mmteb-miracl.py

# coding=utf-8
# Copyright 2020 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.

# Lint as: python3

import json
import datasets
from collections import defaultdict
from dataclasses import dataclass
from typing import Dict

_CITATION = '''@article{10.1162/tacl_a_00595,
 author = {Zhang, Xinyu and Thakur, Nandan and Ogundepo, Odunayo and Kamalloo, Ehsan and Alfonso-Hermelo, David and Li, Xiaoguang and Liu, Qun and Rezagholizadeh, Mehdi and Lin, Jimmy},
 title = "{MIRACL: A Multilingual Retrieval Dataset Covering 18 Diverse Languages}",
 journal = {Transactions of the Association for Computational Linguistics},
 volume = {11},
 pages = {1114-1131},
 year = {2023},
 month = {09},
 abstract = "{MIRACL is a multilingual dataset for ad hoc retrieval across 18 languages that collectively encompass over three billion native speakers around the world. This resource is designed to support monolingual retrieval tasks, where the queries and the corpora are in the same language. In total, we have gathered over 726k high-quality relevance judgments for 78k queries over Wikipedia in these languages, where all annotations have been performed by native speakers hired by our team. MIRACL covers languages that are both typologically close as well as distant from 10 language families and 13 sub-families, associated with varying amounts of publicly available resources. Extensive automatic heuristic verification and manual assessments were performed during the annotation process to control data quality. In total, MIRACL represents an investment of around five person-years of human annotator effort. Our goal is to spur research on improving retrieval across a continuum of languages, thus enhancing information access capabilities for diverse populations around the world, particularly those that have traditionally been underserved. MIRACL is available at http://miracl.ai/.}",
 issn = {2307-387X},
 doi = {10.1162/tacl_a_00595},
 url = {https://doi.org/10.1162/tacl\_a\_00595},
 eprint = {https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl\_a\_00595/2157340/tacl\_a\_00595.pdf},
}'''

surprise_languages = ['de', 'yo']
new_languages = ['es', 'fa', 'fr', 'hi', 'zh'] + surprise_languages
languages = ['ar', 'bn', 'en', 'es', 'fa', 'fi', 'fr', 'hi', 'id', 'ja', 'ko', 'ru', 'sw', 'te', 'th', 'zh'] + surprise_languages

languages2filesize = { 
 'ar': 5,
 'bn': 1,
 'en': 66,
 'es': 21,
 'fa': 5,
 'fi': 4,
 'fr': 30,
 'hi': 2,
 'id': 3,
 'ja': 14,
 'ko': 3,
 'ru': 20,
 'sw': 1,
 'te': 2,
 'th': 2,
 'zh': 10,
 'de': 32,
 'yo': 1,
}

_DESCRIPTION = 'dataset load script for MIRACL'

_DATASET_URLS = {
 language: {
 'dev': [
 f'https://huggingface.co/datasets/miracl/miracl/resolve/main/miracl-v1.0-{language}/qrels/qrels.miracl-v1.0-{language}-dev.tsv',
 ],
 'corpus': [
 f'https://huggingface.co/datasets/miracl/miracl-corpus/resolve/main/miracl-corpus-v1.0-{language}/docs-{i}.jsonl.gz' for i in range(n)
 ],
 'queries': [
 f'https://huggingface.co/datasets/miracl/miracl/resolve/main/miracl-v1.0-{language}/topics/topics.miracl-v1.0-{language}-dev.tsv',
 ],
 } for language, n in languages2filesize.items()
}

def load_topic(fn: str) -> Dict[str, str]:
 """
 Load topics from a file.
 Args:
 fn: file path
 Returns:
 A dictionary from query id to query text.
 """
 qid2topic = {}
 with open(fn, encoding="utf-8") as f:
 for line in f:
 qid, topic = line.strip().split('\t')
 qid2topic[qid] = topic
 return qid2topic


