Visualization of a Text Network Structure Using X3D; Final rept. Apr 2007-Nov 2007

摘要

This report describes a dynamic extensible three-dimensional (3-D) (X3D) scene graph for visualizing text documents. It was developed at the U.S. Army Research Laboratory as a tool for social network analysis: specifically, to view a network of nodes where a node is a noun within a news article. Keyword and target nodes are represented as X3D spheres of different radii and color. Keyword node attraction to a specific target node is dynamic by a Java class access to scene content from an X3D script node. The magnitude of the direction vector, or speed, between keyword and target is currently proportional to frequency of the word within the document. This technique can be modified/replaced easily for a more comprehensive examination, such as with an algorithm that considers word selection across multiple documents. X3D code for a particular example is given and viewed using the Xj3D 2.0 browser from Yumetech, Inc.