Versioning of Learning Objects

Preview:

DESCRIPTION

Versioning of Learning Objects. Christopher Brooks, John Cooke, Julita Vassileva Advanced Research in Intelligent Educational Systems (ARIES) Laboratory Computer Science Department University of Saskatchewan Saskatoon, SK, Canada. Motivation: Nature of Learning Objects. - PowerPoint PPT Presentation

Citation preview

Versioning of Learning Objects

Christopher Brooks, John Cooke, Julita Vassileva

Advanced Research in Intelligent Educational Systems (ARIES) Laboratory

Computer Science DepartmentUniversity of Saskatchewan

Saskatoon, SK, Canada

2

Motivation: Nature of Learning Objects

• Free to use and modify– “open source” (creative commons license, writings of

Stephen Downes)

• Distributed both in use and storage– POOL, POND, SPLASH, Edutella, etc

• Customized for a particular audience, with common backgrounds and goals– Localized versions of learning objects (CanCore,

Le@rning Federation Profile, etc)

• David Wiley’s reusability paradox– “the more reusable a learning object is, the harder its

use is to automate [combination and delivery]”

3

Motivation: Lecture Learning Objects

• Consider anecdotal evidence about lecture notes:– Shared between lecturers for no cost– Modified to fit the presentation style of the

new lecturer– Modified to contain different content based on

course goals • e.g. examples in Java instead of C for an

introductory computer science course

4

Approach: Authoring Learning Objects

• These observations suggest that it is important to support the authoring and editing of learning objects

– Automated reusability is highly complex for both a content management system and for learning object authors

5

Approach: Versioning

• the act of recording the evolution of software artifacts– Well studied: Software Configuration Management,

Hypermedia, Knowledge Representation

• Some important aspects of versioning for LOs:– Learning objects are considered immutable, changes

just create new versions– Older learning objects can be retrieved and versioned

again, creating revision trees– Versions are often meant to coexist with one another

(variants)

6

Approach: What to capture

• Two kinds of versioning need to be captured:– Semantics: How has the meaning of this learning

object changed with respect to the one it was derived from?

– Syntax: How can I retrieve and display a previous version of this learning object?

• Three pieces of a learning object need to be versioned:– The content– The display style– The metadata

7

Approach: Problems with Current MD

• Current metadata specifications tend to focus on human readable values instead of machine readable values

• Further, they don’t capture how the learning object has changed

• LOM– Relation: Identifies whether the version is a variant or

revision– Version: Identifies a human readable tag– Status: One of draft, final, revised, unavailable

8

Approach: Metadata Model at a Glance

• IEEE LOM compliant– Does not duplicate or redefine elements that already

exist

• Content changes (deltas) are stored as a tuple– The kind of change, which identifies

(programmatically) how a value can be applied to reproduce the change

• E.g. XML:NodeAdded or ASCII:CharactersInserted

– The value of the change• E.g. <b>Object databases are also known…</b>

9

Approach: Metadata Model at a Glance

• Semantics related to the change are also captured– The perspective of the change, which identifies

which metadata key is effected by the change• Encoded as a path through a metadata tree• E.g. “LOMv1.0:General:Coverage”• E.g. “DC:Coverage”

– The nature of the change, from a predefined vocabulary (machine understandable)

• Subset, superset, equivalent, etc.

10

Approach: Implementation

• Very little work has been done on syntactic interoperability of a learning object– authoring environments extremely diverse – E.g. Word, PowerPoint, Photoshop, Notepad, etc.

• We define a subset of XHTML to be used to mark up content– Allows for easily refactoring most existing learning

objects• We provide a vocabulary for content change

based around the DOM• The history of a LO can be viewed by

applying XSLT

11

Conclusions

• The LOM is insufficient for capturing version changes in learning objects

• We have proposed a model that can capture both the syntax of a change as well as the semantics associated with that change

• While this model is LO format neutral, it is being applied to lecture based learning objects using a subset of XHTML

12

Questions?

Contact information:http://www.cs.usask.ca/research/research_groups/aries/

Christopher Brooks:cab938@mail.usask.ca

www.cs.usask.ca/~cab938

John Cookecooke@cs.usask.cawww.cs.usask.ca/~cooke

Julita Vasselivajiv@cs.usask.cawww.cs.usask.ca/~jiv

13

Why not use CVS?

• CVS, and most other software versioning environments are:– Centralized

• Source code is usually shared amongst a small community• Learning objects are meant to be distributed to a large

community

– Agent unfriendly• Semantics are not stored in a machine readable format• Workflow (searching) of learning objects not supported by

CVS

• Decentralized CVS variants?– tend to decentralize artifact storage only, and still

have centralized metadata catalogues

Recommended