Self-Organized Learning

Networks for Lifelong LearningRTD Programme 2003-2008

Rob Koper, Peter Sloep, Colin Tattersall, Peter van Rosmalen

Educational Technology Expertise CentreOpen University of the Netherlands

www.learningnetworks.org

Hannover, November, 24th 2003

Overview

Introduction to the Programme

Rob Koper

Semantic Representation of Nodes (IMS LD)

Rob Koper

Agent technologies to support teaching functions in learning networks

Peter Sloep

Navigation in Learning Networks

Colin Tattersall

AlfaNet (use of LD in a context of agent technologies and collaborative learning)

Peter van Rosmalen

Introduction

Open University of the Netherlands

Started in 1984; National (Public) Institute Two missions:

1. provide open distance education and 2. innovate education (in general)

Open distance education

6 faculties, 23000 students (age avg 32), 9 bachelor/master programmes; students can make a

free selection of courses during their life 20 study centres in Netherlands and Belgium Develop self study materials in multidisciplinary

teams Deliver education through a variety of technologies

(print, cd-rom/DVD, telephone, internet, face to face contact sessions, practicals, etc.)

Innovation

Educational Technology Expertise Center RTD programme into Learning Technologies RTD programmes into LT

1998-2002: Educational Modeling (EML, IMS LD, Edubox)2003-2008: Learning Networks for LifeLong Learning

Positioning:a. Major expertise ‘educational technology’ b. Major focus is: innovation through development of new learning technologiesc. Between Educational Science and ICT technologiesd. Bring in educational requirements that are specific enough to be implemented in ICT environments

Learning Networks Programme

2003-2008

Basic activities

1. RTD projects in several themes

2. Standardization activities

3. International Expert groups

4. EU projects and national RTD projects

Objective of ProgrammeDevelop a coherent set of learning technology models, specifications & tools to establish a new effective, efficient, attractive and accessible approach for higher, distributed lifelong learning, called learning networks.

Network in the interpretation of:

1. Network of interacting persons and resources: heterogeneous lifelong learners, experts, tutors, learning resources and tools in some knowledge domain

2. Network of interacting distributed devices (e.g. computers, mobiles, …)

3. Network of interacting providers for lifelong learning resources and services (institutions, libraries, publishers, associations, companies, …)

Programme addresses two key issues

1. Establish the emergence of lifelong learning into a distributed, heterogeneous network of learners, providers and software agents

2. Help staff members to do their work more effective and efficient (minimize staff work load)

Issue 1: Lifelong Learning

Some general questions: How do lifelong learners learn? What do

we know of the learning behaviours and preferences of persons during their lifetime and career

How can we support lifelong learners with new learning technologies?

Issue 2: Efficiency of SupportBasic Question.

How can we:

make learners more productive, responsible, adapt to prior knowledge, provide freedom of navigation (learner),

produce high quality learning resources (knowledge) provide more formative feedback on the productions

(assessment) can involve more experts and practitioners, handle

heterogeneity in groups (community) …

without increasing (or better: decreasing) the workload for the staff members involved.

Main instruments in programme

Models, principles and rules to establish self-organized, distributed lifelong learning

agent technologies (in context of semantic web) to support the actors in the learning process (learners, tutors/experts, developers) and

interoperability specifications and standards (e.g. for portable learner dossiers, competencies, architectures, etc.)

Main programme themes

1. Development and use of Activity Nodes How to design, create, share, use units of learning in the Learning Network

2. Learner Positioning in Learning NetworksHow to position new and existing learners in a Learning Network independent of curriculum or institution

3. Navigation in Learning NetworksHow to navigate in Learning Networks, using & exchanging recorded learning tracks, learning routes and learning patterns in Learning Networks

IMS Learning Design

In Short New standard from IMS (februari 2003)

www.imsglobal.org Based on our previous work on EML (Educational

Modelling Language; published december 2000) Objective is to model complete Units of Learning

that can be transferred to different systems and contain the compete description of its designed content and process.

Provides an integrated framework for different other IMS specifications (incl. LOM, QTI, LIP, CP, RCD, SS)

What is a Learning Design?

The learning design specifies the specific workflow and content in the learning process:

which role has to performs which activities, using which resources and services in which order in order to attain the learning objectives in the best way, taking care of individual differences

(LD is an instance of a pedagogical model: a concrete application of a pedagogical model for a specific target group, for specific learning objectives and a specific domain)

Content Packaging & Learning Design

PACKAGE

Manifest

Physical Files

The actual content: HTML, Media, Activity descriptions, Collaboration and other files

Meta-data

Organizations:Organization

Resources:Resource

(sub)Manifest

Unit of Learning

Manifest

Physical Files

The actual content: HTML, Media, Activity descriptions, Collaboration and other files

Meta-data

Organizations:L. Design

Resources:Resource

(sub)Manifest

Why IMS LD?

Pedagogical meta model Offers a level of abstraction enabling different educational

models to be described, including:Learner, Knowledge, Assessment, Community Centered Approaches (in the different ‘schools’)

Software which knows about the meta-model can interpret specific models—model an approach to learning (eg problem based learning) and have it executed (‘played’)

Complete specification of a course (not only the resource part; needed for automation and interoperability)

Moves the focus from Learning Objects to Learning Activities

Some References

IMS LD (download www.imsglobal.org) www.learningnetworks.org (EML) See: list with recent journal articles/books/chapters

Agents for Support Activities (ASA)

Peter van Rosmalen , Peter Sloep

November, 2003

Rationale for ASA

1. Support staff lends support to many different kinds of Learning Activities.

2. This puts quite a strain on the support staff.3. From an institutional point of view this

means that providing support for learners rapidly becomes unaffordable.

