Measuring Similarity of Two Embeddings

When assessing the similarity of two vectors, is it correct to sum the differences between each value pair and treat this as a distance measure? For example:

let v1 = [.1,.2,.3…];
let v2 = [.1,-.5,3…];
let dist = 0.0;
for (let i = 0; i<v1.length; i++){
dist + = Math.Abs(v1[i] - v2[i]);
}

Is dist a measure of the semantic difference between the two embeddings?

You need to use dot product How the Dot Product Measures Similarity | by Tivadar Danka | Towards Data Science

2 Likes