Singular Value Decomposition

The final step

m x n content-word-transcript matrix A, of rank r, can be expressed as the product:

A = U * S * V^T

U is m x r content-word matrix

S is r x r diagonal matrix

V^T is r x n transcript matrix