Premises

Establish learning related interactions between distributed actors and distributed resources in a Learning Network. Do so efficiently: minimally maintaining the intensity and learning quality of the interactions without increasing staff workload.

Objective

To develop learning technologies (agents) that help tutors support their students in learning networks by

1. Building an abstract change model that provides entry points for the development of tools 

2. Developing functional prototypes of these tools and test them in pilots.

Outcomes

• a model of how tutors will be supported in their support activities for the Activity Nodes in a Learning Network

• prototypical software modules that qualify as generic support agents for tutors

• a model of how agents operate within the context of a design specified in IMS-LD.

Some details

Focus on the tutor: support the tutor, not the learners directly

Focus on agents that will build upon language technologies (e.g. support for e-mail answering and essay grading)

Navigation in Learning Networks

Colin Tattersall

Navigation in learning networks

Exploiting collective learner interactions to help learners select paths through learning networks towards their educational goals. “Others who went before you proceeded that way to

reach their educational goals”. A feedback loop which guides learners in deciding what

to do next. The idea is that an individual’s chances of reaching his

or her goals are improved through insights on how others have successfully reached their goals.

Aim: to improve educational yield using principles of self-organisation

Educational yield

m learners n learners

Time period t

t0 t1

Educational yield is the percentage of learners which successfully meet certain criteria in a given timeframe. Success might be ‘accumulating study points’, undergoing an oral examination, etc

Yield = ((n/m) * 100)

Positioning in Learning Networks

the point of departure (which competencies an individual already possesses) destination (which competencies are desired to be gained) the assessment of whether the destination has actually been reached (i.e. testing whether competencies have been mastered)

Goals = “Destination” = a position in a learning network which reflects the mastery of certain competencies;

Navigation vs Positioning

Positioning: “I have these competencies and I want those” “I am here and I want to be there”

Navigation: how to get from here to there (Must carry out all units of learning in a certain order) Must carry out all units of learning but can vary order

Travelling Learner Problem

Can select which units of learning to perform vs. which to skip but must follow a particular order

Can select which units of learning to perform and in which order

Emergent,macro-levelinformation

Track1: AN1, AN4, AN6: learner1, learner6, …;Track2: AN1, AN6: learner4, learner9,…;Track3: AN1, AN10, AN2, AN4: learner7;Track4: AN1, AN2: learner99, learner77, …

Learner/activity interactiondata

Micro-levelinteractions

•Which information should be fed back (eg, success rate, time taken)•How? As abstract directed graphs? Landscapes of competencies?•When? Always show everything?

Feedbackinself-organisinglearningnetworks

Feedback to learners Feedback to providers

Presentationof collectivelearnerbehaviour

Filters

Interaction data

Inspiration: self-organisation by ants

Learning Tracks & Roadmap

Tracks are left behind by learners like the pheromones left behind by ants

The intensity of the track reflects chances of success, number of attempts, time taken, … ?

Feedback to help answering …

Learners How did other learners progress in this learning network from

where am I now?

Which path through the learning network offer the most chance of success?

What has been the fastest path taken by others through this Learning Network?

Providers What percentage of learners followed the learning route(s)

prescribed in the curriculum through the learning network?

Is the learning route the most efficient way to progress through the learning network or are learners identifying better paths?

Where are learners slowing down or dropping out?

November,2003

ALFANET

Active Learning For Adaptive Internet

Peter van Rosmalen Open University of the Netherlands

Project AimsAlfanet aims to develop new methods and services for active and adaptive e-learning. The project’s target is to deliver a tested set of components for e-learning providers that will provide significantly enhanced individual learning, through technologies with adaptive features and approaches.

Key issues:

• Adaptation – individual needs design- & runtime:

Links, contents & collaboration

• Feedback loop for the design

• Agent-supported architecture

• Standards: IMS-LD, ……

• Partners: SAGE, UNED, EDP, KLETT, ACE-BNET, OUNL

Core Components and Standards

OpenACS - (UNED) provides facilities for collaborative learning.

IMS-LD - to enable advanced pedagogical designs including adaptation- to enable communication between the different actors – designers, tutors & agents

IMS-LD authoring tool- (ACE-BNET)

IMS-LD engine- (OUNL)

Agents

Learning Adaptation Model (UNED)- will support the learner in collaborative learning, navigation and content selection.

Audit (OUNL)- will support the design team with feedback concerning the initial design and the actual use/results

Multi-Agent Pedagogical Model (OUNL)-will support the design team with the selection & use of LD-models.- will support the learner during the execution of selected activities.

IMS LD

IMS LD

properties

IMS-LD-engineAgents:• Audit• Adaptation• MAPM

Tutor:• Designed role• Observator role

Authoring tool: design time

Presentation layer

Adaptation based on the design

Adaptation based on runtime monitoring of agents and tutors

Audit feedback to the design based on runtime monitoring

Unit of LearningIMS-LD

Technical ArchitectureLD En

J2EE Application Server

Security Layer Presentation Layer

Dispatcher

Tracker

Server

Services

Data

Common Repositories

Authoring Tool

Object Model

1…n

Current Status

Current Status

Alfanet version 1 (1 January-2004)

Integrated:- IMS LD Authoring Tool- OpenACS (collaborative framework)- IMS LD level A engine

Partly integrated / partly demonstrators- Learning Adaptation- Audit- MAPM

Evaluation round 1 (January-March 2004)- design and learner evaluation of two courses