Looks like author is comparing document vectors using nearest neighbour. I thought cosine similarity was considered the best measurement for closeness of word

Looks like author is comparing document vectors using nearest neighbour. I thought cosine similarity was considered the best measurement for closeness of word and document vectors.