Noise Reduction Methods for Distantly Supervised Biomedical Relation Extraction
Please cite the paper if you use the codes:
@INPROCEEDINGS{Li2017-zu, title = "Noise Reduction Methods for Distantly Supervised Biomedical Relation Extraction", booktitle = "Proceedings of the {BioNLP} 2017 workshop", author = "Li, Gang and Wu, Cathy H and Vijay-Shanker, K", pages = "184--193", year = 2017, conference = "BioNLP 2017" }
- python 2.7 (should work with 3.6 but not tested)
- bash terminal
- Download data:
wget http://biotm.cis.udel.edu/biotm/projects/ds/data.tar.gz
- Uncompress data folder:
tar -zxvf data.tar.gz
- Create a python virtual environment:
virtualenv env
- Activate virtual environment:
. env/bin/activate
- Install python modules:
pip install -r python_modules.txt
This is optional as all the result files are already included in this repo. Run the experiments if you want to regenerate them.
First activate the virtual environment: . env/bin/activate
sh script/single_instance.sh
sh script/multi_instance.sh
Figures are stored in eval/figures
python src/compute_scores.py
python src/draw_scores.py
python src/draw_curves.py