Trying to implement a Recurrent Convolutional Neural Networks for Discourse Compositionality described in http://nal.co/papers/KalchbrennerBlunsom_CVSC13
still in process
numpy
theano
gensim
python-dateutil
nltk
It's able to get the swda dataset from http://compprag.christopherpotts.net/swda.html Notice that swap.py is modified a little in this project to make sure it runs correctly.