After training a model using word2vec I'd now like to store the trained model with the word serving as a key and the vector as its value. However I'm not sure how I'll be able to implement in this way a k-nearest neighbour search. What would be the correct way to get around this?
$\begingroup$
$\endgroup$
$\begingroup$
$\endgroup$
A most popular way of obtaining the approximate nearest neighbors is the locality sensitive hash. And here are some practical results. Then once you have the neighboring keys, it's straightforward to use a key-value store to retrieve the corresponding words.
