Collaborative knowledge visualisation for cross community knowledge exchange [Elektronische Ressource] / von Jasminko Novak

Jasminko Novak

Sujets

• Management

Informations

• Publié par: universitat_duisburg-essen
• Publié le: 01 janvier 2006
• Nombre de lectures: 60
• Langue: English
• Poids de l'ouvrage: 18 Mo

Extrait

Collaborative Knowledge Visualisation for
Cross-Community Knowledge Exchange



Von der Fakultät für Ingenieurwissenschaften der
Universität Duisburg-Essen
zur Erlangung des akademischen Grades eines
Doktors der Ingenieurwissenschaften
genehmigte Dissertation


von


Dipl.-Ing. Jasminko Novak

aus Derventa














Referent: Prof.Dr-Ing. Jürgen Ziegler
Korreferent: Prof.Dr. Heinz Ulrich Hoppe
Tag der mündlichen Prüfung: 16. Oktober 2006


Abstract
The notion of communities as informal social networks based on shared interests or common practices has been
increasingly used as an important unit of analysis when considering the processes of cooperative creation and
sharing of knowledge. While knowledge exchange within communities has been extensively researched, different
studies observed the importance of cross-community knowledge exchange for the creation of new knowledge and
innovation in knowledge-intensive organizations. Especially in knowledge management a critical problem has
become the need to support the cooperation and exchange of knowledge between different communities with highly
specialized expertise and activities. Though several studies discuss the importance and difficulties of knowledge
sharing across community boundaries, the development of technological support incorporating these findings has
been little addressed.
This work presents an approach to supporting cross-community knowledge exchange based on using knowledge
visualisation for facilitating information access in unfamiliar community domains. The theoretical grounding and
practical relevance of the proposed approach are ensured by defining a requirements model that integrates
theoretical frameworks for cross-community knowledge exchange with practical needs of typical knowledge
management processes and sensemaking tasks in information access in unfamiliar domains. This synthesis
suggests that visualising knowledge structures of communities and supporting the discovery of relationships between
them during access to community spaces, could provide valuable support for cross-community discovery and sharing
of knowledge. This is the main hypothesis investigated in this thesis.
Accordingly, a novel method is developed for eliciting and visualising implicit knowledge structures of individuals and
communities in form of dynamic knowledge maps that make the elicited knowledge usable for semantic exploration
and navigation of community spaces. The method allows unobtrusive construction of personal and community
knowledge maps based on user interaction with information and their use for dynamic classification of information
from a specific point of view. The visualisation model combines Document Maps presenting main topics, document
clusters and relationships between knowledge reflected in community spaces with Concept Maps visualising personal
and shared conceptual structures of community members. The technical realization integrates Kohonen’s self-
organizing maps with extraction of word categories from texts, collaborative indexing and personalised classification
based on user-induced templates. This is accompanied by intuitive visualisation and interaction with complex
information spaces based on multi-view navigation of document landscapes and concept networks.
The developed method is prototypically implemented in form of an application framework, a concrete system and a
visual information interface for multi-perspective access to community information spaces, the Knowledge Explorer.
The application framework implements services for generating and using personal and community knowledge maps
to support explicit and implicit knowledge exchange between members of different communities. The Knowledge
Explorer allows simultaneous visualisation of different personal and community knowledge structures and enables
their use for structuring, exploring and navigating community information spaces from different points of view.
The empirical evaluation in a comparative laboratory study confirms the adequacy of the developed solutions with
respect to specific requirements of the cross-community problem and demonstrates much better quality of knowledge
access compared to a standard information seeking reference system. The developed evaluation framework and
operative measures for quality of knowledge access in cross-community contexts also provide a theoretically
grounded and practically feasible method for further developing and evaluating new solutions addressing this
important but little investigated problem.








