Level of Data - A Concept for Knowledge Discovery in Information Spaces

The visualization of large volumes of abstract information requires mechanisms to support the user by knowledge discovering. Therefore we developed the level of data concept. Therein the abstract nature of the data and the diverse user community has been taken into account. By using the iceberg metaphor, we classified the information space into several layers. The different backgrounds and analysis purposes of the users yield various views on the data. With these different perspectives on the iceberg, the levels of data can be pointed out. For proving the level of data concept we developed a system for the interactive visualization of shares prices as a representative example of abstract information. With the integration of this concept it is possible that any user can perform any analysis within single system. 1 INTRODUCTION Information visualization is a new research and development focus in the visualization community. The main difference between scientific visualization and information visualization is the kind of data, on which they operate (a detailed discussion can be found in [5]). In the application area of information visualization you have to take into consideration that the interesting data are mostly abstract, like information structures and relationships. Herewith the fundamental problem of information visualization is pointed out. This special class of data has no natural and obvious physical representation, so that intuitive compre-hensible visual metaphors have to be developed. A second problem in the information visualization, in contrast to the scientific visualization, is the widespread area of applications. This implies, that a diverse user community has to be considered, different levels of background and analysis purposes. Several approaches exist to represent multivariate relations and information in general, like the 'worlds-within-worlds' approach [2], the 'Information Visualizer' [9], the MineSet tool [1] and the work of Risch [7]. However, the aim of the visualization of information is not only to display as much data as possible [10], but also to perform a data preprocessing to provide an analysis and exploration of the information. The work of Campo [3] with automatic management of abstraction levels and the degree of interest function from [4] are first steps in this direction. To carry on these ideas we introduce a concept for a method to prepare and visualize the information in a way, that it is tailored to every kind of user. Our level of data concept will support the user by knowledge discovering in information spaces. Hence …