Skip to main content
Data Science

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

2
  • $\begingroup$ Thank you for your time. But I don't think the question means "Why should we use Cosine similarity" so using it is not always a good choice. Especially that embeddings are produced by some deep neural network. I don't think their embeddings are related to each other in some metric space, except the model has been trained to produce meaningful cosine embeddings space. In other words, I think It's quite none sense if we try to get some raindom embeddings produced by a model then compute cosine distance. Then the number is meaningless. but for some statistical embedding method like TF-IDF $\endgroup$ Commented Aug 22, 2024 at 3:08
  • $\begingroup$ then yes cosine metric is proved to be more effective than Euclidean distance $\endgroup$ Commented Aug 22, 2024 at 3:08