Text embedding methods at different levels for tweets text classification
To test my implementation, go to MyImplementation and download all the files Run the AAc for Convolutional Neural Network (CNN) based on character embedding, AAs for subword embedding, AAw for word embedding.
To test the byte-pair-encoding algorithm, run the bash file bpeAA. (by Sennrich https://github.com/rsennrich/subword-nmt)
The Prasha's Implementation requires Keras==1.0 and theano backend to run. Make sure the dependencies are satisfied.