SeLeNe : Self eLearning Networks

1
SeLeNe Self e-Learning Networks

• Alex Poulovassilis, Birkbeck, Univ. of London
• One-year Accompanying Measure for
• IST V.1.9 CPA9 GRID Technologies and their
  Applications

2
SeLeNe Motivation

• Information and Knowledge Grids will require
  semantic integration of heterogeneous information
  sources
• This will need
• common metadata standards
• automatic reconciliation and integration of
  metadata
• definition of personalised views over the whole
  distributed information resource
• change notification and propagation from source
  to derived objects

3
SeLeNe Aims

• The aim of SeLeNe is to investigate the technical
  feasibility of using Semantic Web technology to
  support this functionality, with e-Learning as
  the application testbed
• Todays learners are
• geographically distributed
• heterogeneous in backgrounds and learning needs
• A wealth of learning resources are increasingly
  being made available on the Web, again
• geographically distributed
• heterogeneous in content, aim, quality

4
SeLeNe Aims

• We define a Self e-Learning Network to be
• a distributed repository of educational
  metadata describing learning objects available on
  the WWW, collaboratively built and used by anyone
  who wishes to use existing learning objects or
  to construct new learning objects, in any
  knowledge domain

5
SeLeNe Objectives

• The SeLeNe Project has 4 main objectives, each of
  which is being addressed by one Workpackage
  (running Nov 2002 Oct 2003)
• conduct a study of on-line educational resources
  and metadata
• identify technologies for managing distributed,
  evolving RDF repositories
• identify technologies for syndication and
  personalisation of educational resources
• design the high-level system architecture of a
  Self e-Learning Network this is likely to be
  based on a combination of Grid and P2P components

6
Syndication and Personalisation

• This aspect will include
• semantic reconciliation and integration of
  heterogeneous metadata describing learning
  objects and learners
• support of structured and unstructured querying
  over this metadata
• maintaining personal and group profiles which
  will have an effect on selection, ranking and
  presentation of query results
• facilities for defining derived LOs, as views
  over other, base or derived, LOs
• change notification and propagation from source
  LOs to derived Los

7
(No Transcript)
8
SeLeNe Consortium Members

• Birkbeck
• database and web technologies ECA rules, data
  integration, web personalisation
• Institute of Education, University of London
• educational metadata, users requirements for
  e-learning environments
• FORTH, Crete
• semantic web technology - RDF Suite and
  extensions P2P
• LRI, University of Paris Sud
• semantic integration and ontologies
• University of Cyprus
• distributed systems and Grid computing

9
SeLeNe Consortium Members

• Birkbeck
• George Loizou, Alex Poulovassilis, Mark Levene,
  Peter Wood, Kevin Keenoy (RA), George Papamarkos
  (RA)
• Institute of Education, University of London
• Don Peterson, Richard Noss, Kevin Keenoy (joint
  BBK/IOE RA)
• FORTH, Crete
• Vassilis Christophides, Dimitris Plexousakis,
  Panos Constantopoulos, Miltos Stratakis, Aimilia
  Maganaraki, Giorgios Zacharioudakis
• LRI, University of Paris Sud
• Nicholas Spyratos, Phillipe Rigaux
• University of Cyprus
• George Samaras, Kyriakos Karenos

10
Main Features of SeLeNe

• SeLeNe is intended for both supervised and
  unsupervised learning, and also for the shades of
  grey in between
• We envisage many self e-learning networks
  (SeLeNes) running world-wide, created by and
  serving different communities of users
• New nodes will be able to join a SeLeNe by
  contacting a known node
• The SeLeNe will allow access to LOs that have
  been registered with it and to their metadata

11
Authorisation, Access Control, Pricing

• There will be a need for authorisation and access
  control for different categories of LOs and users
• Also a need for pricing of different types of
  access to different categories of LOs by
  different categories of users
• We envisage a distributed catalogue of all
  metadata, with one fragment per node of the
  network likely to be a distributed RDF
  repository

