Metadatengeleitete Navigation in bildbasierten wissenschaftlichen Experimentaldaten

While evaluating image based scientific experiments, experts must analyse and interprete large amounts of image and symbolic data. Visualizing experimental data supports domain experts in their evaluation tasks. Modern laboratory processes produce many images and measurements. Software supplying appropriate views of these data assists experts and enables the experiments' evaluation. The standard approach is developing individual software for each customer and domain. This development process is timeconsuming both for implementation itself and for communicating the detailed requirements to the development team. Additionally requirements change with the scientific progress in the application domains. This effort must be reduced. With this goal in mind, this thesis introduces BioZoom. BioZoom is an approach showing how a generic software solution supports the evaluation of image based scientific experiments. BioZoom consists of information zooming for fast and flexible view configuration by the end user and reuse of appropriate views. BioZoom uses metadata. Metadata describes entities and relationships from the application domain of image based experiments derivable from structured laboratory processes. With BioZoom, end users can compose views interactively with an information zooming tool and reuse task-relevant views. Users of BioZoom bookmark those views while exploring experimental data. While exploring users navigate from an experimental data overview to relevant subsets, further detailing the subsets in appropriate manner. Images are visually integrated with metadata - e.g. image position visualizes relationships between image data, and localisable entities are embedded in their image context. BioZoom generalizes bookmarked views with regard to the structure of the experimental data displayed within them. By reusing these view schemes users build a library of relevant view types. A view scheme will be instantiated in navigation automatically with the data available. Thus the user can easily and directly - starting from an initial data set - navigate to the view scheme required for the work task at hand. A large part of individual analysis software's functionality is realized in this end-user-development-approach. Illustrating this approach the thesis discusses two individual applications for visual evaluation and quality control drawn from an industry background. The applications are compared with BioZoom's views and navigational paths. It is shown that the generic end-user-development approach BioZoom can replace those specialised custom made applications.