KNOWLEDGE VISUALIZATION – Seeing and understanding
Gaining insights into relationships hidden in complex data is part of our daily work. In the area "Knowledge Visualization" we develop innovative methods of visualization in order to present and communicate complex information and interrelationships in a clear and compelling way.
The area of Knowledge Visualization develops cutting-edge visualization methods for analyzing large amounts of data and helping users understand complex relationships between them. Our goal is to bring together various sources of information via meaningful visualizations, to promote visual thinking, and guide the sense-making process. By combining interactive visualization with automatic analysis, we facilitate the discovery of unexpected phenomena, verification of known facts and deriving of new knowledge out of massive data. Our visual analytics technologies guide the user through the process of knowledge generation by providing best visual means for acquiring information from the analytical process and initiating new analytical steps based on the user’s current needs. Moreover, we strive to make knowledge accessible at all times, to deliver it on time and in the format that enhances the individual’s potential to analyze and explore interconnections. A myriad of factors, such as data characteristics, user profile, the task at hand and the device capabilities, must be taken into account to deliver knowledge in the visual form that is easily understood and smoothly integrated into the individual’s thinking process. Augmented reality and other ubiquitous computing technologies provide knowledge at the right moment and seamlessly integrate it into the working scenario of the user.
Within the framework of the CODE project, we developed user-friendly web-based interfaces for discovery, exploration and visual analysis of Linked Open Data (LOD). Although LOD is a great source of semantic information, it remains underutilized due to difficulties in accessing the data. Our Query Wizard interface makes search and selection in LOD as simple as a Google search and spreadsheet manipulation. The selected data set can be visualized with a single mouse click by using the Visualization Wizard, which automatically selects the most appropriate visual representation.
In EEXCESS we investigate adaptive visual interfaces for recommender systems. While recommendation engines provide relevant user context data (i.e., task, preferences, etc.), our visualization engine suggests visual representations suitable for the data. Additionally, the visualization engine configures visual characteristics (UI layout, sizes, colors, etc.) to deliver the best possible visual experience.
WebGraph is a Web-based interactive graph visualization which employs geometry optimizations and data aggregation techniques to maintain the clarity of the representation, which otherwise may be compromised by too many nodes and overlapping edges. Interaction techniques that utilize semantic and structural information contained in the data intelligently support navigation and exploration of knowledge space. WebGraph is applied in several projects with such partners as CDS or AutomationX.
Within the DIVINE project we have developed a web-based dynamic topography knowledge map visualization (also known as information landscape) and applied it to dynamic media repositories to analyze changes in their thematic structure. Moreover, we developed algorithms for thematic cluster analysis that can capture changes in growing collections of documents.