README.md

---
license: cc-by-4.0
language:
- ru
- mdf
- myv
task_categories:
- translation
size_categories:
- 100K<n<1M
configs:
- config_name: default
  data_files:
  - split: train
    path: data/subset=*/split=train/*.parquet
  - split: validation
    path: data/subset=*/split=dev/*.parquet
  - split: test
    path: data/subset=*/split=test/*.parquet
- config_name: mdf_rus
  data_files:
  - split: train
    path: data/subset=mdf_rus/split=train/*.parquet
  - split: validation
    path: data/subset=mdf_rus/split=dev/*.parquet
  - split: test
    path: data/subset=mdf_rus/split=test/*.parquet
- config_name: myv_rus
  data_files:
  - split: train
    path: data/subset=myv_rus/split=train/*.parquet
  - split: validation
    path: data/subset=myv_rus/split=dev/*.parquet
  - split: test
    path: data/subset=myv_rus/split=test/*.parquet
- config_name: myv_mdf
  data_files:
  - split: train
    path: data/subset=myv_mdf/split=train/*.parquet
  - split: validation
    path: data/subset=myv_mdf/split=dev/*.parquet
  - split: test
    path: data/subset=myv_mdf/split=test/*.parquet
---


# "e-mordovia-articles-2024": a parallel news dataset for Russian, Erzya and Moksha

This is a semi-aligned dataset of Russian, Erzya and Moksha news articles, crawled from https://www.e-mordovia.ru.

## Dataset Description

### Dataset Summary

This is a dataset of news articles collected from https://www.e-mordovia.ru, the official portal of the state authorities of the Republic of Mordovia.

The articles have been paired by the following algorithm:
1. Calculate similarities between all articles of source and target languages
2. Filter pairs of articles with similarities bigger than 0.6
3. Filter pairs of articles where the publication dates differ by less than 30 days
4. Validate each pair of articles by native speakers.

Then articles have been split into sentences and automatically aligned on the sentence level.

The earliest publication date among the aligned articles is 2016-11-21, and the latest is 2024-10-02. 
The publication dates of the source language articles were used to select the dev and test splits, 
each approximately comprising 5% of the total articles. Articles published after 2024-02-01 form the 
validation split; those after 2024-05-25 form the test split.

The dataset contains unaligned sentences and sentences that were aligned poorly. 
To extract a relatively high-quality parallel part, it is recommended to filter the sentence pairs by `sim>=` ([TBA]). 

Different paired articles contain duplicate aligned sentences so for processing it may be necessary to drop duplicates based on the src_sent and tgt_sent columns.

### Supported Tasks and Leaderboards

The dataset is intended to be used as training data for machine translation models.

### Languages

The languages are Russian (`rus`), Erzya (`myv`) and Moksha ('mdf'). 
Erzya and Moksha are languages from the Mordvinic branch of the Uralic family. They are two of the three official languages in Mordovia, alongside with Russian.

All languages are written in the Cyrillic script.
 
## Dataset Structure

### Data Instances

A typical datapoint is a pair sentences (if they were aligned) or a single sentence in one of the languages (if the algorithm failed to align them).

Each sentence is associated with a document (a news article).

A typical aligned instance may look like
```Python
{'src_sent_id': 0.0,
 'src_sent': '16 апрельста Артём Здунов ётафтсь рабочай васедема "Сколково" Фондть Правлениянц Председателенц, Россиянь Федерациянь Советть видеса интеллектуальнай собственностень кизефкснень коряс Советть членонц Игорь Дроздовонь мархта.',
 'tgt_sent_id': 0.0,
 'tgt_sent': '16 апреля Артём Здунов провел рабочую встречу с Председателем Правления Фонда "Сколково", членом Совета по вопросам интеллектуальной собственности при Совете Федерации России Игорем Дроздовым.',
 'sim': 0.7039214883531842,
 'sim_pnlz': 0.34841506047894927,
 'src_doc_link': 'https://www.e-mordovia.ru/mkh/for-smi/all-news/skolkovo-fondt-pravleniyants-predsedatelets-igor-drozdov-respublikat-potentsialots-pyak-otsyu-/',
 'tgt_doc_link': 'https://www.e-mordovia.ru/glava-rm/novosti/predsedatel-pravleniya-fonda-skolkovo-igor-drozdov-potentsiafbddgn/',
 'src_doc_hash': '4ef6081d024d9604',
 'tgt_doc_hash': '91208a045de3de28',
 'docs_sim': 0.6881643989786337,
 'src_id': 2117}
```
An unaligned instance may look like
```Python
{'src_sent_id': 16.0,
 'src_sent': 'Активисттне макссесть лама кизефкс.',
 'tgt_sent_id': nan,
 'tgt_sent': None,
 'sim': nan,
 'sim_pnlz': nan,
 'src_doc_link': 'https://www.e-mordovia.ru/mkh/for-smi/all-news/1-iyuntsta-saranskyaysa-yetaftovs-od-lomanen-festival/',
 'tgt_doc_link': 'https://www.e-mordovia.ru/glava-rm/novosti/1-iyunya-v-saranske-proshel-molodezhnyy-festival-/',
 'src_doc_hash': '88deac0d4700895f',
 'tgt_doc_hash': '231c20a5f989549e',
 'docs_sim': 0.46148286872766453,
 'src_id': 1052}
```

