Transfer-learning is one possible approach :

 1. Design and implement a neural net to match [Google Word2Vec's design][1]
 (In terms of number of layers, activation functions and etc.,). 
 2. Pre-initialize weights with these vectors 
 3. Retrain with domain-specific corpus 

[This][2] is an implementation that can be used as base and modified for step #1 


 [1]: https://github.com/dav/word2vec/blob/master/src/word2vec.c
 [2]: https://github.com/tensorflow/models/blob/master/tutorials/embedding/word2vec.py