APPLICATIONS: Computer & Information Sciences Large Datasets & Visualization

TeraScope: Visual Tera Mining

TeraScope is a massively parallelized set of information visualization tools for Visual Data Mining. It consists of new algorithms and tools to interactively query and mine terabyte datasets, correlate the data, and then visualize the data using parallelized rendering software on tiled displays. TeraScope’s main foci are to develop techniques for creating TeraMaps (visualizations that summarize rather than plot enormous datasets) and to develop a distributed memory cache that collects pools of memory from optically connected computer clusters. These caches are used by TeraScope to bridge the impedance mismatch between large and slow distributed data stores and fast local memory. TeraScope currently works with Project DataSpace’s distributed data servers; however, it can be adapted to work with other database systems. Acknowledgment: DataSpace, developed by UIC/s NCDM; USA National Science Foundation awards EIA-9802090, ANI-9730202, ANI-0129527 and ACI-9619019 (to the National Computational Science Alliance); and, Quanta. Contact Jason Leigh Electronic Visualization Laboratory (EVL) University of Illinois at Chicago (UIC), USA spiff@evl.uic.edu Collaborators Jason Leigh, Charles Zhang, EVL at UIC, USA Robert Grossman, Sivakumar Harinath, Marco Mazzucco, National Center for Data Mining (NCDM) at UIC, USA www.evl.uic.edu/cavern/teranode/terascope www.dataspaceweb.net