12
Creating LOs

• Users will be able to register base LOs and their
  associated metadata with a SeLeNe, and also to
  define and register derived LOs
• At a minimum, the metadata for a base or derived
  LO will be represented using the RDF/S binding of
  IEEE LOM. Other metadata may also be associated
  with a LO e.g. relating to domain or
  topic-specific ontologies
• Users will be able to make local replicas of LOs
  to which they are allowed access, plus their
  metadata. The birth catalogue will keep track
  of all the replicas of an original LO

13
LO Replication

• Whenever an original LO is updated or deleted by
  its author, the SeLeNe will be able to notify all
  nodes which have made replicas
• A user of a replicated LO may choose to be
  notified about different classes of updates to
  the source LO and/or to other replicas
• Similarly, the author of the original LO may
  choose to be notified about changes made by
  someone else to a replica of the LO
• If changes to a LO O are not applied to some
  replica R then R ceases to be a replica of O
  similarly, if the author of the original declines
  to apply updates made to a replica

14
Change notification and propagation

• SeLeNe will notify delta changes to a LO to those
  users who have subscribed to these changes, if
  the changes can be derived from the LOs content
  and metadata otherwise, subscribers will be
  notified of the whole new LO
• A derived LO, O, is dependent on the LOs in terms
  of which it is directly defined, dep(O). O may
  be either virtual (no local replication of
  dep(O)) or materialised (local replication of
  dep(O)) or indeed a mixture of virtual and
  materialised
• Locally, SeLeNe will be able to propagate changes
  from dep(O) to O, generating the new metadata for
  O

15
Retrieving LOs

• We envisage several alternatives will be
  available to users to retrieve LOs accessible to
  them on a SeLeNe
• Browsing through the metadata, or through a view
  thereof
• Keyword search
• Annotated keyword search
• Specification of their learning needs and current
  levels of knowledge a formalism will be needed
  for specifying this information and for matching
  it with the LO metadata
• We also envisage SeLeNe maintaining both personal
  and group profiles which will have an effect on
  selection, ranking and presentation of results

16
Retrieving LOs

• Users will have a choice of having either
• a disconnected set of LOs returned, or
• sets of LOs connected into trails
• A trail is a coherent sequence of LOs reflecting
  a possible navigation of the LO space suitable
  for the users expressed needs and profile
• See the paper by Mark Levene and Don Peterson on
  Trail Records and Ampliative Learning for
  further discussion of the role of trails in
  learning (www.dcs.bbk.ac.uk/mark/download/trail_e
  pistemics.pdf)

17
Trails

• Trails will be first-class objects, with their
  own metadata, and facilities for creating,
  storing, modifying and exchanging them
• We envisage 3 types of trails
• authored either individually or collectively
• derived from the LO metadata
• emergent from repeated individual or collective
  navigation of the LO space

18
Some SeLeNe Usage Scenarios

• Some of the usage scenarios we envisage for
  SeLeNe are
• An individual author creating and registering a
  new base LO or derived LO and its associated
  metadata
• A group of authors collaborating in creating a LO
  base, derived, or a mixture thereof
• An individual learner retrieving and using LOs,
  and maintaining her/his own trail space
• A group of learners collaborating in retrieving
  and using LOs and in maintaining a shared trail
  space
• A group of teachers and learners collaborating in
  authoring and using a set of LOs and a shared
  trail space.

19
Conclusions Novel Features of SeLeNe

• Support and reconciliation of multiple ontologies
  of learning
• Definition of derived LOs as views of base LOs
• Change notification and change propagation from
  base to derived LOs
• Support of trails as first-class objects
• Support for unsupervised and supervised learning
• Support for collaborative authoring and learning
• Support for a variety of query modes browsing,
  keyword search, annotated keyword search,
  specification
• Support for personalisation of query results,
  with both individual and group profiles