Sequential-Sentence-Classification

code for our paper Sequential Sentence Classification with a Boundary-aware Dual Biaffine Model for Biomedical Documents

Requirement

All the requirements are listed in the requirement.txt

Dataset

Pubmed-pico

This dataset is introduced by Jin, Di, and Peter Szolovits. "PICO Element Detection in Medical Text via Long Short-Term Memory Neural Networks." Proceedings of the BioNLP 2018 workshop. 2018. And it can be find in github from "PubMed PICO Element Detection Dataset"

Pubmed-rct

This dataset is introduced by Franck Dernoncourt, Ji Young Lee. PubMed 200k RCT: a Dataset for Sequential Sentence Classification in Medical Abstracts. International Joint Conference on Natural Language Processing (IJCNLP). 2017. And it can be find in github from "PubMed 200k RCT dataset"

Run

The command for runing the model on pubmed-rct dataset allennlp train config/biobert_biaffine_pubmed-rct.jsonnet -s saved_models/pubmed-rct/ --include-package codes --force .

The command for runing the model on pubmed-pico dataset allennlp train config/biobert_biaffine_pubmed-pico.jsonnet -s saved_models/pubmed-pico/ --include-package codes --force .

After training, the results can be found in folder 'saved_models/pubmed-rct/' and 'saved_models/pubmed-pico/' respectivly.