Text this: A performance comparison of euclidean, manhattan and minkowski distances in k-means clustering