Transcript Theme 3 presentation

Theme 3: Active and Adaptive Learning Objects
“Influenced by and Influencing Social Computing”
Jim Greer, Gordon McCalla, Ralph Deters, Julita Vassileva
Department of Computer Science
University of
Saskatchewan
Introduction
University of
Saskatchewan
Introduction 3
Specific projects:
▪ Jim Greer 4
▪ Julita Vassileva 4
▪ Ralph Deters 7
▪ Gord McCalla 8
Conclusions 2
 Why social computing?
 Our deployed learning environments
have convinced us that there is an
increasing social dynamic to be captured
 This dynamic has two sides relevant to
educational technology research:
 It’s important. Learners collaborate just-intime all of the time, and expect nothing less.
Access to email, chat, and instant messaging
within a learning environment has changed
the ways learners do this online.
 We can capture it. Learners are turning
increasingly to technology to engage in their
learning activities, and we have the option to
be in the thick of it all.
Social Computing in E-Learning
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 We teach/learn in a [usage] data
saturated environment
 The tools in iHelp capture this attention
metadata
 iHelp Courses, a standards-based research
LCMS
 iHelp Discussions, an asynchronous forum
system
 iHelp Chat, a synchronous forum system
 iHelp Share, online collaborative code
annotation groupware (demo at poster
session)
 We aim to capture fine grained attention
metadata
 Who reads what? [post/object/chat]
 How long do they read it?
Making Sense of Data
 Of course, usage data isn’t the only data
of interest
 Content data
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 We are working with theme 1 (SFU)
theme 4 (Waterloo) to dig into this data
a bit more
 Can ontologies help to organize and provide
deductive reasoning over our collected data?
 Can ontologies provide a bridge between real
usage data and learning standards (e.g. IMS
LD)?
 How do user inputs (e.g. collaborative
tagging) compare to automatic metadata
generation?
 Can content metadata be leveraged alongside
content interaction metadata?
Awareness and Assistance in iHelp
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions






Are my friends around?
Who is doing what?
Where do I stand?
Who is willing and able to help?
Is an instructor available?
What resources might fit my needs right
now?
 Who is at risk?
 How healthy is the learning
environment?
 What kinds of interactions are occurring?
Specific Projects - Jim Greer
 Christopher Brooks
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Basic Approach
 Real systems, real learners
 Large scale deployments
 Collaboration in a safe environment
 Building respect for privacy
 Enabling and utilizing publicity
iHelp Share
 The iHelp project is still ongoing
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Collaborative document annotation
 Programmer help
 Writing help
 Augmented by chat and discussion or voice
 Why not collaborative editing?
 Research opportunities
 Willingness to collaborate
 Tutor training
 Demo by Stephen Damm, student
Privacy in e-Learning
 Virtual learning communities may not be
a circle of close friends
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 How to protect privacy
 Add pseudonymity
 Building trust through reputation
 But without full identity
 Reliable sharing of reputation data
 How?
 What about fusing partial reputations?
 What about transfer of reputation?
 Poster by Mohd Anwar, PhD student
E-Portfolios to Learner Models
 Learner model is a detailed cognitive
representation of a learner
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Can an e-portfolio initialize a learner
model?
 What information can be automatically
extracted?
 How can evidence be used to support claims
about cognitive abilities?
 The process of “evidencing”
 Reflection has its benefits
 User study
 Poster by Zinan Guo, MSc student
Specific Projects – Julita Vassileva

Comtella – a community for sharing
 Participation is the key problem!
 Previous (now deployed) approaches:
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Incentive mechanisms: rewarding participation
through social visibility, status and privileges
 Successful, but do not necessarily help learning
(students “optimize” their participation to yield the
rewards)
 New approaches:
 Making the system immediately useful –
embedding sharing into Personal Information
Management (PIM) – in blogs
 Exploiting/Fostering interpersonal relationships to
generate recommendations of RSS
 Bridging communities – in this way even small
communities can reach a “critical mass” since the
community doesn’t need to provide all the services
and users don’t need to start from scratch
Collaborating through blogs
 Sharing? Why?
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Need to be useful for self first, then to others
 Sharing by default? Personal info management
 Need to be convenient, manage access seamlessly
 Blogs – personal info space
 Currently – open for everyone to see (like a
homepage), e.g. MySpace, LifeJournal
 Managing access rights – very much needed
 Who sees what? Delegating access rights to groups.
 Collaborating – allowing others to modify blog
 Prototype – a blog system allowing users to
manage access rights to their blogs
 Special language: user groups, access rights
packages (roles), item groups (rooms)
 Usability evaluation
 See Indratmo’s poster (PhD student)
Social networks for recommending content
 Current recommender systems:
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Content based, Collaborative and Hybrid
 Collaborative recommender systems use data about past
user actions (rating, buying), correlates it and finds
users who have liked similar things in the past 
recommendations
 However, recommendations are faceless “people who in
the past have bought similar things like you bought this
item”.
 Information spreads using social networks
 Diseases spread also using social networks
 Open model of the relationships of influence
between users,




show it to users,
allow users to add /remove people of influence they wish
use these relationships to recommend content
applied to recommend RSS
 Evaluation: outperforms classic collaborative
filtering even in a static database
 Applicable also for recommending new items
 See Andrew Webster’s Poster (MSc student)
Bridging online communities
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Currently, online communities are “islands”.
 Can we enable users to seamlessly jump across
communities, without abandoning their old
communities?
 Three problems:
 Identity management across communities
 Translation of user data across communities
 Negotiation of policies across communities
 Exploring solutions in the Comtella system
 Mutli-community, multi-node framework
 Different user roles, rights and priviledges
 Communities and nodes are autonomous, with own
policies.
 Decentralized user modelling
 See Tariq Muhammad’s poster (MSc student)
Scalability & Mobility – Ralph Deters
 How to enable scalable solutions?
University of
Saskatchewan
 Open, agile, manageable, etc… with great
performance
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 How to support users of mobile devices?
 Support the mobile learner, anytime,
anywhere
Scalability & Manageability
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
 Delivering/Accessing Learning Objects
via Web Services
 SOAP is Expensive
 How to handle large volumes of requests?
Conclusions
 Scheduling of Requests
 Dmytro Dyachuk’s Poster
Scalability & Manageability
 Defining Learning Workflows
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions




