/anlp_rbert

Entity Relation Extraction using R-BERT

Primary LanguageJupyter Notebook

anlp_rbert

Entity Relation Extraction using R-BERT

In the google colab notebook, entity relations are extracted from medical data about diseases. Bert is customized for the task, using methods from the following paper:

Enriching Pre-trained Language Model with Entity Information for Relation Classification https://arxiv.org/abs/1905.08284.

The code for the implementation was borrowed from: https://github.com/wang-h/bert-relation-classification

This custom R-Bert model is then fine tuned for our data and used to predict what kind of relationships hold between entities in our test set.

Data: https://www.kaggle.com/kmader/figure-eight-medical-sentence-summary

TRAINED MODEL: https://drive.google.com/open?id=1D12EiIglMuyRqQB1wKpddSl80KHI4N7N