This work is built on progressbar, Keras and Theano

The best way to configure the environment is to use Anaconda [https://www.continuum.io/downloads], after install Anaconda, then install the Theano [https://github.com/Theano/Theano] and Keras [https://github.com/fchollet/keras]

Here is the introduction of each file:
data directory: include the pre-trained word embedding
log directory: include the output log

the rest of .py files are used to train the model. 
for convenient, we preprocess the original DDI data, and transform each sentence into xml file, which could be found in xml directory

please contact me if you have any problem
my email: [email protected]