### Data Fields

- `src_sent_id`: id of the sentence in the source language of the pair (or empty)
- `src_sent`: the source language sentence (or empty)
- `tgt_sent_id`: id of the sentence the target language article (or empty)
- `tgt_sent`: the target language sentence (or empty)
- `sim`: similarity of the sentences (or empty): a product of their LaBSE cosine similarities and their shortest-to-longest ratio of character lengths.
- `sim_pnlz`: a penalized similarity of the sentences (or empty); based on whether this number is positive, the decision was made to align the sentences or not.
- `src_doc_link`: link to the source language article
- `tgt_doc_link`: link to the target language article
- `src_doc_hash`: unique identifier of the source language article
- `tgt_doc_hash`: unique identifier of the target language article
- `docs_sim`: the similarity between the documents, computed as the aggregated similarity of the individual sentences in them
- `src_id`: numeric id of the document pair
- `src_lang`: ISO 639-3 code for the source language in the pair (e.g., "myv" for Erzya, "mdf" for Moksha)
- `tgt_lang`: ISO 639-3 code for the target language in the pair (e.g., "rus" for Russian, "mdf" for Moksha)

### Data Splits

The dataset is separated in three subsets:

1. myv-rus
2. mdf-rus
3. myv-mdf

All subsets are split into test, validation and train subsets. The publication dates of the source language articles were used to select 
the dev and test splits, each approximately comprising 5% of the total articles. Articles published after 2024-02-01 form the validation split;
those after 2024-05-25 form the test split.

## Dataset Creation

### Curation Rationale

The dataset has been curated in order to boost the quality of machine translation for Erzya or Moksha.

### Source Data

#### Initial Data Collection and Normalization

The data has been scraped from the https://www.e-mordovia.ru website. 

The Erzya and Russian articles were considered as a possible pair if their included the same image.
Then each candidate pair of articles was split into sentences with the [razdel](https://github.com/natasha/razdel) Python package 
and aligned using the [slone_nmt](https://github.com/slone-nlp/myv-nmt) package,
the [slone/LaBSE-en-ru-myv-v2](https://huggingface.co/slone/LaBSE-en-ru-myv-v2) and
and the `achapaev/LaBSE-en-ru-mdf-v1` [TBA] sentence encoders.

No other text preprocessing has been performed.

#### Who are the source language producers?

The language producers are the writers and the editors of the e-mordovia.ru portal, typically anonymous.

### Annotations

The dataset does not contain any additional annotations.

### Personal and Sensitive Information

The dataset may contain personal names of the people figurating in the news. 
However, as all these news are public anyway, we do not see any additional risks in using this data for training NLP models.

## Considerations for Using the Data

### Social Impact of Dataset

We hope that by enabling better machine translation from and to Erzya and Moksha, this dataset would contribute to improve the digital presence and overall prestige of the Erzya and Moksha languages.

### Discussion of Biases

As a portal of a public authority in Russian Federation, a country not very famous for its freedom of speech, 
*e-mordovia* may be presenting its own particular coverage and interpretation of some internal, federal and international events. 
Therefore, while we recommend using the collected dataset for teaching the Erzya and Moksha languages to models (and maybe even to humans), 
its usage as a source of non-linguistic knowledge should be approached with extreme caution.

### Other Known Limitations

The original articles were not originally intended to serve as parallel texts, and the writers who translated them between Erzya and Russian might have altered the content someitmes.
Our automatic metrics of sentence similarity were designed to filter out such cases, these metrics are error-prone themselves.
Therefore, whereas the aligned part of the dataset seems to be highly parallel, the translations should be treated as noisy rather than 100% accurate.
Filtering by the `sim` and `doc_sim` fields is recommended for obtaining a highly parallel subset.

## Additional Information

### Dataset Curators

The dataset was collected and aligned by Artem Chapaev using the data and algorithm from the [previous dataset](https://huggingface.co/datasets/slone/e-mordovia-articles-2023).

### Licensing Information

According to the note on the source website (https://www.e-mordovia.ru/podderzhka-portala/ob-ispolzovanii-informatsii-sayta/):

> All materials of the official website of the state authorities of the Republic of Mordovia can be reproduced in any media, on Internet servers or on any other media without any restrictions on the volume and timing of publication.
> This permission applies equally to newspapers, magazines, radio stations, TV channels, websites and Internet pages.
> The only condition for reprinting and retransmission is a link to the original source. No prior consent to reprint is required from the editorial board of the official website of the state authorities of the Republic of Mordovia.

Here is the original note in Russian:

> Все материалы официального сайта органов государственной власти Республики Мордовия могут быть воспроизведены в любых средствах массовой информации, на серверах сети Интернет или на любых иных носителях без каких-либо ограничений по объему и срокам публикации.
> Это разрешение в равной степени распространяется на газеты, журналы, радиостанции, телеканалы, сайты и страницы сети Интернет.
> Единственным условием перепечатки и ретрансляции является ссылка на первоисточник. Никакого предварительного согласия на перепечатку со стороны редакции официального сайта органов государственной власти Республики Мордовия не требуется.

In short, you can use and even redistribute the data, but when redistributing, you must give a reference to the original website, https://www.e-mordovia.ru.

Based on this note, we concluded that a CC-BY license would be appropriate for this dataset.

### Citation Information

[TBD]