---

System Requrement:

python 2.7, theano, keras, gensim

sklearn     #for load KB features*

maybe acquired: h5py, libshortexts

---

How to use

1. We have extracted all (cheimcal,disease) pairs from BiocreativeV CDR data and stored them in "data/" such as "train_full" . Knowledge-Bases features ("*.svm") also store in that folder.

     "train_full" and "test_full": full sentences

     "train_only_between" and "test_only_between": segment between two entities.

2. Run:

   Test our model with KB features:

  python pos_fea_cnn.py data/train_full data/test_full

Test our model without KB features: 

python pos_cnn.py data/train_full data/test_full

Trainditional CNN model with KB features: 

python standard_fea_cnn.py data/train_only_between data/test_only_between

Trainditional CNN model without KB features: 

  python standard_cnn.py data/train_only_between data/test_only_between

Any questions pls connect [email protected]