Due to technological advances in the acquisition, storage and transmission of data, ever increasing large data repositories are emerging. In areas such as engineering, business and science, large amounts of potentially relevant data are produced. Recently, institutional providers allow public access to large repositories of data emerging from public research, e.g., via data repositories operated by libraries or individual research institutions. Also, public administration data of all sorts is increasingly published by public bodies on all administrative levels, following the so-called open data trend. To appropriately access and understand such data, a novel search and analysis methodology is required to make best use of such data repositories. A main challenge herein is the fact that such data repositories typically are heterogeneous in nature, and not in all cases consistently annotated with metadata. Also, data quality and content may vary drastically among and within the respective repositories.pIn this project, we want to develop novel visual-interactive search and analysis approaches for large repositories of heterogeneous data from the research data and open data domains. We will focus on methods for time- and text-oriented data, considering the heterogeneous nature of the respective data sources. Following Shneiderman’s Visual Information Seeking Mantra, we will develop visual overview techniques which allow exploration and understanding of the structure and overall composition of respective data repositories. To that end, novel visual representations based on data clustering in heterogeneous data environments will be researched. We will also research novel visual search methods, in which users are enabled to navigate in heterogeneous data collections, and formulate search queries in an intuitive visual way, e.g., relying on sketch-based and query-by-example approaches. The developed methods will be exemplarily tested on selected research and open data repositories.
