” ontology.


65

Andrew Webster

Theme 3: MADMUC Lab, University of Saskatchewan


Keep
-
Up: A Collaborative Re
commender Feature for E
-
Learning Systems


E
-
Learning systems are big places. The amount of available information is often overwhelming and may
discourage students’ learning activities. We present the design of a collaborative recommender feature,
Keep
-
Up,
that allows students to personalize, monitor and share content, giving them a

exible strategy to
deal with information overload. The central concept of Keep
-
Up is to support and augment word of mouth
communication rather than attempting to completely automate it as traditional recommender systems do.
We recognize learning is a soci
al activity and exploit the similarity associations produced by collaborative

ltering algorithms to make students aware of each other. That is, the implicit social network formed by
shared interest/activity is made visible, allowing students to adjust the
ir ego
-
centric network for each and
every interest/activity. Recommendation is ultimately achieved by seeding information items into these
networks and controlling the spread based on models of information diffusion.


Poster Abstracts

66

Christopher Brooks

Theme 3: ARIES
Lab, University of Saskatchewan


Visualizing Asynchronous Interactions in Educational Environments


Virtually all e
-
learning systems provide some kind of asynchronous discussion tools to support peer
interaction. This interaction is seen by many educationa
l science researchers to be paramount in providing
motivation for students as well as being important in the creation of knowledge (e.g. Social Development
Theory). This poster describes our initial work at developing a customized sociogram visualization t
o help
instructors understand the asynchronous interactions that take place among students in an e
-
learning
system.


67

Tariq Muhammad, Julita Vassileva

Theme 3: MADMUC Lab, University of Saskatchewan


User policies to support inter
-
linking in learning com
munities


Online learning communities can bene

t from communication across their borders. Users in the
communities with interoperable user identity can join new communities without having to start from
scratch. However, any such inter
-
community collaborati
on requires compatible rights of access and
policies. Currently, the user identity in online communities is designed for authentication of users and
veri

cation of their access rights. These identity models focus on exclusion rather than inclusion. There h
as
been a transition in the approach to addressing this problem from proprietary solutions to open standards in
order to enhance such collaborations. We point out opportunities available in interoperable user identities
and access policies for online commu
nity designers, and suggest custom solutions that focus on higher
level functional requirements of online learning communities. Our design of online communities in
Comtella demonstrates how interoperable user identity and access policies can be used for co
llaboration
between learning communities.


68

Dong Liu, Julita Vassileva

Theme 3: MADMUC Lab, University of Saskatchewan


Service
-
Oriented Learning Processes


A learning process is a sequence of activities to achieve certain learning objectives. A learning

organization
should be able to provide

exible combinations of learning objectives and learning processes for personal
requests. Therefore, it is valuable to virtualize the atomic learning activities as services, and to compose
complicate learning processes based on them. The organizations and
institutes will be easy to share their
learning resources in this way. The educational institutes will be more competent in the education market
by supplying

exible courses and training. The Web Services Business Execution Language (WSBPEL)
speci

es a lan
guage for business process behavior based on web services. We can apply WSBPEL to
implementing learning processes. This poster presents how to represent learning processes using WSBPEL,
and discusses the management issues of the approach.


Poster Abstracts

69

Indratmo, Ju
lita Vassileva

Theme 3: ARIES Lab, University of Saskatchewan


Extending Blogs for Knowledge Management Tools


The nature of knowledge management in everyday live is informal, lightweight, and opportunistic.
Knowledge sharing often occurs in lounges and ha
llways, as opposed in planned, formal meetings.
Organizations, however, cannot capture and reuse knowledge exchanged during such informal
conversations, unless there is a persistent representation of the knowledge. In this project, we propose blogs
as a to
ol to capture and facilitate knowledge reuse in organizations. The style of blog entries is similar to
that of knowledge management: informal and conversational. Blogs enable interaction between authors and
their audiences. Furthermore, blogging software p
rovides an easy way to post entries, thereby allowing
users to focus on their writing. Despite its potential, however, the current blogging software lacks certain
functions to support knowledge management in organizational settings. In this poster, we outl
ine our
approach to extending blogs for knowledge management tools, focusing on the management of access
control in blogs.


