Why use dot as a measure of similarity？

[celia01]

in utils.py, why use the embedding's dot in the function getSimilarity,

[suanrong]

They use dot in the paper to measure similarity, while the loss function optimize the Euclid distance. It makes me confused.

[lijunweiyhn]

Sorry, but where indicate the calculation of similarity in the original paper? I saw the code use dot, too. Thanks

[lijunweiyhn]

Sorry, but where indicate the calculation of similarity in the original paper? I saw the code use dot, too. Thanks

[suanrong]

Sorry, the author did not mention in the paper. I once asked him and he told me he have tried using Euclidean distance to measure similarity but get worse performance.

[lijunweiyhn]

Thanks, so am I. From the loss function, Euclidean distance may be the most reasonable way, but when I get the embedding to link prediction task(my own dataset), the results were embarrassing.