|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
"""MultiSim is a growing collection of Text Simplfication datasets in multiple languages. Each dataset is a set of complex and simple sentence pairs.""" |
|
|
|
import pandas as pd |
|
import os |
|
from collections import defaultdict |
|
|
|
import datasets |
|
|
|
_CITATION = """\ |
|
@inproceedings{ryan-etal-2023-revisiting, |
|
title = "Revisiting non-{E}nglish Text Simplification: A Unified Multilingual Benchmark", |
|
author = "Ryan, Michael and |
|
Naous, Tarek and |
|
Xu, Wei", |
|
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", |
|
month = jul, |
|
year = "2023", |
|
address = "Toronto, Canada", |
|
publisher = "Association for Computational Linguistics", |
|
url = "https://aclanthology.org/2023.acl-long.269", |
|
pages = "4898--4927", |
|
abstract = "Recent advancements in high-quality, large-scale English resources have pushed the frontier of English Automatic Text Simplification (ATS) research. However, less work has been done on multilingual text simplification due to the lack of a diverse evaluation benchmark that covers complex-simple sentence pairs in many languages. This paper introduces the MultiSim benchmark, a collection of 27 resources in 12 distinct languages containing over 1.7 million complex-simple sentence pairs. This benchmark will encourage research in developing more effective multilingual text simplification models and evaluation metrics. Our experiments using MultiSim with pre-trained multilingual language models reveal exciting performance improvements from multilingual training in non-English settings. We observe strong performance from Russian in zero-shot cross-lingual transfer to low-resource languages. We further show that few-shot prompting with BLOOM-176b achieves comparable quality to reference simplifications outperforming fine-tuned models in most languages. We validate these findings through human evaluation.", |
|
} |
|
""" |
|
|
|
|
|
|
|
_DESCRIPTION = """\ |
|
MultiSim is a growing collection of Text Simplfication datasets in multiple languages. Each dataset is a set of complex and simple sentence pairs. |
|
""" |
|
|
|
|
|
_HOMEPAGE = "https://github.com/XenonMolecule/MultiSim" |
|
|
|
|
|
_LICENSE = """MIT License |
|
|
|
Copyright (c) 2023 Michael Ryan |
|
|
|
Permission is hereby granted, free of charge, to any person obtaining a copy |
|
of this software and associated documentation files (the "Software"), to deal |
|
in the Software without restriction, including without limitation the rights |
|
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell |
|
copies of the Software, and to permit persons to whom the Software is |
|
furnished to do so, subject to the following conditions: |
|
|
|
The above copyright notice and this permission notice shall be included in all |
|
copies or substantial portions of the Software. |
|
|
|
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
|
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
|
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
|
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
|
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
|
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE |
|
SOFTWARE.""" |
|
|
|
_SUBCORPORA = { |
|
"NewselaEN": { |
|
"path": "./data/English/Newsela EN", |
|
"language": "en" |
|
}, |
|
"WikiAutoEN": { |
|
"path": "./data/English/WikiAuto", |
|
"language": "en" |
|
}, |
|
"ASSET": { |
|
"path": "./data/English/ASSET", |
|
"language": "en" |
|
}, |
|
"Simplext": { |
|
"path": "./data/Spanish/Simplext", |
|
"language": "es" |
|
}, |
|
"NewselaES": { |
|
"path": "./data/Spanish/Newsela ES", |
