23 October 2011
Exploratory data visualization calls for iterative analyses, but very large databases are often far too slow to allow interactive exploration. Incremental, approximate database queries exchange precision for speed: by sampling from the full database, the system can resolve queries rapidly. As the sample gets broader, the precision increases at the cost of time. As the precision of the sample value can be estimated, we can represent the range of possible values. This range may be visually represented using uncertainty visualization techniques. This paper outlines the current literature in both incremental approximate queries and in uncertainty visualization. The two fields mesh well: incremental techniques can collect data in interactive time, and uncertainty techniques can show bounded error.
|Published in||IEEE Symposium on Large Data Analysis and Visualization|