def load_qrels(fn: str) -> Dict[str, Dict[str, int]]:
 """
 Load qrels from a file.
 Args:
 fn: file path
 Returns:
 A dictionary from query id to a dictionary from doc id to relevance score.
 """
 if fn is None:
 return None

 qrels = defaultdict(dict)
 with open(fn, encoding="utf-8") as f:
 for line in f:
 qid, _, docid, rel = line.strip().split('\t')
 qrels[qid][docid] = int(rel)
 return qrels


class MMTEBMIRACL(datasets.GeneratorBasedBuilder):
 BUILDER_CONFIGS = [datasets.BuilderConfig(
 version=datasets.Version('1.0.0'),
 name=lang, description=f'MIRACL qrels in language {lang}.'
 ) for lang in languages
 ] + [
 datasets.BuilderConfig(
 version=datasets.Version('1.0.0'),
 name=f'corpus-{lang}', description=f'corpus of MIRACL dataset in language {lang}.'
 ) for lang in languages
 ] + [ 
 datasets.BuilderConfig(
 version=datasets.Version('1.0.0'),
 name=f'queries-{lang}', description=f'queries of MIRACL dataset in language {lang}.'
 ) for lang in languages
 ]

 def _info(self):
 name = self.config.name
 
 if name.startswith('corpus-'):
 features = datasets.Features({
 'docid': datasets.Value('string'),
 'title': datasets.Value('string'),
 'text': datasets.Value('string'),
 })
 elif name.startswith("queries-"):
 features = datasets.Features({
 'query_id': datasets.Value('string'),
 'query': datasets.Value('string'),
 })
 else:
 features = datasets.Features({
 'query_id': datasets.Value('string'),
 'docid': datasets.Value('string'),
 'score': datasets.Value('int32'),
 })

 return datasets.DatasetInfo(
 # This is the description that will appear on the datasets page.
 description=_DESCRIPTION,
 # This defines the different columns of the dataset and their types
 features=features, # Here we define them above because they are different between the two configurations
 supervised_keys=None,
 # Homepage of the dataset for documentation
 homepage='https://project-miracl.github.io',
 # License for the dataset if available
 license=None,
 # Citation for the dataset
 citation=_CITATION,
 )

 def _split_generators(self, dl_manager):
 name = self.config.name

 if name.startswith('corpus-'):
 language = name.replace('corpus-', '')
 downloaded_files = dl_manager.download_and_extract(
 _DATASET_URLS[language]['corpus'])
 splits = [
 datasets.SplitGenerator(
 name='corpus',
 gen_kwargs={
 'filepaths': downloaded_files,
 },
 ),
 ]
 elif name.startswith('queries-'):
 language = name.replace('queries-', '')
 downloaded_files = dl_manager.download_and_extract(
 _DATASET_URLS[language]['queries'])
 splits = [
 datasets.SplitGenerator(
 name='queries',
 gen_kwargs={
 'filepaths': downloaded_files,
 },
 ),
 ]
 else:
 language = name
 downloaded_files = dl_manager.download_and_extract(
 _DATASET_URLS[language]['dev'])
 splits = [
 datasets.SplitGenerator(
 name='dev',
 gen_kwargs={
 'filepaths': downloaded_files,
 },
 ),
 ]
 
 return splits

 def _generate_examples(self, filepaths):
 name = self.config.name

 if name.startswith('corpus-'):
 for filepath in sorted(filepaths):
 with open(filepath, encoding="utf-8") as f:
 for line in f:
 data = json.loads(line)
 yield data['docid'], data
 
 elif name.startswith('queries-'):
 for filepath in filepaths:
 qid2topic = load_topic(filepath)
 for qid in qid2topic:
 data = {}
 data['query_id'] = qid
 data['query'] = qid2topic[qid]
 yield qid, data

 else:
 for filepath in filepaths:
 qrels = load_qrels(filepath)
 for qid in qrels:
 for docid in qrels[qid]:
 data = {}
 data['query_id'] = qid
 data['docid'] = docid
 data['score'] = qrels[qid][docid]
 yield f"{qid}.{docid}", data