|
"language": "es" |
|
}, |
|
"Terence": { |
|
"path" : "./data/Italian/Terence", |
|
"language": "it" |
|
}, |
|
"Teacher": { |
|
"path": "./data/Italian/Teacher", |
|
"language": "it" |
|
}, |
|
"SimpitikiWiki": { |
|
"path": "./data/Italian/Simpitiki Italian Wikipedia", |
|
"language": "it" |
|
}, |
|
"AdminIt": { |
|
"path": "./data/Italian/AdminIT", |
|
"language": "it" |
|
}, |
|
"PaCCSS-IT": { |
|
"path": "./data/Italian/PaCCSS-IT Corpus", |
|
"language": "it" |
|
}, |
|
"CLEAR" : { |
|
"path" : "./data/French/CLEAR Corpus", |
|
"language": "fr" |
|
}, |
|
"WikiLargeFR": { |
|
"path" : "./data/French/WikiLargeFR Corpus", |
|
"language": "fr" |
|
}, |
|
"EasyJapanese": { |
|
"path": "./data/Japanese/Easy Japanese Corpus", |
|
"language": "ja" |
|
}, |
|
"EasyJapaneseExtended": { |
|
"path": "./data/Japanese/Easy Japanese Extended", |
|
"language": "ja" |
|
}, |
|
"PorSimples" : { |
|
"path": "./data/Brazilian Portuguese/PorSimples", |
|
"language": "pt-br" |
|
}, |
|
"TextComplexityDE" : { |
|
"path": "./data/German/TextComplexityDE Parallel Corpus", |
|
"language": "de" |
|
}, |
|
"GEOLinoTest" : { |
|
"path" : "./data/German/GEOLino Corpus", |
|
"language": "de" |
|
}, |
|
"GermanNews" : { |
|
"path" : "./data/German/German News", |
|
"language": "de" |
|
}, |
|
"CBST": { |
|
"path" : "./data/Basque/CBST", |
|
"language": "eu" |
|
}, |
|
"DSim": { |
|
"path": "./data/Danish/DSim Corpus", |
|
"language": "da" |
|
}, |
|
"SimplifyUR": { |
|
"path": "./data/Urdu/SimplifyUR", |
|
"language": "ur" |
|
}, |
|
"RuWikiLarge": { |
|
"path" : "./data/Russian/RuWikiLarge", |
|
"language": "ru" |
|
}, |
|
"RSSE" : { |
|
"path": "./data/Russian/RSSE Corpus", |
|
"language": "ru" |
|
}, |
|
"RuAdaptLit" : { |
|
"path": "./data/Russian/RuAdapt Literature", |
|
"language": "ru" |
|
}, |
|
"RuAdaptFairytales" : { |
|
"path": "./data/Russian/RuAdapt Fairytales", |
|
"language": "ru" |
|
}, |
|
"RuAdaptEncy" : { |
|
"path" : "./data/Russian/RuAdapt Ency", |
|
"language": "ru" |
|
}, |
|
"TSSlovene" : { |
|
"path" : "./data/Slovene/Text Simplification Slovene", |
|
"language": "sl" |
|
} |
|
} |
|
|
|
_LANGUAGES = { |
|
"English":'en', |
|
"Spanish":'es', |
|
"Italian":'it', |
|
"French" : 'fr', |
|
"Japanese": 'ja', |
|
"Brazilian Portuguese": 'pt-br', |
|
"German": 'de', |
|
"Basque": 'eu', |
|
"Danish": 'da', |
|
"Urdu": 'ur', |
|
"Russian": 'ru', |
|
"Slovene": 'sl' |
|
} |
|
|
|
|
|
class MultilingualSimplification(datasets.GeneratorBasedBuilder): |
|
"""MultiSim is a growing collection of Text Simplfication datasets in multiple languages. Each dataset is a set of complex and simple sentence pairs.""" |
|
|
|
VERSION = datasets.Version("1.0.0") |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
BUILDER_CONFIGS = [ |
|
|
|
datasets.BuilderConfig(name="WikiAutoEN", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="ASSET", version=VERSION, description="TODO: Descriptions"), |
|
|
|
|
|
datasets.BuilderConfig(name="Terence", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="Teacher", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="SimpitikiWiki", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="AdminIt", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="PaCCSS-IT", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="CLEAR", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="WikiLargeFR", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="EasyJapanese", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="EasyJapaneseExtended", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="PorSimples", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="TextComplexityDE", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="GEOLinoTest", version=VERSION, description="TODO: Descriptions"), |
