Group project for the NLP subject in our AI degree.

• data: contains the training and test data, as well as preprocessed data useful for our models.
• latex: necessary files used for the creation of the report in latex.
• models: last trained checkpoints of the CRF and LSTM models.
• src: contains all the python scripts and jupyter notebooks.
  • preprocesing.py: functions for data preprocessing, incuding tokenization and lemmatization.
  • blindNegex.py: code for the Full Scope NegEx, the baseline model.
  • negex.py: implementation of NegEx as proposed in the original paper, using medical terms extracted from the training set.
  • crf0.py: code for the first implementation of CRF.
  • crf.py: code for the second and official implementation of CRF.
  • lstm.py: code for the BiLSTM implementation.
  • eval.py: code to evaluate the model predictions
  • .ipynb files: python scripts used during the development of this project, most of them to test and demonstrate the funcitonality of their corresponding .py files.

• Amritpal Singh
• Oscar Arrocha
• Mustapha El Aichouni
• Eric López