K-Means - no change to inter-cluster density (precision)

Question

I have ran a k-means algorithm on a corpus of classical literature documents using Apache Mahout. The aim is to find the best parameters for the algorithm by altering the di