|
|
|
|
|
|
|
|
|
datasets.BuilderConfig(name="RuWikiLarge", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="RSSE", version=VERSION, description="TODO: Descriptions"), |
|
|
|
datasets.BuilderConfig(name="RuAdaptFairytales", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="RuAdaptEncy", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="TSSlovene", version=VERSION, description="TODO: Descriptions"), |
|
|
|
datasets.BuilderConfig(name="English", version=VERSION, description="TODO: Descriptions"), |
|
|
|
datasets.BuilderConfig(name="Italian", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="French", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="Japanese", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="Brazilian Portuguese", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="German", version=VERSION, description="TODO: Descriptions"), |
|
|
|
|
|
|
|
datasets.BuilderConfig(name="Russian", version=VERSION, description="TODO: Descriptions"), |
|
datasets.BuilderConfig(name="Slovene", version=VERSION, description="TODO: Descriptions"), |
|
|
|
datasets.BuilderConfig(name="all", version=VERSION, description="TODO: Descriptions"), |
|
] |
|
|
|
DEFAULT_CONFIG_NAME = "all" |
|
|
|
def _info(self): |
|
|
|
features = datasets.Features( |
|
{ |
|
"original": datasets.Value("string"), |
|
"simple": datasets.Sequence(feature={"simplifications" : datasets.Value("string")}) |
|
} |
|
) |
|
return datasets.DatasetInfo( |
|
|
|
description=_DESCRIPTION, |
|
|
|
features=features, |
|
|
|
|
|
|
|
|
|
homepage=_HOMEPAGE, |
|
|
|
license=_LICENSE, |
|
|
|
citation=_CITATION, |
|
) |
|
|
|
def _split_generators(self, dl_manager): |
|
|
|
|
|
|
|
|
|
|
|
|
|
filepaths = [] |
|
if (self.config.name == 'all'): |
|
for subcorpus in _SUBCORPORA: |
|
filepaths.append(_SUBCORPORA[subcorpus]['path']) |
|
elif (self.config.name in _LANGUAGES): |
|
lang_code = _LANGUAGES[self.config.name] |
|
for subcorpus in _SUBCORPORA: |
|
if _SUBCORPORA[subcorpus]['language'] == lang_code: |
|
filepaths.append(_SUBCORPORA[subcorpus]['path']) |
|
elif (self.config.name in _SUBCORPORA): |
|
filepaths = [_SUBCORPORA[self.config.name]['path']] |
|
else: |
|
print("Invalid configuration name: " + self.config.name + ". Try 'all', 'English', 'ASSET', etc.") |
|
return [ |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
|
|
gen_kwargs={ |
|
"filepaths": filepaths, |
|
"split": "train", |
|
}, |
|
), |
|
datasets.SplitGenerator( |
|
name=datasets.Split.VALIDATION, |
|
|
|
gen_kwargs={ |
|
"filepaths": filepaths, |
|
"split": "val", |
|
}, |
|
), |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TEST, |
|
|
|
gen_kwargs={ |
|
"filepaths": filepaths, |
|
"split": "test" |
|
}, |
|
), |
|
] |
|
|
|
|
|
def _generate_examples(self, filepaths, split): |
|
|
|
|
|
df = pd.DataFrame() |
|
|
|
if (len(filepaths) > 1): |
|
for filepath in filepaths: |
|
if os.path.exists(filepath + "_" + split + ".csv"): |
|
df = pd.concat([df, pd.read_csv(filepath + "_" + split + ".csv")]) |
|
|
|
|
|
df = df.sample(frac=1, random_state=3600).reset_index(drop=True) |
|
else: |
|
if os.path.exists(filepaths[0] + "_" + split + ".csv"): |
|
df = pd.read_csv(filepaths[0] + "_" + split + ".csv") |
|
|
|
if len(df) > 0: |
|
for key, row in df.iterrows(): |
|
|
|
original = row["original"] |
|
simple = [] |
|
for label,content in row.items(): |
|
if label != "original" and type(content) != float: |
|
simple.append({"simplifications": content}) |
|
yield key, { |
|
"original": original, |
|
"simple": simple |
|
} |