On Interoperability of Ontologies for Webbased Educational Systems

1
On Interoperability of Ontologies for Web-based
Educational Systems

• Yannis Kalfoglou, Bo Hu, Dave Reynolds

This work is sponsored by CROSI (Capturing,
Representing and Operationalising Semantic
Integration) a joint research project between the
University of Southampton and HP Labs _at_Bristol
with the support of AKT consortium
2
overview

• Interoperability and ontologies
• WBESs needs KE support
• Global vs. multiple ontologies
• K-Sharing demand technology supply
• Empowering user communities
• Evaluation feedback
• Working examples
• Engineering effort benefits
• WBESs interoperability issues
• (Semantic) Web vision WBESs role

3
Interoperability facts

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Heterogeneity due to independently developed
  systems
• Homogeneous groups have been resolving it in
  familiar, closed environments, but
• endemic characteristic of the (Semantic) Web
  distributed, decentralised systems
• Furthermore communities introduce and use their
  own semantics
• Need to achieve structural as well as semantic
  interoperability

4
Interoperability solutions

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• need to expose and share information and
  knowledge between disparate systems
• Ontologies as the means for knowledge sharing and
  reuse
• In WBESs advocated as a common vocabulary for
  domain knowledge representation
• Legacy KE technology which created a niche
  ontology engineering

5
Global ontology - pros cons

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• In the 90s dominant approach for KE use of a
  well crafted, top-level, formal ontology
• Effective once it is ready reduces sharing and
  reuse costs
• Caveats designing the perfect ontology
  syndrome
• Irreconcilable arguments among engineers at
  design time
• Inappropriate interpretations and usage at
  deployment time
• Hard to maintain not stable over time
  obsolete knowledge
• 10-15 years later
• (Semantic) Web made ontologies far easier to
  publish, reach and access than KEs ever thought
  it would be possible
• ontology engineering tools appeal to large
  audiences, not only ontology engineers
• available informal ontologies outnumber by far
  formal ones
• Consequently, ontology supply outstrip demand.

6
Multiple ontologies - issues

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Proliferation of ontologies meant
• scalability authoring, deployment, need to be
  catered for.
• Interoperability of ontologies themselves
• interoperability is no longer based on a single,
  consensual ontology but uses multiple ontologies.
• need to achieve interoperability at the
  underpinning ontology level before achieving
  interoperability of WBESs (using those
  ontologies)
• Call for ontology to ontology interoperability
  ontology mapping
• Ontology mapping is not a solved problem though
• a lot of potential solutions available, but
• not integrated at the design phase
• not easy to replicate integrate with other
  components
• not automated, and not easy to maintain.

7
Using learners

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• learners access and use learning material
• In KE typically encode it as learning objects
  (preferably as an ontology)
• KEs model, encode and offer those to end users
  for immediate consumption
• But this, arguably, ignores input from learners
• Feedback and evaluation from learners should be
  used when modelling learning objects.

8
Using learners cntd.

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• (Web) users actively involved in various tasks
• FOAF network (typical foaf user not a KE)
• Blogging (users maintain them)
• RSS (decentralised authoring)
• P2P and other Web-based interaction
• (Semantic) Web environment and modus operandi
  encourages active user participation
• WBESs interoperability should exploit this
• Early theoretical investigations promote the role
  of communities in knowledge sharing (Kents work)

9
KM setting

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• AKT consortium five UK universities, 65
  personnel dispersed in five UK cities
• Collaborative tasks, KM at work
• practice what you preach!
• Task help new workers familiarize themselves
  with AKT
• Learning material wide variety of digital media
  A/V streams, Web-casting, story telling,
  newsletters, etc.
• Learning material was semantically annotated and
  encoded in an ontology

10
Engineering effort

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Resulting ontology one of the few, well-crafted,
  state-of-the-art Semantic Web ontologies
• Supports award winning applications (2003 SW
  Challenge winner) and a variety of AKT testbeds
• It has a knock-on effect on sharing and reuse
• But it took us 3 years to build it and is
  expensive to maintain

11
Using small, domain ontologies

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• We worked with smaller, domain- and task-
  specific ontologies
• myPlanet a Web-based, ontology-driven,
  personalised organisational newsletter

• Easier to build and maintain
• Directly appealing to end-users/learners, but
• Long term benefits difficult to quantify (limited
  learning material available)

12
Engaging users

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Ontology-based Communities of Practice (ONTOCOPI)

13
Engaging users cntd.

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Ontology Network Analysis (ONA)

14
Ontology mapping

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

Information Flow based Ontology Mapping
a method and a theory for Ontology mapping
Reference ontology
15
Multi- vs. single ontology

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• more than one ontology needed to support WBESs
• (Semantic) Web made that easier
• difficult to quantify and weigh the benefits of
  multiple vs. single ontology support for a WBES
• If a typical WBES (single domain, std learning
  obj) then a single ontology is good enough
• If a complex WBES (variety of learning obj,
  multi-domain), then multiple ontologies are
  needed
• lack of requirements analysis
• (not in general but specifically for ontology
  support)

16
Semantic Web enabled

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Semantic Web is both a challenge
• which ontology to use and how to choose the right
  one?
• authority and version control, inconsistency,
  incompleteness, trust and provenance
• and an opportunity for WBESs
• abundance of ontologies available for use
• Semantic Web built-in support for
  interoperability (OWL constructs)

17
Semantic interoperability

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Need to focus on semantics, structure alone
  cannot solve the interoperability problem
• Semantic interoperability needed when dealing
  with ontologies (rich in semantics), even more
  when multiple ontologies are used in a WBES
• WBES can inform semantic interoperability
  requirements thanks to their relatively uniform
  domain representation (learning material)

18
Community driven

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• WBESs provide access to large audiences/user base
• These could provide feedback and evaluate WBESs
• The quest is to find ways with which these users
  will inform and be involved in the
  interoperability process.
• then we are realising the goal of having
  operational semantics exposed by their users.

19
Versatile content

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

• Traditional views of WBESs content need to be
  updated.
• Many-forms, multiple-sourced, distributed digital
  content is easily accessible and available,
  nowadays
• Thanks to the (Semantic) Web, it could also be
  semantically annotated.
• WBESs should reflect this versatility and be able
  to accommodate many different kinds of content.
• Typically, accommodate university-like online
  course material
• Standardisation efforts on framework to assist
  integrating this content are already underway (EU
  PROLEARN project)

20
Questions?

• Interoperability ontologies
• Global vs. multiple ontologies
• Empowering user communities
• Working examples
• WBESs interoperability issues

21
Using an ontology to achieve interoperabilityOnto
logies principles, methods and applications1996
- KER 11(2)93-136 Uschold Gruninger
22
Theoretical investigations on the role of
communities in K-SharingInformation Flow
Framework R.Kent, 2003
23
Information Flow Based Ontology Mapping
(IF-Map)IF-Map an ontology mapping method based
on information flow theoryY.Kalfoglou
M.Schorlemmer, JDS (1)198-127, 2003
24
Information Flow Based Ontology Mapping
(IF-Map)IF-Map an ontology mapping method based
on information flow theoryY.Kalfoglou
M.Schorlemmer, JDS (1)198-127, 2003