Adding Turkic X-WMT evaluation set for machine translation (#3605)

* adding turkic xwmt evaluation dataset

* fixed isort issues

* addressing pr comments

* addressing pr comments

* Apply suggestions from code review

* Update README.md

Co-authored-by: Jamshidbek Mirzakhalov <mirzakhalov@Jamshidbeks-MacBook-Pro.local>
Co-authored-by: Jamshidbek Mirzakhalov <mirzakhalov@Jamshidbeks-MBP.lan>
Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com>

Commit from https://github.com/huggingface/datasets/commit/6c89c96617c98d6929f890e776b7da1166f8ba6c

README.md

@@ -0,0 +1,596 @@
+---
+annotations_creators:
+- crowdsourced
+language_creators:
+- found
+languages:
+  az-ba: 
+  - az
+  - ba
+  az-en: 
+  - az
+  - en
+  az-kaa: 
+  - az
+  - kaa
+  az-kk: 
+  - az
+  - kk
+  az-ky: 
+  - az
+  - ky
+  az-ru: 
+  - az
+  - ru
+  az-sah: 
+  - az
+  - sah
+  az-tr: 
+  - az
+  - tr
+  az-uz: 
+  - az
+  - uz
+  ba-az: 
+  - ba
+  - az
+  ba-en: 
+  - ba
+  - en
+  ba-kaa: 
+  - ba
+  - kaa
+  ba-kk: 
+  - ba
+  - kk
+  ba-ky: 
+  - ba
+  - ky
+  ba-ru: 
+  - ba
+  - ru
+  ba-sah: 
+  - ba
+  - sah
+  ba-tr: 
+  - ba
+  - tr
+  ba-uz: 
+  - ba
+  - uz
+  en-az: 
+  - en
+  - az
+  en-ba: 
+  - en
+  - ba
+  en-kaa: 
+  - en
+  - kaa
+  en-kk: 
+  - en
+  - kk
+  en-ky: 
+  - en
+  - ky
+  en-ru: 
+  - en
+  - ru
+  en-sah: 
+  - en
+  - sah
+  en-tr: 
+  - en
+  - tr
+  en-uz: 
+  - en
+  - uz
+  kaa-az: 
+  - kaa
+  - az
+  kaa-ba: 
+  - kaa
+  - ba
+  kaa-en: 
+  - kaa
+  - en
+  kaa-kk: 
+  - kaa
+  - kk
+  kaa-ky: 
+  - kaa
+  - ky
+  kaa-ru: 
+  - kaa
+  - ru
+  kaa-sah: 
+  - kaa
+  - sah
+  kaa-tr: 
+  - kaa
+  - tr
+  kaa-uz: 
+  - kaa
+  - uz
+  kk-az: 
+  - kk
+  - az
+  kk-ba: 
+  - kk
+  - ba
+  kk-en: 
+  - kk
+  - en
+  kk-kaa: 
+  - kk
+  - kaa
+  kk-ky: 
+  - kk
+  - ky
+  kk-ru: 
+  - kk
+  - ru
+  kk-sah: 
+  - kk
+  - sah
+  kk-tr: 
+  - kk
+  - tr
+  kk-uz: 
+  - kk
+  - uz
+  ky-az: 
+  - ky
+  - az
+  ky-ba: 
+  - ky
+  - ba
+  ky-en: 
+  - ky
+  - en
+  ky-kaa: 
+  - ky
+  - kaa
+  ky-kk: 
+  - ky
+  - kk
+  ky-ru: 
+  - ky
+  - ru
+  ky-sah: 
+  - ky
+  - sah
+  ky-tr: 
+  - ky
+  - tr
+  ky-uz: 
+  - ky
+  - uz
+  ru-az: 
+  - ru
+  - az
+  ru-ba: 
+  - ru
+  - ba
+  ru-en: 
+  - ru
+  - en
+  ru-kaa: 
+  - ru
+  - kaa
+  ru-kk: 
+  - ru
+  - kk
+  ru-ky: 
+  - ru
+  - ky
+  ru-sah: 
+  - ru
+  - sah
+  ru-tr: 
+  - ru
+  - tr
+  ru-uz: 
+  - ru
+  - uz
+  sah-az: 
+  - sah
+  - az
+  sah-ba: 
+  - sah
+  - ba
+  sah-en: 
+  - sah
+  - en
+  sah-kaa: 
+  - sah
+  - kaa
+  sah-kk: 
+  - sah
+  - kk
+  sah-ky: 
+  - sah
+  - ky
+  sah-ru: 
+  - sah
+  - ru
+  sah-tr: 
+  - sah
+  - tr
+  sah-uz: 
+  - sah
+  - uz
+  tr-az: 
+  - tr
+  - az
+  tr-ba: 
+  - tr
+  - ba
+  tr-en: 
+  - tr
+  - en
+  tr-kaa: 
+  - tr
+  - kaa
+  tr-kk: 
+  - tr
+  - kk
+  tr-ky: 
+  - tr
+  - ky
+  tr-ru: 
+  - tr
+  - ru
+  tr-sah: 
+  - tr
+  - sah
+  tr-uz: 
+  - tr
+  - uz
+  uz-az: 
+  - uz
+  - az
+  uz-ba: 
+  - uz
+  - ba