70

Ben Daniel, Gordon McCalla,Richard Schwier

Theme 3: ARIES Lab, University of Saskatchewan


Mining Soft Data to Understand the Pro
cess of Information and Knowledge Sharing in Virtual
Learning Communities


The process of information and knowledge sharing in virtual communities is in part, an informal social
process that can be characterized as soft and implicit. Even in most formal vi
rtual learning communities
with explicit social protocols and requirements, learners often engage in the exchange of information, and
sharing personal experiences which are not necessarily required part of the formal process. Typically, in
any learning env
ironment, when we think of learners’ information seeking behaviours, it is quite obvious
that we are thinking of different kinds of information sources such as books, documents, web sites,
databases, knowledge repositories and formal course content. Howeve
r, in online learning environments, a
signi

cant component of learners’ information and knowledge needs consist of implicit informal
relationships with peers or others he/she can tap for various kinds of learning needs. Improving the ways in
which people build relationships with each other in virtu
al learning communities is critical to research and
practice in online learning environments. Building on our research over the last

ve years, this particular
work presents social network techniques and qualitative methods to analyze soft data in order to

understand
informal patterns of interactions among learners in virtual learning communities. The main focus of the
research is two fold: to understand the

ow of information and knowledge in a virtual learning community,
and to identify individuals and co
mmunity factors which are critical to supporting knowledge sharing in
virtual learning communities. We think that knowledge drawn from this work can inform the design of
tools and processes to support effective interaction and knowledge sharing in virtual
learning communities.


Poster Abstracts

71

Mohd Anwar, Jim Greer

Theme 3: ARIES Lab, University of Saskatchewan


Trust in Privacy
-
Enhanced E
-
learning


The success or failure of an e
-
learning initiative

whether it be in the form of an online course, an
informal learning
, a blended learning, or the online communities


greatly hinges on privacy and trust. A
disembodied e
-
learning environment allows one to have multiple personae and create multiple electronic
identities (e.g. a man may create a female identity). Through ma
intaining multiple identities (or
pseudoidentities) and restricting their linkage, one could control the conditions for sharing their personal
information and thereby enjoy privacy. On the other hand, trust is an important enabler of effective
communicatio
n among the members (e.g. a learner, an instructor, etc.) of e
-
learning communities, which is
the primary activity of learning. Typically, the knowledge about an identity is used for assessing the
trustworthiness of its progenitor, which raises concerns ov
er privacy. To resolve this mutual exclusiveness
of privacy and trust, a guarantor mediated reputation management together with an identity management
scheme is proposed.


72

Wengang Liu, Gord McCalla

Theme 3: ARIES Lab, University of Saskatchewan


Mining

Patterns from e
-
Learning Environment


Data collected from e
-
Learning system include information about learners' behaviours and learning objects'
contents, and interactive information between learners and systems. We try to use data mining methods to
revea
l latent learning patterns that can be used in various pedagogical purposes such as recommending
helpers, recommending learning objects, and assessing learning behaviours. For example, for purpose of
recommending helpers, we need a pattern such that we can

filter a group of candidates helpers who will
have enough knowledge and willing to provide helps based on specific help requests. In this poster, we will
show a general model to build a patterns repository that can be used to meet different purposes.


73

Dmytro Dyachuk, Ralph Deters

Theme 3: ARIES Lab, University of Saskatchewan


Scheduling of Composite Web Services


Composite Web Services (CWS) aggregate multiple Web Services in one logical unit to accomplish a
complex task (e.g. business process). This a
ggregation is achieved by defining a workflow that orchestrates
the underlying Web Services in a manner consistent with the desired functionality. Since CWS can
aggregate atomic and other CWS they foster the development of service layers and reuse of alrea
dy
existing functionality. An important issue in the deployment of services is their run
-
time performance under
various loads. Due to the complex interactions of the underlying services, a CWS they can exhibit
problematic and often difficult to predict beh
aviours especially in overload situations. This poster focuses
on the use of request scheduling as a means of improving CWS performance in overload situations.
Different scheduling policies are evaluated for the CWS workflow patterns sequence and
splitsync
hronization. In addition the poster presents a novel scheduling policy (ALWKR) that extends
classical LWKR in using workflow topology information.


Poster Abstracts

74

Xin Liu, Ralph Deters

Theme 3: MADMUC Lab, University of Saskatchewan


Caching & Model Web Services


The

Service
-
Oriented Architecture (SOA) is emerging as a new paradigm for building large, open,
looselycoupled systems. Unlike its predecessors (e.g. CORBA), SOA enables organizations to expose their
legacy systems in a way that supports the easy and fast cre
ation of new services. But the widespread use of
SOA raises questions for clients of mobile devices, such as laptops or PDAs, if and how they can
participate in SOA. Mobile devices often rely on wireless networks that are characterized by low
bandwidth and

unreliable connectivity. To overcome these problems, it is necessary to use caching and pre
-
fetching techniques. To adapt to the constantly changing environment and client context, the caching
systems must be able to detect the changes, and exploit knowle
dge about the services and the clients. This
poster proposes a model
-
driven dual caching system for nomadic clients in SOA, which is context
-
aware
by modeling behaviors of the client, the server and the wireless environment.


