I have understood K-means algorithm, I wonder how does this algorithm applied to cluster products in amazon (I want to know technicality not the theory)