Exploiting metadata, ontologies and semantics to design/enhance new end-user experiences for adaptive pervasive computing environments
- 1. Ahmet Soylu, PhD
KU Leuven, Belgium
web: www.ahmetsoylu.com
e-mail: ahmet.soylu@kuleuven-kulak.be
Senseable City Lab @ MIT, USA, 10 July 2012 (Online Pecha Kucha Style)
- 2. Introduction
Individual
user-models
adaptivity = intelligence
Collective
context
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 2
- 3. Problem Statement
End-user perspective Developer/development perspective
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 3
- 4. Objectives
to provide abstract development approaches and methods
to facilitate sustainable and rapid development and
management (conceptual)
to enable end-users to be aware of relevant context,
conceive adaptive behaviors, and to be involved in
adaptation process (conceptual)
to enable end-users to form their own personal and pervasive
environments (apps., appliances etc.) and to orchestrate
them (practical)
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 4
- 5. Approach
Web as a pervasive computing framework
Application space, communication space, and information space
Physical devices with digital presence => widgets for encapsulation
A uniform approach based on high level abstractions
Abstractions as a medium of development, adaptivity, and
interoperability
Interdisciplinary Research on Technology, Education and Communication, KU Leuven Campus Kortrijk, Belgium, www.itec-research.eu 5
- 6. Approach
Overall research framework: (1) abstract development, (2) widgetization/ubiquitous
access, (3) personal environments, (4) end-user involvement.
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 6
- 7. The Two-use of Ontologies
Individual application level
Perspectives
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 7
- 8. The Two-use of Ontologies
Individual application level
Models and ontologies
Three modeling paradigms
A uniform approach
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 8
- 9. The Two-use of Ontologies
Individual application level
Towards a methodology
Practical challenges:
- Logic layer
- OWL to Java/SQL etc.
A methodology for merging Model Driven Development and Ontologies
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 9
- 10. Widgetization
Collective level
Ubiquitous Web Navigation
An approach, a method and a set of
heuristics for specifying, extracting,
presenting embedded semantic data
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 10
- 11. Widgetization
Collective level
Data oriented and document oriented considerations
Semantic information network map of an example web application
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 11
- 12. Widgetization
Collective level
Prototype
An example navigation session of a user seeking a particular offer
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 12
- 13. Personal Environments
Collective level
Widget-based UI mashups
Widgets as building blocks!
Widget Interoperability : Data and application Interoperability
Platform and Framework: Standard components and services
Orchestration: Manual and automated
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 13
- 14. Personal Environments
Collective level
Application Interoperability Data Interoperability
Functional widget interfaces (FWI) Use of domain knowledge for data interoperability
Widget triggered by an event of another widget Data is copied from one widget to another by an end-user
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 14
- 15. Personal Environments
Collective level
Platform
The platform is composed of two
primary layers : a run-time system
and a backend system.
The run-time resides at the client
(e.g., browser) and is responsible
for the operational tasks and the
delivery of standard platform
services (e.g., preference
management) to the widget
instances.
The backend system resides at the
server side and is responsible for the
persistence and decision making.
We extended W3C standards for widgets
accordingly.
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 15
- 16. Personal Environments
Collective level
Behavior mining
Workflow/process mining for
the extraction of behavioral
patterns and Colored Petri nets
for the representation, share,
Petri nets for widget orchestration: transitions refer to widget functions
and validation & verification
A behavioral pattern consists
of a triggering action and one
or at most two parallel or
alternative actions.
The patterns used in automated orchestration: OR and Sequence
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 16
- 17. Personal Environments
Collective level
Prototype
An example scenario for system-driven widget orchestration
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 17
- 18. Conclusions
The Two-use of ontologies
A conceptual framework and approach,
Practical directions,
A uniform approach is yet to be realized.
Widgetization
A methodology,
Heuristics for the end-user consumption,
Other automated annotation mechanisms.
Personal Environments and Orchestration
Interoperability framework,
A generic platform,
Automated/manual widget orchestration,
Online learning and concept –drift problem.
