InfoSky: Visualisation of Hierarchical Knowledge Spaces

Partner: Hyperwave Research & Development Ges.m.b.H

Employing today’s standard means of navigation and visualisation for large knowledge spaces, the identification of trends, relations and similarities between documents or search results is no trivial task. Knowledge Management Systems add to the problem by structuring documents in hierarchies, which are even less suited for explorative analysis in standard user interfaces. InfoSky is an interactive system for the exploration of large, hierarchically structured document collections. InfoSky employs a planar graphical representation with variable magnification like a real-world telescope. The hierarchical structure is reflected using recursive subdivision into Voronoi polygons. At each level of the hierarchy documents and subcollections are positioned according to the similarity of their content using a force-directed placement technique. Documents are assumed to have significant textual content, which can be extracted with specialised tools. The hierarchical structure is exploited for greater performance. Force-directed placement is applied recursively at each level on the objects at that level rather than on the whole corpus. The concept of InfoSky has been explored in detail during the years 2001 and 2002, with several publications and a patent for the principle as a result. In 2003, an advanced system prototype featuring full server-client-architecture and incorporating features such as user rights has been built.

top

IMB- Intelligente Multimedia Bibliothek

Partner: JOANNEUM RESEARCH Forschungsgesellschaft mbH

Having created a framework of tools for semantic annotation and retrieval and storage and indexing of multimedia documents it was decided that in 2003 an extensive market study had to be done. Investigating lots of different products and vendors in different areas of interest allowed us to redefine the context of our prototype and find further usage scenarios for semantic multimedia annotation and retrieval based on MPEG-7.

top

KnowMiner - Multipurpose, Generic, Cross-media Retrieval and Visualisation Framework

Strategic Research Project

Depending on the results of the V-Markt project we will chose the directions for further developing the WebRat framework from the year 2002. Focusing at the identified core areas it should become an adaptable, configurable, general-purpose knowledge discovery framowork. Special attention should be given to handling very large amount of data, hierarchically structure repositories, and data enriched with metadata. Some functionality and analysis methods from the InsoSky project as well as metadata support techniques from the IMB Project will also be integrated. Modules from all these systems will be brought together under a single unified architecture. A number of chosen prototype applications will be developed as proof of concept, for testing the system, and for identifying weaknesses of the concept.

top

onAIR: APA Intelligent Retrieval - Implementation

Partner: APA Defacto

Reliably locating information within a database containing 20 million documents is a challenging task when only standard information re-trieval techniques may be used. In the preceding project, several en-hancements to retrieval algorithms and implementation techniques used in the current version of APA Defacto News Search have been suggested. In cooperation with APA IT, these enhancements have now been implemented prototypically.
New or updated functionality included searching for similar docu-ments, algorithms for analysing search results, algorithms for auto-matically extending current search terms and algorithms from the field of information visualisation. In cooperation with our partner the user interface was specified and prototypically implemented. A variety of visualisation techniques, ranging from simple one-dimensional dia-grams to complex 3D models, was explored and the impact of several navigation techniques was tested.
Finally, the developed client/server prototype was evaluated. Using special test scenarios it was demonstrated that the implemented algo-rithms yield satisfying results. In addition, benefits and drawbacks of the prototypical user interface were identified through user testing and results were integrated into the design proposal for the Apa Online Manager.

top

V-Markt - Study on the Market Demand for Visual Tools for Accessing Complex Knowledge Domains

Strategic Research Project

The inceasing complexity of information and knowledge spaces in the last years has created new research challenges, with visual access systems being a particularly promising field. WebRat, a prototype visualisation system, adressing some of these problems, was implemented in the year 2002. The system was presented in a number of publications and definitely generated interest. The aim of the V-Markt project is to study the market and discover if and what kinds of visual tools for accessing knowledge repositories are requested on the market.

top