ETEREX-REBEL

End-to-end Temporal Relation Extraction in the clinical domain as a sequence-to-sequence task with the REBEL framework

This repository contains the code for the modelling and evaluation of our BART-based model that is capable of performing end-to-end temporal relation extraction in clinical narratives as a sequence-to-sequence task. This work was presented in Text2Story 2023.

The pre-trained model and part of the code are based on the REBEL framework (Cabot, 2021). The corpus was retrieved from the i2b2 2012 Temporal Relation Task (Sun, 2013).

References

Huguet Cabot, P.-L., & Navigli, R. (2021). REBEL: Relation Extraction By End-to-end Language generation. Findings of the Association for Computational Linguistics: EMNLP 2021, 2370–2381. doi: https://doi.org/10.18653/v1/2021.findings-emnlp.204

Sun, W., Rumshisky, A., & Uzuner, O. (2013). Evaluating temporal relations in clinical text: 2012 i2b2 Challenge. In Journal of the American Medical Informatics Association (Vol. 20, Issue 5, pp. 806–813). Oxford University Press (OUP). https://doi.org/10.1136/amiajnl-2013-001628

Citation

Please cite the paper if you use this resource in any way:

@article{saiz2023end, title={End-to-End Temporal Relation Extraction in the Clinical Domain [FULL]}, author={Saiz, Jos{'e} Javier and Altuna, Bego{~n}a}, year={2023} }

License

ETEREX-REBEL is licensed under the CC BY-SA-NC 4.0 license. The text of the license can be found here.