# coding=utf-8 # Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """ The main aim of MESINESP2 is to promote the development of practically relevant semantic indexing tools for biomedical content in non-English language. We have generated a manually annotated corpus, where domain experts have labeled a set of scientific literature, clinical trials, and patent abstracts. All the documents were labeled with DeCS descriptors, which is a structured controlled vocabulary created by BIREME to index scientific publications on BvSalud, the largest database of scientific documents in Spanish, which hosts records from the databases LILACS, MEDLINE, IBECS, among others. MESINESP track at BioASQ9 explores the efficiency of systems for assigning DeCS to different types of biomedical documents. To that purpose, we have divided the task into three subtracks depending on the document type. Then, for each one we generated an annotated corpus which was provided to participating teams: - [Subtrack 1 corpus] MESINESP-L – Scientific Literature: It contains all Spanish records from LILACS and IBECS databases at the Virtual Health Library (VHL) with non-empty abstract written in Spanish. - [Subtrack 2 corpus] MESINESP-T- Clinical Trials contains records from Registro Español de Estudios Clínicos (REEC). REEC doesn't provide documents with the structure title/abstract needed in BioASQ, for that reason we have built artificial abstracts based on the content available in the data crawled using the REEC API. - [Subtrack 3 corpus] MESINESP-P – Patents: This corpus includes patents in Spanish extracted from Google Patents which have the IPC code “A61P” and “A61K31”. In addition, we also provide a set of complementary data such as: the DeCS terminology file, a silver standard with the participants' predictions to the task background set and the entities of medications, diseases, symptoms and medical procedures extracted from the BSC NERs documents. """ import json import os from typing import Dict, List, Tuple import datasets from .bigbiohub import text_features from .bigbiohub import BigBioConfig from .bigbiohub import Tasks _LANGUAGES = ['Spanish'] _PUBMED = False _LOCAL = False _CITATION = """\ @conference {396, title = {Overview of BioASQ 2021-MESINESP track. Evaluation of advance hierarchical classification techniques for scientific literature, patents and clinical trials.}, booktitle = {Proceedings of the 9th BioASQ Workshop A challenge on large-scale biomedical semantic indexing and question answering}, year = {2021}, url = {http://ceur-ws.org/Vol-2936/paper-11.pdf}, author = {Gasco, Luis and Nentidis, Anastasios and Krithara, Anastasia and Estrada-Zavala, Darryl and Toshiyuki Murasaki, Renato and Primo-Pe{\~n}a, Elena and Bojo-Canales, Cristina and Paliouras, Georgios and Krallinger, Martin} } """ _DATASETNAME = "bioasq_2021_mesinesp" _DISPLAYNAME = "MESINESP 2021" _DESCRIPTION = """\ The main aim of MESINESP2 is to promote the development of practically relevant \ semantic indexing tools for biomedical content in non-English language. We have \ generated a manually annotated corpus, where domain experts have labeled a set \ of scientific literature, clinical trials, and patent abstracts. All the \ documents were labeled with DeCS descriptors, which is a structured controlled \ vocabulary created by BIREME to index scientific publications on BvSalud, the \ largest database of scientific documents in Spanish, which hosts records from \ the databases LILACS, MEDLINE, IBECS, among others. MESINESP track at BioASQ9 explores the efficiency of systems for assigning DeCS \ to different types of biomedical documents. To that purpose, we have divided the \ task into three subtracks depending on the document type. Then, for each one we \ generated an annotated corpus which was provided to participating teams: - [Subtrack 1 corpus] MESINESP-L – Scientific Literature: It contains all \ Spanish records from LILACS and IBECS databases at the Virtual Health Library \ (VHL) with non-empty abstract written in Spanish. - [Subtrack 2 corpus] MESINESP-T- Clinical Trials contains records from Registro \ Español de Estudios Clínicos (REEC). REEC doesn't provide documents with the \ structure title/abstract needed in BioASQ, for that reason we have built \ artificial abstracts based on the content available in the data crawled using \ the REEC API. - [Subtrack 3 corpus] MESINESP-P – Patents: This corpus includes patents in \ Spanish extracted from Google Patents which have the IPC code “A61P” and \ “A61K31”. In addition, we also provide a set of complementary data such as: \ the DeCS terminology file, a silver standard with the participants' predictions \ to the task background set and the entities of medications, diseases, symptoms \ and medical procedures extracted from the BSC NERs documents. """ _HOMEPAGE = "https://zenodo.org/record/5602914#.YhSXJ5PMKWt" _LICENSE = 'Creative Commons Attribution 4.0 International' _URLS = { _DATASETNAME: { "subtrack1": "https://zenodo.org/record/5602914/files/Subtrack1-Scientific_Literature.zip?download=1", "subtrack2": "https://zenodo.org/record/5602914/files/Subtrack2-Clinical_Trials.zip?download=1", "subtrack3": "https://zenodo.org/record/5602914/files/Subtrack3-Patents.zip?download=1", }, } _SUPPORTED_TASKS = [Tasks.TEXT_CLASSIFICATION] _SOURCE_VERSION = "1.0.6" _BIGBIO_VERSION = "1.0.0" class Bioasq2021MesinespDataset(datasets.GeneratorBasedBuilder): """\ A dataset to promote the development of practically relevant semantic indexing tools for biomedical content in non-English language. """ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION) BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION) BUILDER_CONFIGS = [ BigBioConfig( name="bioasq_2021_mesinesp_subtrack1_all_source", version=SOURCE_VERSION, description="bioasq_2021_mesinesp source schema subtrack1", schema="source", subset_id="bioasq_2021_mesinesp_subtrack1_all", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack1_only_articles_source", version=SOURCE_VERSION, description="bioasq_2021_mesinesp source schema subtrack1", schema="source", subset_id="bioasq_2021_mesinesp_subtrack1_only_articles", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack2_source", version=SOURCE_VERSION, description="bioasq_2021_mesinesp source schema subtrack2", schema="source", subset_id="bioasq_2021_mesinesp_subtrack2", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack3_source", version=SOURCE_VERSION, description="bioasq_2021_mesinesp source schema subtrack3", schema="source", subset_id="bioasq_2021_mesinesp_subtrack3", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack1_all_bigbio_text", version=BIGBIO_VERSION, description="bioasq_2021_mesinesp BigBio schema subtrack1", schema="bigbio_text", subset_id="bioasq_2021_mesinesp_subtrack1_all", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack1_only_articles_bigbio_text", version=BIGBIO_VERSION, description="bioasq_2021_mesinesp BigBio schema subtrack1", schema="bigbio_text", subset_id="bioasq_2021_mesinesp_subtrack1_only_articles", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack2_bigbio_text", version=BIGBIO_VERSION, description="bioasq_2021_mesinesp BigBio schema subtrack2", schema="bigbio_text", subset_id="bioasq_2021_mesinesp_subtrack2", ), BigBioConfig( name="bioasq_2021_mesinesp_subtrack3_bigbio_text", version=BIGBIO_VERSION, description="bioasq_2021_mesinesp BigBio schema subtrack3", schema="bigbio_text", subset_id="bioasq_2021_mesinesp_subtrack3", ), ] DEFAULT_CONFIG_NAME = "bioasq_2021_mesinesp_source" def _info(self) -> datasets.DatasetInfo: if self.config.schema == "source": features = datasets.Features( { "abstractText": datasets.Value("string"), "db": datasets.Value("string"), "decsCodes": datasets.Sequence(datasets.Value("string")), "id": datasets.Value("string"), "journal": datasets.Value("string"), "title": datasets.Value("string"), "year": datasets.Value("string"), } ) elif self.config.schema == "bigbio_text": features = text_features return datasets.DatasetInfo( description=_DESCRIPTION, features=features, homepage=_HOMEPAGE, license=str(_LICENSE), citation=_CITATION, ) def _split_generators(self, dl_manager) -> List[datasets.SplitGenerator]: """Returns SplitGenerators.""" if "subtrack1" in self.config.name: track = "1" elif "subtrack2" in self.config.name: track = "2" else: track = "3" urls = _URLS[_DATASETNAME][f"subtrack{track}"] if self.config.data_dir is None: try: data_dir = dl_manager.download_and_extract(urls) except ConnectionError: raise ConnectionError( "Could not download. Save locally and use `data_dir` kwarg" ) else: data_dir = self.config.data_dir if track == "1": top_folder = "Subtrack1-Scientific_Literature" elif track == "2": top_folder = "Subtrack2-Clinical_Trials" else: top_folder = "Subtrack3-Patents" if track == "1": if "all" in self.config.name: train_filepath = "training_set_subtrack1_all.json" else: train_filepath = "training_set_subtrack1_only_articles.json" else: train_filepath = f"training_set_subtrack{track}.json" dev_filepath = f"development_set_subtrack{track}.json" test_filepath = f"test_set_subtrack{track}.json" split_gens = [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "filepath": os.path.join( data_dir, top_folder, "Train", train_filepath ), }, ), datasets.SplitGenerator( name=datasets.Split.VALIDATION, gen_kwargs={ "filepath": os.path.join( data_dir, top_folder, "Development", dev_filepath ), }, ), datasets.SplitGenerator( name=datasets.Split.TEST, gen_kwargs={ "filepath": os.path.join( data_dir, top_folder, "Test", test_filepath ), }, ), ] # track 3 doesn't have Train data if track == "3": return split_gens[1:] return split_gens def _generate_examples(self, filepath) -> Tuple[int, Dict]: """Yields examples as (key, example) tuples.""" if self.config.schema == "source": with open(filepath) as fp: data = json.load(fp) for key, example in enumerate(data["articles"]): yield key, example elif self.config.schema == "bigbio_text": with open(filepath) as fp: data = json.load(fp) for key, example in enumerate(data["articles"]): yield key, { "id": example["id"], "document_id": "NULL", "text": example["abstractText"], "labels": example["decsCodes"], }