|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
""" |
|
We introduce BIOMRC, a large-scale cloze-style biomedical MRC dataset. Care was taken to reduce noise, compared to the |
|
previous BIOREAD dataset of Pappas et al. (2018). Experiments show that simple heuristics do not perform well on the |
|
new dataset and that two neural MRC models that had been tested on BIOREAD perform much better on BIOMRC, indicating |
|
that the new dataset is indeed less noisy or at least that its task is more feasible. Non-expert human performance is |
|
also higher on the new dataset compared to BIOREAD, and biomedical experts perform even better. We also introduce a new |
|
BERT-based MRC model, the best version of which substantially outperforms all other methods tested, reaching or |
|
surpassing the accuracy of biomedical experts in some experiments. We make the new dataset available in three different |
|
sizes, also releasing our code, and providing a leaderboard. |
|
""" |
|
|
|
import itertools as it |
|
import json |
|
|
|
import datasets |
|
|
|
from .bigbiohub import qa_features |
|
from .bigbiohub import BigBioConfig |
|
from .bigbiohub import Tasks |
|
|
|
_LANGUAGES = ["English"] |
|
_PUBMED = True |
|
_LOCAL = False |
|
_CITATION = """\ |
|
@inproceedings{pappas-etal-2020-biomrc, |
|
title = "{B}io{MRC}: A Dataset for Biomedical Machine Reading Comprehension", |
|
author = "Pappas, Dimitris and |
|
Stavropoulos, Petros and |
|
Androutsopoulos, Ion and |
|
McDonald, Ryan", |
|
booktitle = "Proceedings of the 19th SIGBioMed Workshop on Biomedical Language Processing", |
|
month = jul, |
|
year = "2020", |
|
address = "Online", |
|
publisher = "Association for Computational Linguistics", |
|
url = "https://www.aclweb.org/anthology/2020.bionlp-1.15", |
|
pages = "140--149", |
|
} |
|
""" |
|
|
|
_DATASETNAME = "biomrc" |
|
_DISPLAYNAME = "BIOMRC" |
|
|
|
_DESCRIPTION = """\ |
|
We introduce BIOMRC, a large-scale cloze-style biomedical MRC dataset. Care was taken to reduce noise, compared to the |
|
previous BIOREAD dataset of Pappas et al. (2018). Experiments show that simple heuristics do not perform well on the |
|
new dataset and that two neural MRC models that had been tested on BIOREAD perform much better on BIOMRC, indicating |
|
that the new dataset is indeed less noisy or at least that its task is more feasible. Non-expert human performance is |
|
also higher on the new dataset compared to BIOREAD, and biomedical experts perform even better. We also introduce a new |
|
BERT-based MRC model, the best version of which substantially outperforms all other methods tested, reaching or |
|
surpassing the accuracy of biomedical experts in some experiments. We make the new dataset available in three different |
|
sizes, also releasing our code, and providing a leaderboard. |
|
""" |
|
|
|
_HOMEPAGE = "https://github.com/PetrosStav/BioMRC_code" |
|
|
|
_LICENSE = "License information unavailable" |
|
|
|
_BASE_URL = "https://huggingface.co/datasets/biomrc/resolve/main/data/" |
|
_URLS = { |
|
"large": { |
|
"A": { |
|
"train": _BASE_URL + "biomrc_large/dataset_train.jsonl.gz", |
|
"val": _BASE_URL + "biomrc_large/dataset_val.jsonl.gz", |
|
"test": _BASE_URL + "biomrc_large/dataset_test.jsonl.gz", |
|
}, |
|
"B": { |
|
"train": _BASE_URL + "biomrc_large/dataset_train_B.jsonl.gz", |
|
"val": _BASE_URL + "biomrc_large/dataset_val_B.jsonl.gz", |
|
"test": _BASE_URL + "biomrc_large/dataset_test_B.jsonl.gz", |
|
}, |
|
}, |
|
"small": { |
|
"A": { |
|
"train": _BASE_URL + "biomrc_small/dataset_train_small.jsonl.gz", |
|
"val": _BASE_URL + "biomrc_small/dataset_val_small.jsonl.gz", |
|
"test": _BASE_URL + "biomrc_small/dataset_test_small.jsonl.gz", |
|
}, |
|
"B": { |
|
"train": _BASE_URL + "biomrc_small/dataset_train_small_B.jsonl.gz", |
|
"val": _BASE_URL + "biomrc_small/dataset_val_small_B.jsonl.gz", |
|
"test": _BASE_URL + "biomrc_small/dataset_test_small_B.jsonl.gz", |
|
}, |
|
}, |
|
"tiny": { |
|
"A": {"test": _BASE_URL + "biomrc_tiny/dataset_tiny.jsonl.gz"}, |
|
"B": {"test": _BASE_URL + "biomrc_tiny/dataset_tiny_B.jsonl.gz"}, |
|
}, |
|
} |
|
|
