Here i present the workflow which automatically model informal learning communities. For this purpose we created a stereotype model that is based on community of practice theory and a number of models that can specify different cases in communities, e.g. question-answer community. In the presentation I focused on presenting the opensource community modeling service that allows to create models using RESTful queries and the analysis service that performs a community analytics and detect communities, their evolution, investigates texts to define community learners' emotions and attitudes. In results you can see two models of one community where based on models you can understand community insights, problems, issues and goals. The work was evaluated by i* experts and can be further extended by overlapping community detection, further analysis of texts and other interesting directions, We call to collaborations using our opensource services for modeling and visualizing learning communities.
MULTIDISCIPLINRY NATURE OF THE ENVIRONMENTAL STUDIES.pptx
On Modeling Learning Communities
1. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-1
TeLLNet
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
Zinayida Petrushyna, Ralf Klamma and Milos Kravcik
EC-TEL 2015,
Toledo,
September 16, 2015
On Modeling Learning Communities
2. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-2
TeLLNet
Outline
Motivation and Research Question
Data Collection
Methods
Results
Evaluation
Future Work and Discussions
3. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-3
TeLLNet
Informal learning communities need digital media to realize
social context
Motivation
Motivation
Methodology
Results
Evaluation
Outlook
But online community data is huge and
heterogeneous Verbert et. al., 2012
Is learning analytics a solution?
Problem Learning Analytics Community Modeling
Define learner roles + +
Predict learner success + +
Classify a type of a community +/- +
Estimate community needs +/- +
Find a solution to a problem +/- +
Forecast community changes +/- +
4. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-4
TeLLNet
Modeling Informal Online Learning Communities
An efficient data management solution
A service for learning community analysis
Community borders and roles of users
Kleanthous & Dimitrova, 2007, 2010
Learning community/user goals, attitudes, interests
Modeling
Refinement
Monitoring
Analysis
Petrushyna et al. 2014
Motivation
Methodology
Results
Evaluation
Outlook
A service for automatic modeling of communities
A community of practice is a stereotype model Wenger, 1998
Other models from a model repository Petrushyna et al. 2010
Call to stakeholders‘ action
5. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-5
TeLLNet
• # posts ≈ 429K ; # users ≈ 21K; # threads ≈ 68K
• Max depth(thread)=318 posts, Avg depth=6 posts
• 360 characters in avg in a post
• 132 communities with > 3 members
Research Example
TOEFL
GMAT
GRE
Tests Informal learning communities in URCH subforums
Motivation
Methodology
Results
Evaluation
Outlook
6. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-6
TeLLNet
Community detection
– Define time intervals based on events of communities
𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑗𝑗 = 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑗𝑗, 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑗𝑗 where 𝑗𝑗 ∈ 𝐽𝐽, where J is a
set of events, e.g., exams
– Modularity-based community detection
Newman and Girvan, 2004
Community Detection and Evolution
Motivation
Methodology
Results
Evaluation
Outlook
Community evolution Palla et al. 2007
Mapping of communities using modified
Jaccard index
𝑆𝑆𝑆𝑆 𝑆𝑆 𝐶𝐶𝑖𝑖𝑗𝑗
, 𝐶𝐶𝑟𝑟𝑘𝑘
= max
𝐶𝐶𝑖𝑖𝑗𝑗
⋂𝐶𝐶𝑟𝑟 𝑘𝑘
𝐶𝐶𝑖𝑖𝑗𝑗
,
𝐶𝐶𝑖𝑖𝑗𝑗
⋂𝐶𝐶𝑟𝑟 𝑘𝑘
𝐶𝐶𝑟𝑟 𝑘𝑘
≥ 𝑡𝑡ℎ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟
Gliwa et al. 2012
Social network analysis measures Wassermann and Faust, 1994
– Detecting patterns using closeness, betweenness
– Patterns: questioners, answering persons, newbies and usual users
7. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-7
TeLLNet
i* Modeling Approach for Informal Learning
Community Modeling
Dependency
resource
Goal
Softgoal
Task
Agent Role
Depender
Agent
Dependee
Agent
Learning
resource
Learning
goal
Acceptance
Support
learning
process
Learner A Expert
Community Learner
Motivation
Methodology
Results
Evaluation
Outlook
+ point out dependencies
between human and non-
human agents
+ models can be created
using XML-based format
+ models can be extended
to describe the rationale
of agents
+ emphasize agents,
their types and roles
+ indicate intentions
in social networks
8. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-8
TeLLNet
i*-REST Opensource RESTful WebServices
Model creation
− Strategic Dependency i*
− API related to the iStarML
− Models are resources (REST)
− Model validation
− Storage and versioning in model repository
Model visualization
− From iStarML to SVG
− Easy to embed into a Web page
− JS extension allows user interactions
Wanted to
solve
equation QuestionerThe
community
with 44 users
User_id=69588
Usual user
User_id=69561
Forget to
write
debrief
Newbie
User_id=69575How to
solve
Goal
Role
Depender
Agent
Dependee
Agent
Motivation
Methodology
Results
Evaluation
Outlook
9. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-9
TeLLNet
Modeling of an URCH
Learning Community Evolution
01-10.12.2004 08-17.12.2004
# posts = 471
# users = 22
# adjacent nodes = 43
# high influence users = 13
# low influence users = 2
need to learn
want to write
take to solve
started to take practice
prepared to take beast
trying to learn stuff
# posts = 226
# users = 20
# adjacent nodes = 15
# high influence users = 4
# low influence users = 4
how to answer
instructed to take writing
supposed to answerplan to take GRE
take to solve
Motivation
Methodology
Results
Evaluation
Outlook
10. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-10
TeLLNet
Evaluation of Automatic Creation Process
of i* Models
i* experts evaluated the process
25 subjects, 86% agree thatMotivation
Methodology
Results
Evaluation
Outlook
– community stakeholders can
understand community situations
better using i* models
– emphasizing community
requirements facilitate developers’
work of community information
systems
Social
Network
Analysis
Community
Detection
and
Evolution
Intent
Analysis
Named
Entities
Retrieval
i* models can be abstract and not
straightforward
Training is required before stakeholders
can use models
11. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-11
TeLLNet
Conclusion and Outlook
Informal online learning community modeling workflow
Integration of community analysis and modeling for
automatic creation of models
Future Work
Near-real time collaborative modeling
Derntl et al. 2013, Nicolaescu et al. 2013
Further analysis of communities texts and learner roles
Better usability for stakeholders
Overlapping community detection Shahriari et al. 2015
Application on MOOCs
Motivation
Methodology
Results
Evaluation
Outlook
12. Lehrstuhl Informatik 5
(Information Systems)
Prof. Dr. M. Jarke
I5-ZP-915-12
TeLLNet
Call to Collaborations
Extending and applicating services for modeling
learning communities
Search on github
iStarMLModel-Service
http://tinyurl.com/modelingService
Visualizing service iStarMLVisualizer-Service
http://tinyurl.com/visualizingService