+  uz-en: 
+  - uz
+  - en
+  uz-kaa: 
+  - uz
+  - kaa
+  uz-kk: 
+  - uz
+  - kk
+  uz-ky: 
+  - uz
+  - ky
+  uz-ru: 
+  - uz
+  - ru
+  uz-sah: 
+  - uz
+  - sah
+  uz-tr: 
+  - uz
+  - tr
+licenses:
+- mit
+multilinguality:
+- translation
+pretty_name: turkic_xwmt
+size_categories:
+- n<1K
+task_categories:
+- conditional-text-generation
+task_ids:
+- machine-translation
+source_datasets:
+- extended|WMT 2020 News Translation Task
+---
+
+# Dataset Card for turkic_xwmt
+
+## Table of Contents
+- [Table of Contents](#table-of-contents)
+- [Dataset Description](#dataset-description)
+  - [Dataset Summary](#dataset-summary)
+  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
+  - [Languages](#languages)
+- [Dataset Structure](#dataset-structure)
+  - [Data Instances](#data-instances)
+  - [Data Fields](#data-fields)
+  - [Data Splits](#data-splits)
+- [Dataset Creation](#dataset-creation)
+  - [Curation Rationale](#curation-rationale)
+  - [Source Data](#source-data)
+  - [Annotations](#annotations)
+  - [Personal and Sensitive Information](#personal-and-sensitive-information)
+- [Considerations for Using the Data](#considerations-for-using-the-data)
+  - [Social Impact of Dataset](#social-impact-of-dataset)
+  - [Discussion of Biases](#discussion-of-biases)
+  - [Other Known Limitations](#other-known-limitations)
+- [Additional Information](#additional-information)
+  - [Dataset Curators](#dataset-curators)
+  - [Licensing Information](#licensing-information)
+  - [Citation Information](#citation-information)
+  - [Contributions](#contributions)
+
+## Dataset Description
+
+- **Homepage:** 
+- **Repository:**[Github](https://github.com/turkic-interlingua/til-mt/tree/master/xwmt)
+- **Paper:** [https://arxiv.org/abs/2109.04593](https://arxiv.org/abs/2109.04593)
+- **Leaderboard:** [More Information Needed]
+- **Point of Contact:** [turkicinterlingua@gmail.com](mailto:turkicinterlingua@gmail.com)
+
+### Dataset Summary
+
+To establish a comprehensive and challenging evaluation benchmark for Machine Translation in Turkic languages, we translate a test set originally introduced in WMT 2020 News Translation Task for English-Russian. The original dataset is profesionally translated and consists of sentences from news articles that are both English and Russian-centric. We adopt this evaluation set (X-WMT) and begin efforts to translate it into several Turkic languages. The current version of X-WMT includes covers 8 Turkic languages and 88 language directions with a minimum of 300 sentences per language direction.
+
+### Supported Tasks and Leaderboards
+
+[More Information Needed]
+
+### Languages
+
+Currently covered languages are (besides English and Russian):
+- Azerbaijani (az)
+- Bashkir (ba)
+- Karakalpak (kaa)
+- Kazakh (kk)
+- Kirghiz (ky)
+- Turkish (tr)
+- Sakha (sah)
+- Uzbek (uz)
+
+## Dataset Structure
+
+### Data Instances
+
+A random example from the Russian-Uzbek set:
+```