75

Sha Wang, Jiying Zhao

Theme

5: University of Ottawa


Image quality evaluation based on watermarking


As a practical and novel application of watermarking, this paper presents a digital watermarking
-
based
image quality evaluation method that can accurately estimate image quality in t
erms of the classical
objective metrics, such as PSNR, wPSNR, and Watson JND, without the need for the original image. In
this method, a watermark is embedded into the Discrete Wavelet Transform (DWT) domain of the original
image using a quantization metho
d. Considering {that different images have} different frequency
distributions, the vulnerability of the watermark for {the image} is adjusted using automatic control. After
the auto
-
adjustment, the degradation of the extracted watermark can be used to esti
mate image quality in
terms of the classical metrics with high accuracy. We calculated PSNR, wPSNR, and Watson JND quality
measures for JPEG compressed images and compared the values with those estimated using the
watermarking
-
based approach. We found that

the calculated and estimated measures of quality to be highly
correlated, suggesting that the proposed method can provide accurate measures of image quality under
JPEG compression. {Furthermore, given} the similarity between JPEG and MPEG
-
2, this achievem
ent has
paved the road for the practical and accurate quality evaluation of MPEG
-
2 compressed video. We believe
that this achievement is of great importance to video broadcasting.


75B

Masoud Makrehchi

Theme 4: University of Waterloo


Learning Social Netwo
rks Using Support Vector Classifiers


Learning social network from incomplete relationship data is proposed. Automatic generation of a social
network requires extracting pair
-
wise relations of the individuals. In this research, it is assumed that a small
s
ubset of relations between the individuals is known. With this assumption, the social network extraction is
translated into a text classification problem. The relations of the individuals are modeled by merging their
document vectors and the given relation
s are used as labels of training data. By this transformation, a text
classifier such as SVM is used for learning the unknown relations. It is also explained that there is a link
between the intrinsic sparsity of social networks and class distribution imba
lance of the training data. In
order to re
-
balance the unbalanced training data, a minority class downsampling strategy is employed. The
proposed framework is applied to a true FOAF (Friend Of A Friend) database and evaluated by macro
-
averaged F
-
measure.


Poster Abstracts

76

Huiyan Qi, Jiying Zhao

Theme 5: University of Ottawa


Human Visual System Based Adaptive Digital Image Watermarking


It is known that there is a trade off between the imperceptibility and robustness of a watermark. Trying to
deal with this problem, a n
ew adaptive digital image watermarking method is proposed. The new spatial
masking is built according to the image features such as the brightness, edges, and region activities. With
the same watermarking energy, the quality of watermarked image using adap
tive masking is much better
than the one without using adaptive masking. We analyze that the PSNR cannot distinguish the qualities of
two images with the same watermarking energy. We propose to use the wPSNR to evaluate the image
quality. The watermark is
detected by the key
-
dependent method without knowing any information of the
original image. In addition, we also extend this proposed spatial masking to the DCT domain by searching
the extreme value of the quadratic function subject to the bounds on the va
riables.


77

Anis Masmoudi, Gilbert Paquette, Roger Champagne

Theme 6: LICEF Research Center, Télé
-
université / UQAM.


Reuse of Software Components through a Metadata
-
based Repository


Our work aims to improve reuse of existing software components within t
he LORNET (Learning Object
Repositories’ NETworks) project. In previous work, we suggested a platform independent set of metadata,
called SOCOM (SOftware COmponent Metadata) that describes the main specifications of software
components, and is used to buil
d a software component repository. We are currently designing a set of
services that requests any SOCOM Components Repository. We developed a SOCO ontology which helps
us to semantically reference software components and to study software components aggreg
ation’s
feasibility. We graphically designed the SOCO ontology using the MOT+ OWL editor and we manage
instances with the Protégé editor. Afterwards, we design our distributed software architecture that allows
engineer to reference, to search and to aggreg
ate software components within the LORNET operating
system, TELOS. DIG specifications and DL (Descriptive Logics) reasoners are used to implement search
operations.


78

Isabelle Savard

Theme 6: LICEF TELUQ/UQAM


Cultural Diversity and the Reuse of Learning

Objects


The interoperability and reusability of pedagogical resources bestow multiple opportunities upon
instructional designers. They increase the possibility of improving access to significant quality material for
learners in different cultural context
s. Hence, the importance of indexing culture sensitive Learning Objects
(LO) to optimizetheir availability to as many users as possible. This raises many questions regarding: 1) the
possibilities of integrating cultural dimensions into the learning design
process; 2) the effect this would
have on the tasks of instructional designers. An investigation was conducted at the University of Mauritius
Virtual Centre for Innovative Learning Technologies (VCILT) in the summer of 2006. The poster will
present the res
earch project and the initial data collected from instructional designers regarding:

1) the learning design process;

2) the sharing and reusing of LOs;

3) the cultural factors likely to influence learning.