Text Mining Action Set

Topics are defined by the U matrix that contains orthogonal dimensions produced by the SVD. As in factor analysis, you can improve your topic interpretability by using a rotation to increase the weights for the most important terms in your topic. Two types of rotations are available: varimax and promax. They are borrowed from the FACTOR procedure and are described in the section Simplicity Functions for Rotations (SAS/STAT User's Guide).