The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider
removing the
loading script
and relying on
automated data support
(you can use
convert_to_parquet
from the datasets
library). If this is not possible, please
open a discussion
for direct help.
Dataset Card for MEDIQA NLI
Natural Language Inference (NLI) is the task of determining whether a given hypothesis can be inferred from a given premise. Also known as Recognizing Textual Entailment (RTE), this task has enjoyed popularity among researchers for some time. However, almost all datasets for this task focused on open domain data such as as news texts, blogs, and so on. To address this gap, the MedNLI dataset was created for language inference in the medical domain. MedNLI is a derived dataset with data sourced from MIMIC-III v1.4. In order to stimulate research for this problem, a shared task on Medical Inference and Question Answering (MEDIQA) was organized at the workshop for biomedical natural language processing (BioNLP) 2019. The dataset provided herein is a test set of 405 premise hypothesis pairs for the NLI challenge in the MEDIQA shared task. Participants of the shared task are expected to use the MedNLI data for development of their models and this dataset was used as an unseen dataset for scoring each participant submission.
Citation Information
@misc{https://doi.org/10.13026/gtv4-g455,
title = {MedNLI for Shared Task at ACL BioNLP 2019},
author = {Shivade, Chaitanya},
year = 2019,
publisher = {physionet.org},
doi = {10.13026/GTV4-G455},
url = {https://physionet.org/content/mednli-bionlp19/}
}
- Downloads last month
- 51