Nonnegative Matrix and Tensor Factorization for Discussion Tracking - Text Mining: Classification, Clustering, and Applications - page 111

This tool has a mouseover feature that enables a user to hold the mouse
over any pixel in the matrix (reordered or otherwise) and determine which
term and which document the pixel corresponds to. Figure 5.9 is a screenshot
from the vismatrix tool.
FIGURE 5.9 (SEE COLOR INSERT FOLLOWING PAGE 130.) :
Pixel plot of the reordered Enron term-by-document matrix with term and
document labels.
Notice the upper lefthand corner contains the word touchdown ,which
represents the term (term ID#6635) being pointed to, and the identifier
dean-cinfo84 , which represents the document ID. This document, document
3819, was email message #84 saved by an Enron employee named Dean in
his cinfo folder. Scrolling over pixels in a dense block causes the term and
document labels to change in this area of the vismatrix tool. The human brain
can quickly process many terms at once. As a result, the user can attach a