Data Science Stack Exchange is a question and answer site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field. It only takes a minute to sign up.
Sign up to join this community
Word2Vec generates a feature vector for each word. I was wondering if exist something2vec that generates one feature vector for one document. In this way I could compute the distance between two documents by only comparing two feature vectors.
Thanks.
Yes, there are such things, but they aren't really related to word2vec.
First of all, Bag of Words is the old "something2vec" approach of representing documents in a vector space. Of course, the resulting vectors are very sparse, and distance does not always behave well in such cases, and it is better to use dot-product based similarities like cosine.
But you can reduce the dimensionality of this vector space with SVD or NMF (Non-Negative Matrix Factorization), and this will give you what you want: dense vectors of small dimensionality for which you can compute the distances.
Then you can go further and use Topic Modelling, which also will give you a vectorial representation of documents, and you can use it for comparing them.
how about Doc2vec. Mikolov released the journal after the successful attempt of word2vec. A single feature vector for a whole paragraph or sentence or document (depending upon your use case) could be generated. Personally, I didn't get quite good results with it for sentence classification problems. Rather I went with averaged word vectors for sentences.