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 18
- 19. Future Work
End-user “programming” for pervasive spaces Visualizing networked interaction in pervasive
(programming by demonstration) spaces (one-mode vs. two-mode networks)
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgium, www.itec-research.eu 19
- 20. Q&A
Thank you for your attention!
?
For more information and relevant publications:
www.ahmetsoylu.com
The Mechanical Turk
Interdisciplinary Research on Technology, Education and Communication, KU Leuven - KULAK, Belgi um, www.itec-research.eu 20
Editor's Notes
- Good afternoon!I have recently obtained my PhD from KU Leuven. In this presentation, I am going to provide an overview of my PhD work.My PhD focuses on responsive personal and pervasive environments and mainly utilizes abstractions to this end.
- Technology has augmented the end-user environments by merging physical and digital worlds. Now, end-users are confronted with overwhelming amount of data, applications, and appliances.Novel approaches are required to better manage, organize, and interact with these entities at individual application and collective environment levels.
- In this respect, adaptive and pervasive computing domains have emerged.However, there are certain problems.From end-user point of view, such systems are strongly invasive and non-transparent.From development point of view, such systems are becoming extremely complex and therefore are hard to develop and manage.
- To address these challenges, we aim to:
- Concerning our approach …
- Concerning our overall research framework…We use ontologies for automated code generation.Ontologies allow us to realize end-user awareness, software intelligibility, and end-user control.We harvest semantics embedded into application interfaces to widgetize traditional applications and to form personal environments.
- Regarding our work at individual application level…From knowledge representation point of view, we use ontologies to formalize application knowledge for automated development.From logic point of view, we use formalized context ontologies to realize dynamic adaptations. With formal ontologies, we can communicate relevant context and reasoning logic to the end-users.
- To realize such an approach, we proposed merging Model Driven Development and Ontologies.Literature reveals that the integration of prominent modeling paradigms is of importance.Upon such integration, we proposed an approach using ontologies for acquiring domain knowledge and to move form ontologies to more specific models.
- We proposed a conceptual methodology which is adapted from model driven development.The methodology employs domain ontologies as an input for model driven development.Domain ontologies along with generic ontologies are used as a context ontology at run-time.However, there are certain problems regarding the practical applicability of a merged approach in terms of logic layer and transformations.
- At collective level, our first practical study concerns the widgetization of traditional applications.Our goal is to enable application servers to directly serve semantic information for the end-user consumption.To this end, we proposed …
- We combine data oriented and document oriented considerations. A document oriented approach provides high provenance while a data oriented approach provides high precision. Our approach allows end-users to navigate semantic information network of web applications by following HTML and data links.
- We implemented a prototype.At the left hand-side, we see the example website which is composed of three pages.At the right hand-side, we see an example navigation session with our prototype.The end-user experiments and computational experiments revealed that the approach is promising.
- Our second practical study concerns the realization of widget-based personal environments. Widgets are used as building blocks of the personal and pervasive environments as interplaying components.We identified and addressed several challenges.
- For application interoperability, we introduced functional widget interfaces to disclose functionalities of the widgets.For data interoperability, we annotated functional interfaces and the content of widgets with semantic information.On top of this, we built an end-user data mobility facility for manual widget orchestration.
- The platform is composed of :
- For automated widget orchestration , we monitor the event stream of widgets, mine behavioral patterns, and automate these patterns. We employed…
- We implemented a prototype.The snapshot provides an example scenario with a set of extracted patterns.The end-user experiments and computational experiments revealed that our approach is promising.
- We proposed and implemented several approaches and methods for the realization of personal and pervasive environments.However:
- Our future work builds on and complements our previous work.Our first goal is to enable end-users to program their own environments for specific tasks through programming by demonstration idea. Our second goal is to realize a visualization framework for the analysis of networked interaction in pervasive environments to support end-users.
- As a final remark:The future is not only about more intelligent applications.The technology will be more ubiquitous and will extend our sensory, cognitive, and physical abilities as well as the borders of our environments. Therefore, approaches for the efficient integration of human intelligence with machine intelligence and his environment are required.Thank you!