wmt21_metrics_task.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.
"""WMT21 Metrics Shared Task: Segment-Level Data"""

import datasets
import csv

_CITATION = """\
@inproceedings{freitag-etal-2021-results,
    title = {Results of the WMT21 Metrics Shared Task: Evaluating Metrics with Expert-based Human Evaluations on TED and News Domain},
    author = {Freitag, Markus and Rei, Ricardo  and Mathur, Nitika  and Lo, Chi-kiu  and Stewart, Craig  and Foster, George  and Lavie, Alon and Bojar, Ondrej}",
    booktitle = {Proceedings of the Sixth Conference on Machine Translation},
    month = {nov},
    year = {2021},
    address = {Online},
    publisher = {Association for Computational Linguistics},
    url = {https://aclanthology.org/2021.wmt-1.73},
    pages = {733--774}
}
"""

_DESCRIPTION = """\
This shared task will examine automatic evaluation metrics for machine translation. We will 
provide you with MT system outputs along with source text and the human reference translations. 
We are looking for automatic metric scores for translations at the system-level, and segment-level. 
We will calculate the system-level, and segment-level correlations of your scores with human judgements.

We invite submissions of reference-free metrics in addition to reference-based metrics.
"""

_HOMEPAGE = "https://www.statmt.org/wmt21/metrics-task.html"

_LICENSE = "Unknown"


_LANGUAGE_PAIRS = [('bn', 'hi'), ('hi', 'bn'), ('xh', 'zu'), ('zu', 'xh'), ('cs', 'en'), ('de', 'en'), ('de', 'fr'), ('en', 'cs'), ('en', 'de'), ('en', 'ha'), 
('en', 'is'), ('en', 'ja'), ('en', 'ru'), ('en', 'zh'), ('fr', 'de'), ('ha', 'en'), ('is', 'en'), ('ja', 'en'), ('ru', 'en'), ('zh', 'en')]

_URL_BASE = "https://huggingface.co/datasets/muibk/wmt21_metrics_task/resolve/main/"
_URLs = {f"{src_lg}-{trg_lg}": f"{_URL_BASE}{src_lg}-{trg_lg}/train.csv" for src_lg, trg_lg in _LANGUAGE_PAIRS}


class WmtMetricsTaskConfig(datasets.BuilderConfig):
    """BuilderConfig for WMT Metrics Shared Task."""
    def __init__(self, src_lg, tgt_lg, **kwargs):
        super(WmtMetricsTaskConfig, self).__init__(**kwargs)
        self.src_lg = src_lg
        self.tgt_lg = tgt_lg


class Wmt21MetricsTask(datasets.GeneratorBasedBuilder):
    """WMT Metrics Shared Task."""

    BUILDER_CONFIGS = [
        WmtMetricsTaskConfig(
            name=f"{src_lg}-{tgt_lg}",
            version=datasets.Version("1.1.0"),
            description=f"WMT 2021 Metrics Task: {src_lg} - {tgt_lg}",
            src_lg=src_lg,
            tgt_lg=tgt_lg,
        )
        for (src_lg, tgt_lg) in _LANGUAGE_PAIRS
    ]
    BUILDER_CONFIG_CLASS = WmtMetricsTaskConfig

    def _info(self):
        # define feature types
        features = datasets.Features(
            {
                #'source' : datasets.Value("string"),
                #'system_output':datasets.Value("string"),
                'translation': datasets.Translation(languages=(self.config.src_lg, self.config.tgt_lg)),
                'mt_system':datasets.Value("string"),
                'mqm':datasets.Value("float32"),
                'wmt-raw':datasets.Value("float32"),
                'wmt-z':datasets.Value("float32"),
                'pair':datasets.Value("string"),
                'dataset':datasets.Value("string"),
                'sent_id':datasets.Value("int32"),
                'doc_name':datasets.Value("string"),
                'doc_ref':datasets.Value("string"),
                'ref-A':datasets.Value("string"),
                'ref-B':datasets.Value("string"),
                'ref-C':datasets.Value("string"),
                'ref-D':datasets.Value("string")
            }
        )

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            supervised_keys=None,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        pair = f"{self.config.src_lg}-{self.config.tgt_lg}" # string identifier for language pair
        url = _URLs[pair] # url for download of pair-specific train.csv
        data_file = dl_manager.download_and_extract(url) # extract downloaded data and store path in data_file

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": data_file,
                    "source_lg": self.config.src_lg,
                    "target_lg": self.config.tgt_lg,
                }
            )
        ]
    
    def _generate_examples(self, filepath, source_lg, target_lg):
        with open(filepath, encoding="utf-8") as f:
            reader = csv.DictReader(f, delimiter=";") # read each line into dict
            for id_, row in enumerate(reader):
                row["translation"] = {source_lg : row["source"], target_lg: row["system_output"]} # create translation json
                for key in ["source", "system_output"]: # remove obsolete columns
                    row.pop(key)
                row = {k: None if not v else v for k, v in row.items()} # replace empty values
                yield id_, row

# to test the script, go to the root folder of the repo (wmt21_metrics_task) and run:
# datasets-cli test wmt21_metrics_task --save_infos --all_configs