+{"translation": {'ru': 'Моника Мутсвангва , министр информации Зимбабве , утверждает , что полиция вмешалась в отъезд Магомбейи из соображений безопасности и вследствие состояния его здоровья .', 'uz': 'Zimbabvening Axborot vaziri , Monika Mutsvanva Magombeyining xavfsizligi va sog'ligi tufayli bo'lgan jo'nab ketishinida politsiya aralashuvini ushlab turadi .'}}
+```
+### Data Fields
+
+Each example has one field "translation" that contains two subfields: one per language, e.g. for the Russian-Uzbek set:
+- **translation**: a dictionary with two subfields:
+  - **ru**: the russian text
+  - **uz**: the uzbek text
+### Data Splits
+
+<details>
+  <summary>Click here to show the number of examples per configuration:</summary>
+|         |   test |
+|:--------|-------:|
+| az-ba   |    600 |
+| az-en   |    600 |
+| az-kaa  |    300 |
+| az-kk   |    500 |
+| az-ky   |    500 |
+| az-ru   |    600 |
+| az-sah  |    300 |
+| az-tr   |    500 |
+| az-uz   |    600 |
+| ba-az   |    600 |
+| ba-en   |   1000 |
+| ba-kaa  |    300 |
+| ba-kk   |    700 |
+| ba-ky   |    500 |
+| ba-ru   |   1000 |
+| ba-sah  |    300 |
+| ba-tr   |    700 |
+| ba-uz   |    900 |
+| en-az   |    600 |
+| en-ba   |   1000 |
+| en-kaa  |    300 |
+| en-kk   |    700 |
+| en-ky   |    500 |
+| en-ru   |   1000 |
+| en-sah  |    300 |
+| en-tr   |    700 |
+| en-uz   |    900 |
+| kaa-az  |    300 |
+| kaa-ba  |    300 |
+| kaa-en  |    300 |
+| kaa-kk  |    300 |
+| kaa-ky  |    300 |
+| kaa-ru  |    300 |
+| kaa-sah |    300 |
+| kaa-tr  |    300 |
+| kaa-uz  |    300 |
+| kk-az   |    500 |
+| kk-ba   |    700 |
+| kk-en   |    700 |
+| kk-kaa  |    300 |
+| kk-ky   |    500 |
+| kk-ru   |    700 |
+| kk-sah  |    300 |
+| kk-tr   |    500 |
+| kk-uz   |    700 |
+| ky-az   |    500 |
+| ky-ba   |    500 |
+| ky-en   |    500 |
+| ky-kaa  |    300 |
+| ky-kk   |    500 |
+| ky-ru   |    500 |
+| ky-sah  |    300 |
+| ky-tr   |    400 |
+| ky-uz   |    500 |
+| ru-az   |    600 |
+| ru-ba   |   1000 |
+| ru-en   |   1000 |
+| ru-kaa  |    300 |
+| ru-kk   |    700 |
+| ru-ky   |    500 |
+| ru-sah  |    300 |
+| ru-tr   |    700 |
+| ru-uz   |    900 |
+| sah-az  |    300 |
+| sah-ba  |    300 |
+| sah-en  |    300 |
+| sah-kaa |    300 |
+| sah-kk  |    300 |
+| sah-ky  |    300 |
+| sah-ru  |    300 |
+| sah-tr  |    300 |
+| sah-uz  |    300 |
+| tr-az   |    500 |
+| tr-ba   |    700 |
+| tr-en   |    700 |
+| tr-kaa  |    300 |
+| tr-kk   |    500 |
+| tr-ky   |    400 |
+| tr-ru   |    700 |
+| tr-sah  |    300 |
+| tr-uz   |    600 |
+| uz-az   |    600 |
+| uz-ba   |    900 |
+| uz-en   |    900 |
+| uz-kaa  |    300 |
+| uz-kk   |    700 |
+| uz-ky   |    500 |
+| uz-ru   |    900 |
+| uz-sah  |    300 |
+| uz-tr   |    600 |
+</details>
+
+## Dataset Creation
+
+### Curation Rationale
+
+[More Information Needed]
+
+### Source Data
+
+#### Initial Data Collection and Normalization
+
+[More Information Needed]
+
+#### Who are the source language producers?
+
