The document discusses services and the web of data from an engineering perspective. It proposes that as linked data applications increase in complexity, there will need to be increased reuse of pre-existing solutions and components offered as services. Problem-solving methods research focused on decoupling problem-solving knowledge from domains to enable reuse. Infrastructure is needed to support systematically sharing and finding reusable functionality, including through the use of semantic technologies and problem-solving methods. Challenges include balancing overhead and performance with reuse and genericity.

1. Services and the
Web of Data
An Engineering Perspective
Carlos Pedrinaci, Maria Maleshkova
(The Open University)
Thursday, 7 July 2011

2. Acknowledgements
• Dong Liu (OU)
• Ning Li (OU)
• Jacek Kopecky (OU)
• John Domingue (OU)
• SOA4All project
Thursday, 7 July 2011

3. Thursday, 7 July 2011

4. From Linked Data to
Smart Applications
• Linked Data Principles
• Simple, clear, well-established
• Linked Data Applications
• Based on these simple principles and
technologies
• ... and a bunch of hacks on top
Thursday, 7 July 2011

5. We need reusable
Components
The level of complexity and refinement
of Linked Data Apps will be proportional
to our ability to reuse pre-existing
solutions (i.e., components/
functionality)
Thursday, 7 July 2011

6. How much
functionality are
we sharing?
And reusing?
Thursday, 7 July 2011

7. ...this will be often
offered as a Service
For reasons of scale (or control), in the
Semantic Web many of these components
will have to be offered as services that
gather and analyse large quantities of
data to provide advanced functionality (or
their results) online
Thursday, 7 July 2011

8. Developing and Models for capturing services
Registry for sharing and finding these
sharing software Engine for invoking services and providing
a linked data interface over these...
Thursday, 7 July 2011

9. SOA4RE
Mashes Linked Data and
Web APIs data
Finds and Invokes Linked
Services on the Fly
Modularity and
Extensibility as a core
built-in feature
Thursday, 7 July 2011

10. Where are all the
Problem Solvers Gone?
• Applications require both static and
dynamic knowledge
• “To build systems that solve real-world
tasks, however, we must not only specify
our conceptualizations, but also clarify
how problem solving ideally will occur.”
M. A. Musen
Thursday, 7 July 2011

11. Abstracting
Problem-solving
• A pillar of semantic technologies is
genericity. Let’s exploit it!
• This requires decoupling problem-
solving knowledge from the domain
• Problem-Solving Methods research
focussed precisely on this
Thursday, 7 July 2011

12. Applying PSMs over Engines based on problem-specific
vocabularies
domain specific data Used by feeding them with domain-specific
data at invocation time
Thursday, 7 July 2011

13. Suggestion
• We need to support the systematic and
effective reuse of functionality
• We have devised some initial infrastructure to
support this process
• Decoupling problem-solving knowledge from
the domain is essential for further reuse
• PSMs research showed us how to do it
Thursday, 7 July 2011

14. Challenges
• Keep the overhead low
• Exploit the cloud and other people’s
work/services
• Performance vs genericity
• Full complexity of PSMs needs to be
avoided but is also not necessary
• Surrounding solid tooling is necessary
Thursday, 7 July 2011

15. Share your data
Thursday, 7 July 2011

16. ... and your
components!
Thursday, 7 July 2011

17. “... this can’t
work ...”
Thursday, 7 July 2011

18. Bringing the Big Think to the Small Screen
Slide by Tom Gruber, Siri
Thursday, 7 July 2011