Use variety of accessible Learning Objects
How to manage instances of workflows?
How to ensure SLA?
……
 Management of Workflows
 Dong Liu’s Poster
Scalability & Manageability
 Accessing Learning Objects
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions




What happens if some LO are not accessible?
How to use redundancy?
How to ensure more reliable access?
P2P?
 Integration of P2P into Web Services
 Self-organizing
 Dynamic discovery
 Weidong Han - Work completed
Supporting Mobile Learners
 Enabling access without stable networks!
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
Weak connectivity
(Low-bandwidth and
intermittent
connection)
Laptop
Laptop
Strong Connectivity
(High-bandwidth
and reliable
connection)
Laptop
Null Connectivity
(disconnection)
Challenges
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
Wireless Network
Supporting Mobile Learners
 Using a cache to overcome signal loss!
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 How to cache





What to cache?
How to cache?
When to cache?
Location of cache?
…..
 Model-Driven Caching
 Xin Liu’s Poster
Specific Projects: Gord McCalla
 Basic philosophy
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 fragmented learning environments: just
in time learning, mediated by each individual
learner’s various virtual communities
 active learner modelling: model only what
is needed for a particular pedagogical purpose
 ecological approach: each learning object
in a learning object repository has attached to
it all the information known about each
learner who interacted with it and what the
interactions were at a fine-grained level;
these learner model instances can be mined
for interesting pedagogical insight
Research Paper Recommender
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Tiffany Tang, Ph.D., expected 2007
 capturing pedagogical features of
research papers in order to recommend
them to students who are learning about
a research area
 matching these pedagogical features to
models of learners to determine which
papers are appropriate for which
learners
 ties in to ecological approach: can we
capture information about learners’
actual interactions with the learning
material in order to make better
recommendations?
Data Mining of Learner Interactions
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Wengang Liu, M.Sc., expected 2007
 huge amount of interaction data in iHelp
and iHelp courses
 are there patterns in these data?
 one approach: bottom-up from data
trying to find pedagogically useful
patterns, using data mining and
clustering algorithms
 current approach: define pedagogically
interesting aspects of the learner and try
to build metrics to measure these
aspects
 ties in to ecological approach
 see poster at this conference
Building Usable Metadata
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Scott Bateman, M.Sc., expected 2007
 goal is to make the tagging by humans
of learning objects more flexible and
more useful
 one approach: social tagging by the
learners, implemented in OATS system
 another approach: use WordNet as a
closed ontology from which learners
(and teachers) select metadata
vocabulary, implemented in
CommonFolks system
 look for OATS demo, talk on
CommonFolks, and poster
OATS Screen
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
Purpose-Based Open Learner
Modelling
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Collene Hansen, M.Sc., expected 2007
 goal is to open the learner model to the
learner, the teacher, or to other learners
when appropriate
 for various pedagogical purposes active
models of learner(s) can be computed
and displayed appropriately
 can be very informative to learners and
teachers
 system built and now being tested in
courses at U. of S.
 see example, next slide
An Example Purpose and Visualization
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
Enhancing Social Capital
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Julita Vassileva
▪ Ralph Deters
▪ Gord McCalla
Conclusions
 Ben Daniel, Ph.D., expected 2007
 goal is to understand what affects social
capital in virtual learning communities
and in distributed communities of
practice
 many empirical studies carried out,
seeking variables and studying their
affect
 modelling of variable interactions with
Bayes-Nets
 see paper at this conference
The Broader Picture
 In addition to in-lab projects, we are working
with industrial and other partners
University of
Saskatchewan
Introduction
Specific projects:
▪ Jim Greer
▪ Ralph Deters
▪ Julita Vassileva
▪ Gord McCalla
Conclusions
 Parchoma Consulting: Dissemination of the state
of the art in learning object practice. Developed
and deployed content for the Canadian Association
of Prior Learning Assessment (CAPLA), using iHelp
as a basis
 Desire2Learn: Initial meetings on technology
integration, focusing on issues in and around data
mining
 Technology Enhanced Learning: Cooperating
with university endeavours to realise iHelp systems
in a larger scenario, and bringing the concepts of
sociability into the online classroom
 TR Labs: Working with theme 3 on many of the
systems issues that crucially affect performance of
e-learning systems
Conclusions
 The educational technology domain could be a
role model for new methods in psychological and
social sciences research
University of
Saskatchewan
Introduction
Specific projects:
▪ Julita Vassileva
▪ Ralph Deters
▪ Jim Greer
▪ Gord McCalla
Conclusions
 Learning is necessarily situated in the real world –
small experiments and “controlled studies” have
limited impact
 E-learning provides environments that are both
saturated in data about learner interactions and
also about which we can know much about learner
purposes and goals: implies we can carry out fine
grained studies in the real world
 This kind of education research may be a
prototype for fine-grained studies of people in
various kinds of social situations, not just in
learning contexts
 Happy to answer any questions!