Concept Space Concept Space is a set of concept-based search tools [ITO Demo] SOM: Self-Organizing Map SOM is a unsupervised neural network technique to analyze textual or image data. Demo in 2D [Java], 3D VRML (requires Netscape 4+ with VRML) Tile Visual Thesaurus The foreground image is a texture-based visual thesaurus of aerial photographs that include text labels obtained from a digital gazetteer. The small inset window shows related tiles. The background image shows an aerial photo, which indicates the selected texture tile (highlighted in red) and the related tiles (highlighted in blue). Demo Vegetation and Temperature Thesaurus The foreground color map shows an SOM Categorization with the AVHRR data. The color shows the vegetation density and the label indicates temperature. By clicking the top-left region outlined by the pink rectangle, locations in California with these parameters are shown as pink dots. Demo MDS: Multi-Dimensional Scaling Multidimensional Scaling (MDS) is being used to categorize 200 comments from a Group Systems electronic meeting session. The MDS display uses color-coded dots to identify the distribution of the comments from the session. MDS displays have been developed using both automatic indexing and the Arizona Noun Phraser to identify key words and phrases within session comments. User studies are currently being conducted to compare user analysis to computer generated output. Internet Agent The Itsy Bitsy Spider is a dynamic Internet agent. It takes keywords and representative URLs from the user and searches the Internet for the most similar homepages based on links and keyword indexing. Itsy Bitsy Spider uses a genetic algorithm with cross-over operations and mutation procedures to find the best matches.