Projekte

Users of Information Retrieval systems have often been the target group of Human-Computer Interaction researchers. A lot of effort has been spent inventing new forms of visualizations to support the information seeking process. Information Retrieval and Information Visualization are tight coupled fields of research. Together with psychology (which answers questions like "how" do users search) and usability engineering (answering questions like "what" do user expect from user interfaces and their behaviour) the research on improving information seeking systems goes on. This paper will concentrate on a meta-data driven, user-centered approach for the query formulation stage. In contrast to the intense research on result-set visualizations we will focus on the development of a visualization which supports human search behaviour at the query stage. Additionally this visualization proved that it can compete with other visualizations like the scatter-plot as a visual filter in the result-set presentation.

The video-paper describes a java prototype of a generic visual information retrieval system called VisMeB, developed to visualize metadata within various contexts. We reconsidered common visualization strategies and applied them to different domains. The system we describe offers visual assistance in exploring large amounts of datasets from different application domains like geographical metadata, movie databases and the Web. The video demonstrates a scenario using a movie database.

This paper presents a new framework for Metadata Visualization Systems called VisMeB (Visual Metadata Browser). It is based upon redesign ideas from the INSYDER System that were come under extensive evaluations. The aim of our approach is to improve the process of finding relevant information in an intuitive yet multifunctional way. We use a ScatterPlot in combination with a so called SuperTable for visualization. The two techniques are tightly coupled and present unique possibilities of interaction through the use of visual filters.

Es wird ein visuelles Suchsystem zur explorativen Erkundung großer Datenmengen vorgestellt. Das System VisMeB (Visueller Metadaten-Browser) basiert auf einer Reihe von Designstudien, die im Rahmen unterschiedlicher Forschungsprojekte entwickelt worden sind. Der besondere Schwerpunkt lag dabei auf der Entwicklung innovativer Visualisierungs- und Interaktionstechniken, die den Benutzer von der Anfrageformulierung bis zur Ergebnisdarstellung unterstützen. Die anhand eines szenarienbasierten Usability Engineering Ansatzes entwickelten Designstudien wurden einer Reihe von Evaluationen unterzogen. Daraus ergaben sich zahlreiche Redesignvorschläge, die bereits umgesetzt wurden.

This paper will describe ongoing efforts to ease the information retrieval process on metadata, using a SuperTable/Scatterplot framework recently named VisMeB (Visual Metadata Browser). Based on the combination of a SuperTable (in two design variants) and a Scatterplot (also in two design variants), users can follow different search strategies to achieve results. Usability testing has showen, that an integration of our proposed visualizations supports most user s search style.

This paper will present a new visual information retrieval system for metadata and the interaction techniques of the new system. The abundance of information we get while analyzing search results of an arbitrary query has to be channeled. This can be done by different visualizations and filter techniques. We use a Scatterplot in combination with a so called SuperTable to support the process of finding relevant information in an intuitive yet multifunctional way. Visual filters and the interaction between the visualiza-tions play an important role. By examples from a web and a movie database search fea-tures are demonstrated.