Create vectors for classification
Opened this issue · 1 comments
GoogleCodeExporter commented
I'm trying to use word2vec in order to change a bunch of text into vectors with
the use of word2vec.
Ultimately, I want to do this so I can go from text -> vectors ->
classification through the usage of a classification algorithm like a neural
network. Any reading or testing of word2vec, however, does not let me output a
vectorized version of my input text data, but, rather, I can just do word
similarity or some other feature like that.
Am I completely off base with the potential usage of word2vec or is this
actually possible?
Original issue reported on code.google.com by clayclay...@gmail.com on 1 Jun 2015 at 2:37
GoogleCodeExporter commented
Sure this is possible, but you will need to change the code a bit. Simply,
instead of having objective to predict nearby words, you can sum the word
vectors in a sentence (like in CBOW) and predict the label. It should work
well, if you train it for sufficient number of iterations.
Original comment by tmiko...@gmail.com on 2 Jun 2015 at 7:16