The article mentions equivalent ranking from cosine similarity and Euclidean distance. The derivation is very simple. For vectors A and B, the squared Euclidean

The article mentions equivalent ranking from cosine similarity and Euclidean distance. The derivation is very simple. For vectors A and B, the squared Euclidean distance is: (A-B).(A-B) = A.A-2A.B+B.B A and B only interact through a dot product, just like cosine similarity. If A and B are normalized, A.A=B.B=1. For Pearson Correlation, we would just need to center and scale A and B as a pre-processing step.