fredymad / Transformers-For-Negation-and-Speculation

This is the code for Negation and Speculation Cue Detection and Scope Resolution using BERT, XLNet and RoBERTa

Transformers-For-Negation-and-Speculation

1. Transformers_for_Negation_and_Speculation.ipynb:

This is the code for the following papers:

• NegBERT: A Transfer Learning Approach for Negation Detection and Scope Resolution (Published at LREC 2020)
• Resolving the Scope of Speculation and Negation using Transformer-Based Architectures

2. Multitask_Learning_of_Negation_and_Speculation.ipynb:

This is the code for the following paper:

• Multitask Learning of Negation and Speculation using Transformers (Published at LOUHI 2020)

Code has been utilized from the following sources:

• The starter code was taken from this article on Named Entity Recognition with Bert.
• To be able to implement XLNetForTokenClassification and RobertaForTokenClassification for the Transformers library by Huggingface, some code was copy-pasted from that code base.
• The implementation of Early Stopping for PyTorch was adapted from Bjarten's implementation(GitHub).

The colab notebook can be executed directly from Google Colaboratory.

Datasets:

1. BioScope Corpus
2. Sherlock (*sem 2012 Shared Task)
3. SFU Review Corpus

Contributors: Aditya Khandelwal, Benita Kathleen Britto