Open psp3dcg opened 4 years ago
embedding of phrase is not really a thing. Look at how gensim does paragraph2vec, as far as I know, it is the "average" of the words.
I would suggest using the vectors in https://github.com/commonsense/conceptnet-numberbatch
How to get the embedding vector of a word or phrase directly instead of the similarity of two terms? I have seen the document of web api and there seem no desciption of getting the embedding of word or phrase