Metadata for Resource Discovery in Learning Repositories

1
Metadata for Resource Discovery in Learning
Repositories
A.R.D. Prasad, Devika P. Madalli DRTC, Indian
Statistical Institute Bangalore, India ard,
devika_at_drtc.isibang.ac.in T4E Conference,
IIIT-B, 06/08/09
2
E- LearningInstructor Centric to Learner Centric

• In traditional learning scenario, the instructor
  plays the intermediate role between the learner
  and the learning material
• But In an e-learning environment
• Instructors no longer control the delivery of
  material and learners have a possibility to
  combine learning material in courses on their
  own.
• User does not require packaged bundles of
  lessons.

3

• E-learning projects have mostly concentrated on
  putting high quality material in the form of
  lectures, lessons, presentations and also media
  files on disk.
• Regardless of the time or expense put into
  creating advanced training material the content
  is useless, unless it can be searched and
  accessed easily.

4
An Ideal e-learning System

• Learner centric approach
• Flexibility (time space independent)
• Customized and/ or personalized paths
• Non-linear content
• Interactive Learning
• Dynamic content
• Systematic Learning
• Distributed Content

5
Issues in e-learning Environment

• Lack of group and personalized learning spaces
• Presentation of the entire learning material as
  continuous text or media, instead of only the
  relevant information that is actually sought by a
  learner.
• Learning sequence not same for every user
• Re-usability issues
• - Ljiljana Stojanovic, Steffen Staab and Rudi
  Studer, 2007

6
Issues in e-learning environment

• Resource discovery issues and lack of semantic
  interoperability
• Quality assurance
• Ranking and relevance indicators
• Personalized paths and services

7
Content, Context Structure

• content identifies what the learning material is
  about,
• context identifies in which form this topic is
  presented,
• structure is to comprehend the arrangement of the
  learning materials in a learning course

8
ContentContextStructure

• High risk in defining the semantically same
  concept/ topic by two different authors
• Keyword based querying does not really work
  unless they are well defined
• Problems of synonym, antonym, homonym,
  abbreviation, etc.
• Contextual information is missing
• Incapable of meeting the individual skill gaps
• Incapable of delivering content as per
• individual needs

C O N T E N T
Context
Structure
9
A Look at NPTEL
10
(No Transcript)
11
NPTEL

• Multidisciplinary, rich in collection and
  variety.
• Search and Browse Interface-
• by subject (drop-down list of topics/ subject
  enlisting about 8 disciplines)
• by material types (i.e. by Web or by Video),
• by Institutes (7 IITs and IISc),
• by course name,
• by coordinator.

12
(No Transcript)
13
Issues access and retrieval

• Assumption - users would know 'what exactly they
  need'
• How many would really be able term their
  information requirement in formal domain language
  in a search box?
• Semantic gaps between perceived information need
  and expressed information need

14
What more NPTEL can offer?

• Intuitive search and resource discovery features
  enabled through deployment of adequate metadata
  for content description,
• Domain hierarchies in lightweight directory
  structure.
• Support for stemming, parts of names, spelling
  variations through controlled vocabulary.
• Unscrambling the concepts/ disambiguation
• User profile based services

15
Two approaches of resource description

• Metadata based description (using standards like
  LOM)
• Domain driven approach (through domain ontologies)

These two approaches are not exclusive rather
compliment each other for better organization and
retrieval
16
Metadata

• It is the Internet-age term for cataloging
  information that librarians traditionally have
  used to provide 'access points' to items in
  collections.
• A metadata standard advocates use of common set
  of tags that can be applied to describe any
  resource with an aim to facilitate its retrieval
  from a collection
• Metadata ensures interoperability and helps in
  reuse of resources and artifacts

17
Metadata in e-learning

• Learner centric educational architecture
• Domain based knowledge retrieval
• Semantic interoperability, (instructor-learner
  learner-learner peer- learner
  instructor-instructor)
• Re-usability

18
Improved semantics

• Identify works by Ranganathan instead of
  retrieving just everything about/by Ranganathan?
• Relations between pieces of tagged information
• Machine/agent processable information extraction
  is possible
• Interoperability needs to be ensured. (through
  crosswalks, standardization, etc.)

19
Domain ontology driven approach

• Semantic Metadata
• Metadata basically describes the context and
  often not the content of the resource.
• The content should be expressed in a formalized
  representation- ontology.
• Such an holistic approach to resource description
  can be termed as 'Semantic Metadata.

20
Ontology?

• An Ontology is an explicit specification of a
  conceptualization - Gruber
• Ontological categories define the concepts that
  exist in the domain, as well as relationships
  between these concepts.

21
Domain ontology for CS
22

• An LMS based on domain ontology makes it
  relatively easier for the learner to find objects
  using the implicit embedded relations between
  concepts.
• The benefits of using domain ontology for LMS can
  be explained in following scenario

23
Benefits of Ontology Based LMSScenario 1 Issue
of Synonyms

• User need- Learning objects on the topic Local
  Area Network.
• Key word used - LAN
• Results Only Includes the objects indexed with
  the key-word LAN , not possible to retrieve
  those with Local Area Network.
• Reason- System is incapable to establish
  relation between LAN and Local Area Network.
• In Ontology Based LMS
• the relations between abbreviated words and
  abbreviation will be hard-coded
• hence easily retrieves documents indexed with
  either of the forms

24
Benefits of Ontology Based LMSScenario 2 Issue
of Homographs

• User need- Learning objects on the topic cranes
  in the context of Civil engineering.
• Key word used - Cranes
• Results Retrieves all the objects indexed with
  the keyword cranes which includes the bird
  crane also.
• Reason- System is incapable to distinguish
  between Bird- crane and The machine- crane.
• In Ontology Based LMS
• The broader terms of the word 'crane' would
  resolve this issue.
• As the broader terms to crane could be 'building
  material' or 'bird' as the case maybe. (dealing
  with polyhierarchies)

25

• In brief, a system driven by domain ontology
  driven would be capable of handling different
  forms of a concept, e.g. synonyms, homonyms,
  homographs, etc.
• Such an Ontology and domain driven implementation
  should be in compliance with the W3C standards on
  ontology representation

26
Proposal-LOM supported Learning Management System

• OAI-compliant LOM supported repository
• Metadata based search as well as faceted browsing
• Enhanced support for intuitive search

27
Architecture for LOM supported Learning
Management System
28
Ontology and its components

• Knowledge space
• Domain ontology (SKOS (Simple Knowledge
  Organization System))
• Content specific (OWL-DL)
• Learners ontology (OWL-DL)
• Rules
• Protégé for building ontology

29

• In an ideal learning situation system would
  provide a material which is the best quality one
  and best suited to the individual learners
  considering their background knowledge, style of
  learning, their educational level, technical
  facilities, etc.

• Software agents must be able to automatically
  find, discover appropriate service on behalf of
  us
• They must be able to determine automatically how
  to invoke or execute the service
• They must be able to select and combine the
  required information and deliver it

30
In a Nutshell

• Essentially components of e-learning system for
  resource description using
• LOM metadata standard and
• An domain ontology based content representation.
• Such an international standard compliant system
  helps to
• achieve interoperability,
• updating
• versioning
• personalization
• migration and
• reuse of the content

31
Thank You
ard_at_drtc.isibang.ac.in devika_at_drtc.isibang.ac.in