Acknowledgements

The work presented in this thesis has been undertaken during my time as a researcher at the Fraunhofer Institute for
Media Communication. Parts of the reported research were conducted in the collaborative research project “AWAKE
– Networked Awareness and Knowledge Discovery” led by the author and supported by the German Federal Ministry
of Education and Research. (BMBF). The project “CAT/netzspannung.org – an Internet Media Lab for
Communication of Art & Technology” (BMBF) also provided an important experimental context for testing the
developed methods and solutions.
I would like to thank my supervisor Prof. Dr. Jürgen Ziegler for his support, critical feedback and stimulating
discussions: they helped me in distilling the richness of the developed ideas and experimental findings into mature
scientific work. His openness to new ideas paired with his great experience in the design and evaluation of interactive
system were a valuable companion on the often unpredictable journey that has led to this thesis.
My thanks also go to Monika Fleischmann, Head of the MARS Exploratory Media Lab at the Fraunhofer IMK for her
support during the course of this work. We have passed together many a storm and I was privileged to learn from her
ceaseless enthusiasm in crossing the borders between different worlds of knowledge and fields of profession. I
further thank Prof. Dr. Martin Reiser, the IMK Institute Director for his support of the project and of my work as well as
the whole IMK staff. Many thanks to Prof. Dr. Gerhard Schwabe from the University of Zurich for his comprehension
and the opportunity to deliver the finishing touches to this thesis after taking on my new position in his group.
This work would have not been possible without the help of many people. In particular, I have enjoyed the
collaboration with many great people in the AWAKE project. I would like to thank the responsible partners Prof. Dr.
Jürgen Ziegler (Fraunhofer IAO), Prof. Dr. Katharina Morik (University of Dortmund) and Prof. Dr. Bernd Freisleben
(University of Siegen) for their great dedication and the belief in the project. During my stay at the University of
Milano-Bicocca, I have also had the privilege to discuss and confront my ideas with Prof. Dr. Carla Simone, Prof. Dr.
Giorgio de Michelis and Prof. Dr. Federico Butera.
My special appreciation goes to all the members of the project team who have contributed to the development of the
AWAKE system and helped make the project the success that it has become. I have enjoyed many fruitful
discussions with Michael Wurst (University of Dortmund), Stefan Paal (Fraunhofer IMK) and Christoph Kunz
(Fraunhofer IAO) which provided inspiration for some of the important ideas of this work. Michael has also
implemented the supervised classification modules and provided valuable advice on knowledge discovery methods.
Stefan provided the basis for the system middleware and valuable help in keeping the system up & running.
Kresimir Simunic (Fraunhofer IMK) supported the implementation of the Kohonen map, Reni Banov (Db-IT
Consulting), Christoph Seibert (Fraunhofer IMK), Roger Sennert and Jens Wagner (University of Siegen) contributed
to the realization of system integration services. Early interface designs have benefited from the support by Danijela
Djokic and Hartmut Bohnacker (projekttriangle). The final implementation of the Knowledge Explorer and underlying
services was supported by the students Ansgar Himmel, Philipp Wever, Kerstin Schmidt, Stefan Winarzki and Malte
Kloes. My long-time friends, Predrag Peranovic and Kresimir Simunic also provided valuable logistic assistance and
were great hosts during my frequent Milano-Cologne travels.
I also thank all the colleagues from the Fraunhofer IMK and the Dept. of Interactive Systems at the University of
Duisburg-Essen for their help with the preparation and execution of the laboratory experiments. I am indebted to all
test persons for their participation in rather extensive experiments which made the empirical verification possible.
I am grateful to all my friends and colleagues at the Fraunhofer Campus and the MARS Exploratory Media Lab for
their companionship in these years. I will kindly remember the many encounters and conversations with Gabriele
Blome, Carla Valle, Martha Larson, Markus Klann and many others. I thank all the friends in Cologne, Bonn, Rome,
Milan, Zagreb, Zurich, NYC and beyond - fortunately, they are way too many to be mentioned here by name!
Last but not least, I thank my parents for