+[More Information Needed]
+
+### Annotations
+
+#### Annotation process
+
+[More Information Needed]
+
+#### Who are the annotators?
+
+**Translators, annotators and dataset contributors** (in alphabetical order)
+
+Abilxayr Zholdybai  
+Aigiz Kunafin  
+Akylbek Khamitov  
+Alperen Cantez  
+Aydos Muxammadiyarov  
+Doniyorbek Rafikjonov  
+Erkinbek Vokhabov  
+Ipek Baris  
+Iskander Shakirov  
+Madina Zokirjonova  
+Mohiyaxon Uzoqova  
+Mukhammadbektosh Khaydarov  
+Nurlan Maharramli  
+Petr Popov  
+Rasul Karimov  
+Sariya Kagarmanova  
+Ziyodabonu Qobiljon qizi 
+
+### Personal and Sensitive Information
+
+[More Information Needed]
+
+## Considerations for Using the Data
+
+### Social Impact of Dataset
+
+[More Information Needed]
+
+### Discussion of Biases
+
+[More Information Needed]
+
+### Other Known Limitations
+
+[More Information Needed]
+
+## Additional Information
+
+### Dataset Curators
+
+[More Information Needed]
+
+### Licensing Information
+
+[MIT License](https://github.com/turkic-interlingua/til-mt/blob/master/xwmt/LICENSE)
+
+### Citation Information
+
+```
+@inproceedings{mirzakhalov2021large,
+  title={A Large-Scale Study of Machine Translation in Turkic Languages},
+  author={Mirzakhalov, Jamshidbek and Babu, Anoop and Ataman, Duygu and Kariev, Sherzod and Tyers, Francis and Abduraufov, Otabek and Hajili, Mammad and Ivanova, Sardana and Khaytbaev, Abror and Laverghetta Jr, Antonio and others},
+  booktitle={Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
+  pages={5876--5890},
+  year={2021}
+}
+```
+
+### Contributions
+
+This project was carried out with the help and contributions from dozens of individuals and organizations. We acknowledge and greatly appreciate each and every one of them:
+
+**Authors on the publications** (in alphabetical order)
+
+Abror Khaytbaev  
+Ahsan Wahab  
+Aigiz Kunafin  
+Anoop Babu  
+Antonio Laverghetta Jr.  
+Behzodbek Moydinboyev  
+Dr. Duygu Ataman  
+Esra Onal  
+Dr. Francis Tyers  
+Jamshidbek Mirzakhalov  
+Dr. John Licato  
+Dr. Julia Kreutzer  
+Mammad Hajili  
+Mokhiyakhon Uzokova  
+Dr. Orhan Firat  
+Otabek Abduraufov  
+Sardana Ivanova  
+Shaxnoza Pulatova  
+Sherzod Kariev  
+Dr. Sriram Chellappan  
+
+**Translators, annotators and dataset contributors** (in alphabetical order)
+
+Abilxayr Zholdybai  
+Aigiz Kunafin  
+Akylbek Khamitov  
+Alperen Cantez  
+Aydos Muxammadiyarov  
+Doniyorbek Rafikjonov  
+Erkinbek Vokhabov  
+Ipek Baris  
+Iskander Shakirov  
+Madina Zokirjonova  
+Mohiyaxon Uzoqova  
+Mukhammadbektosh Khaydarov  
+Nurlan Maharramli  
+Petr Popov  
+Rasul Karimov  
+Sariya Kagarmanova  
+Ziyodabonu Qobiljon qizi  
+
+**Industry supporters**
+
+[Google Cloud](https://cloud.google.com/solutions/education)  
+[Khan Academy Oʻzbek](https://uz.khanacademy.org/)  
+[The Foundation for the Preservation and Development of the Bashkir Language](https://bsfond.ru/)  
+
+Thanks to [@mirzakhalov](https://github.com/mirzakhalov) for adding this dataset.

