Semantic Interoperability in User Modeling

1
Semantic Interoperability in User Modeling

• AlterEgo (TI)
• Develops the key concepts for cross-domain
  profiling from a user, business and technical
  perspective.
• TU/e focuses on semantic interoperability of
  user profiles between applications.
• Semantic Interoperability
• Applications only have a limited view on the
  user

• Many applications have a part of model on users
  in common, and want to exchange this information
  (collaboration)
• However, there are problems in the collaboration
• - Different ways to name and structure the
  user model information
• - Domain issues
• - Computers do not understand data.
• Goal facility to support user model data
  exchange between applications.

Figure 5. AlterEgo architecture of the UPI
(Universal Profiling Interoperability)

• Instance Translation
• Translating instance data of one application to
  the instance data of another application
• Informal Procedure
• - Build a graph view over the application
  Semantic Web schemas
• - Create a common graph by using graph-matching
  techniques
• - Semi-automatically generate an instance
  mapping recipe

• Motivation
• Developers want to personalize their
  application.
• - Maintain a model for their user but
• Users divide their time over several
  applications
• Single applications have a limited view of the
  user
• User knowledge is fragmented over applications
• Users have to provide the same data over and
  over again
• - Time-consuming
• - Annoying

Figure 4. Using Semantic Web technology to create
instance mappings

• UM Semantic Interoperability Server (UMSIS)
• Storing (and providing) User Model schemas
• Storing (and providing) the mapping between
  schemas
• Ability to apply a schema mapping to an instance
  
• (Semi)-automatically create the mappings
• A component that is independent of the
  architecture of the UPI
• Not fully automatic because of natural language
  ambiguities
• Mapping is application dependent
• Once a mapping between applications has been
  established it can be used indefinitely for all
  users

Figure 1. Illustration of Interoperability
problem with simple CNA data

• Restrictions
• The Web is a heterogeneous environment
• It is near to impossible to enforce applications
  to use a prescribed syntax, structure and
  semantics.
• A large group of legacy applications exist
  that cannot be (easily) adapted.
• Semantic Web
• Provides a standardized structure for syntactic
  interoperability
• Foundation for semantic interoperability
• A lot of Semantic Web Data is already out there
• Generic, i.e. almost all datamodels can be
  written in equivalent Semantic Web form.
• Comprises W3C like RDF(S), OWL, OWL-S, SPRQL,
  etc.

Figure 3. High-level architecture of UMSIS