|
_SUPPORTED_TASKS = [Tasks.QUESTION_ANSWERING] |
|
|
|
_SOURCE_VERSION = "1.0.0" |
|
|
|
_BIGBIO_VERSION = "1.0.0" |
|
|
|
|
|
class BiomrcDataset(datasets.GeneratorBasedBuilder): |
|
"""BioMRC: A Dataset for Biomedical Machine Reading Comprehension""" |
|
|
|
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION) |
|
BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION) |
|
|
|
BUILDER_CONFIGS = [] |
|
|
|
for biomrc_setting in ["A", "B"]: |
|
for biomrc_version in ["large", "small", "tiny"]: |
|
subset_id = f"biomrc_{biomrc_version}_{biomrc_setting}" |
|
BUILDER_CONFIGS.append( |
|
BigBioConfig( |
|
name=f"{subset_id}_source", |
|
version=SOURCE_VERSION, |
|
description=f"BioMRC Version {biomrc_version} Setting {biomrc_setting} source schema", |
|
schema="source", |
|
subset_id=subset_id, |
|
) |
|
) |
|
BUILDER_CONFIGS.append( |
|
BigBioConfig( |
|
name=f"{subset_id}_bigbio_qa", |
|
version=BIGBIO_VERSION, |
|
description=f"BioMRC Version {biomrc_version} Setting {biomrc_setting} BigBio schema", |
|
schema="bigbio_qa", |
|
subset_id=subset_id, |
|
) |
|
) |
|
|
|
DEFAULT_CONFIG_NAME = "biomrc_large_B_source" |
|
|
|
def _info(self): |
|
if self.config.schema == "source": |
|
features = datasets.Features( |
|
{ |
|
"abstract": datasets.Value("string"), |
|
"title": datasets.Value("string"), |
|
"entities_list": datasets.features.Sequence( |
|
{ |
|
"pseudoidentifier": datasets.Value("string"), |
|
"identifier": datasets.Value("string"), |
|
"synonyms": datasets.Value("string"), |
|
} |
|
), |
|
"answer": { |
|
"pseudoidentifier": datasets.Value("string"), |
|
"identifier": datasets.Value("string"), |
|
"synonyms": datasets.Value("string"), |
|
}, |
|
} |
|
) |
|
elif self.config.schema == "bigbio_qa": |
|
features = qa_features |
|
else: |
|
raise NotImplementedError() |
|
|
|
return datasets.DatasetInfo( |
|
description=_DESCRIPTION, |
|
features=features, |
|
homepage=_HOMEPAGE, |
|
license=str(_LICENSE), |
|
citation=_CITATION, |
|
) |
|
|
|
def _split_generators(self, dl_manager): |
|
"""Returns SplitGenerators.""" |
|
|
|
_, version, setting = self.config.subset_id.split("_") |
|
downloaded_files = dl_manager.download_and_extract(_URLS[version][setting]) |
|
|
|
if version == "tiny": |
|
return [ |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
gen_kwargs={"filepath": downloaded_files["test"]}, |
|
), |
|
] |
|
else: |
|
return [ |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
gen_kwargs={"filepath": downloaded_files["train"]}, |
|
), |
|
datasets.SplitGenerator( |
|
name=datasets.Split.VALIDATION, |
|
gen_kwargs={"filepath": downloaded_files["val"]}, |
|
), |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TEST, |
|
gen_kwargs={"filepath": downloaded_files["test"]}, |
|
), |
|
] |
|
|
|
def _generate_examples(self, filepath): |
|
"""Yields examples as (key, example) tuples.""" |
|
|
|
if self.config.schema == "source": |
|
with open(filepath, encoding="utf-8") as fp: |
|
for _id, line in enumerate(fp): |
|
example = json.loads(line) |
|
example["entities_list"] = [ |
|
self._parse_dict_from_entity(entity) for entity in example["entities_list"] |
|
] |
|
example["answer"] = self._parse_dict_from_entity(example["answer"]) |
|
yield _id, example |
|
elif self.config.schema == "bigbio_qa": |
|
with open(filepath, encoding="utf-8") as fp: |
|
uid = it.count(0) |
|
for _id, line in enumerate(fp): |
|
example = json.loads(line) |
|
|
|
|
|
example = { |
|
"id": next(uid), |
|
"question_id": next(uid), |
|
"document_id": next(uid), |
|
"question": example["title"], |
|
"type": "multiple_choice", |
|
"choices": [x.split(" :: ")[0] for x in example["entities_list"]], |
|
"context": example["abstract"], |
|
"answer": [example["answer"].split(" :: ")[0]], |
|
} |
|
yield _id, example |
|
|
|
def _parse_dict_from_entity(self, entity): |
|
if "::" in entity: |
|
pseudoidentifier, identifier, synonyms = entity.split(" :: ") |
|
return { |
|
"pseudoidentifier": pseudoidentifier, |
|
"identifier": identifier, |
|
"synonyms": synonyms, |
|
} |
|
else: |
|
return {"pseudoidentifier": entity, "identifier": "", "synonyms": ""} |
|
|