dummy/az-ba/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ee388fc7e097821af7507d978f3b7de3b02ff9e8c65001ec85307bf84ac1c82
+size 150606

turkic_xwmt.py

@@ -0,0 +1,134 @@
+# coding=utf-8
+# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
+#
+# 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
+"""A Large-Scale Study of Machine Translation in Turkic Languages."""
+import os
+
+import datasets
+
+
+_LANGUAGES = ["az", "ba", "en", "kaa", "kk", "ky", "ru", "sah", "tr", "uz"]
+
+_DESCRIPTION = """\
+A Large-Scale Study of Machine Translation in Turkic Languages
+"""
+
+_CITATION = """\
+@inproceedings{mirzakhalov2021large,
+  title={A Large-Scale Study of Machine Translation in Turkic Languages},
+  author={Mirzakhalov, Jamshidbek and Babu, Anoop and Ataman, Duygu and Kariev, Sherzod and Tyers, Francis and Abduraufov, Otabek and Hajili, Mammad and Ivanova, Sardana and Khaytbaev, Abror and Laverghetta Jr, Antonio and others},
+  booktitle={Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
+  pages={5876--5890},
+  year={2021}
+}
+"""
+
+_DATA_URL = (
+    "https://github.com/turkic-interlingua/til-mt/blob/6ac179350448895a63cc06fcfd1135882c8cc49b/xwmt/test.zip?raw=true"
+)
+
+
+class XWMTConfig(datasets.BuilderConfig):
+    """BuilderConfig for XWMT."""
+
+    def __init__(self, language_pair=(None, None), **kwargs):
+        """BuilderConfig for XWMT.
+
+        Args:
+            for the `datasets.features.text.TextEncoder` used for the features feature.
+          language_pair: pair of languages that will be used for translation. Should
+            contain 2-letter coded strings.
+          **kwargs: keyword arguments forwarded to super.
+        """
+        name = "%s-%s" % (language_pair[0], language_pair[1])
+
+        description = ("Translation dataset from %s to %s") % (language_pair[0], language_pair[1])
+        super(XWMTConfig, self).__init__(
+            name=name,
+            description=description,
+            version=datasets.Version("1.1.0", ""),
+            **kwargs,
+        )
+
+        # Validate language pair.
+        source, target = language_pair
+
+        assert source in _LANGUAGES, ("Config language pair must be one of the supported languages, got: %s", source)
+        assert target in _LANGUAGES, ("Config language pair must be one of the supported languages, got: %s", source)
+
+        self.language_pair = language_pair
+
+
+class TurkicXWMT(datasets.GeneratorBasedBuilder):
+    """XWMT machine translation dataset."""
+
+    BUILDER_CONFIGS = [
+        XWMTConfig(
+            language_pair=(lang1, lang2),
+        )
+        for lang1 in _LANGUAGES
+        for lang2 in _LANGUAGES
+        if lang1 != lang2
+    ]
+
+    def _info(self):
+        source, target = self.config.language_pair
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features(
+                {"translation": datasets.features.Translation(languages=self.config.language_pair)}
+            ),
+            supervised_keys=(source, target),
+            homepage="https://github.com/turkicinterlingua/til-mt",
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        path = dl_manager.download_and_extract(_DATA_URL)
+
+        source, target = self.config.language_pair
+        source_path = os.path.join(path, "test", f"{source}-{target}", f"{source}-{target}.{source}.txt")
+        target_path = os.path.join(path, "test", f"{source}-{target}", f"{source}-{target}.{target}.txt")
+
+        files = {}
+        files["test"] = {
+            "source_file": source_path,
+            "target_file": target_path,
+        }
+
+        return [
+            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs=files["test"]),
+        ]
+
+    def _generate_examples(self, source_file, target_file):
+        """This function returns the examples in the raw (text) form."""
+        source_sentences, target_sentences = None, None
+        source_sentences = open(source_file, encoding="utf-8").read().strip().split("\n")
+        target_sentences = open(target_file, encoding="utf-8").read().strip().split("\n")
+
+        assert len(target_sentences) == len(source_sentences), "Sizes do not match: %d vs %d for %s vs %s." % (
+            len(source_sentences),
+            len(target_sentences),
+            source_file,
+            target_file,
+        )
+
+        source, target = self.config.language_pair
+        for idx, (l1, l2) in enumerate(zip(source_sentences, target_sentences)):
+            result = {"translation": {source: l1, target: l2}}
+            # Make sure that both translations are non-empty.
+            if all(result.values()):
+                yield idx, result