README.md

---
pretty_name: Korpus Malti
configs:
- config_name: shuffled
  data_files:
  - split: train
    path: data/shuffled/train*.jsonl
  - split: validation
    path: data/shuffled/validation*.jsonl
  - split: test
    path: data/shuffled/test*.jsonl
  features:
  - name: text
    dtype: string
  default: true
- config_name: belles_lettres
  data_files: data/belles_lettres/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: author
    dtype: string
  - name: title
    dtype: string
  - name: publisher
    dtype: string
  - name: published
    dtype: string
  - name: copyright
    dtype: string
  - name: translator
    dtype: string
  - name: date
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: blogs
  data_files: data/blogs/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: title
    dtype: string
  - name: url
    dtype: string
  - name: source
    dtype: string
  - name: date
    dtype: string
- config_name: comics
  data_files: data/comics/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: title
    dtype: string
  - name: date
    dtype: string
  - name: url
    dtype: string
- config_name: court
  data_files: data/court/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
  - name: note
    dtype: string
- config_name: eu_docs
  data_files: data/eu_docs/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: publisher
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
  - name: note
    dtype: string
- config_name: gov_docs
  data_files: data/gov_docs/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: publisher
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
  - name: note
    dtype: string
- config_name: government_gazzette
  data_files: data/government_gazzette/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
- config_name: law_eu
  data_files: data/law_eu/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
  - name: notes
    dtype: string
- config_name: law_mt
  data_files: data/law_mt/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
- config_name: legal
  data_files: data/legal/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: source
    dtype: string
- config_name: nonfiction
  data_files: data/nonfiction/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: title
    dtype: string
  - name: publisher
    dtype: string
  - name: editor
    dtype: string
  - name: bookref
    dtype: string
  - name: date
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: parliament
  data_files: data/parliament/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: date
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: filename
    dtype: string
  - name: url
    dtype: string
  - name: note
    dtype: string
- config_name: press_eu
  data_files: data/press_eu/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: title
    dtype: string
  - name: date
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: press_mt
  data_files: data/press_mt/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: title
    dtype: string
  - name: subtitle
    dtype: string
  - name: date
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: speeches
  data_files: data/speeches/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: theses
  data_files: data/theses/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: title
    dtype: string
  - name: date
    dtype: string
  - name: source
    dtype: string
- config_name: umlib_oar
  data_files: data/umlib_oar/*.jsonl
  features:
  - name: text
    list: string
  - name: categories
    list: string
  - name: title
    dtype: string
  - name: publishers
    list: string
  - name: filenames
    list: string
  - name: num_files
    dtype: int64
  - name: date
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: web_general
  data_files: data/web_general/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: subcategory
    dtype: string
  - name: year
    dtype: string
  - name: source
    dtype: string
  - name: url
    dtype: string
- config_name: wiki
  data_files: data/wiki/*.jsonl
  features:
  - name: text
    list: string
  - name: category
    dtype: string
  - name: title
    dtype: string
  - name: url
    dtype: string
  - name: id
    dtype: string
language:
- mt
multilinguality:
- monolingual
size_categories:
- 10M<n<100M
annotations_creators:
- no-annotation
language_creators:
- found
source_datasets:
- original
task_categories:
- text-generation
- fill-mask
task_ids:
- language-modeling
- masked-language-modeling
license:
- cc-by-nc-sa-4.0
---
# Korpus Malti 🇲🇹

General Corpora for the Maltese Language.

This dataset is composed of texts from various genres/domains written in Maltese.


## Versions

This dataset is updated from time to time, and the latest version is obtained unless otherwise specified.
Consult the [changelog](CHANGELOG.md) for a detailed overview of each version released.

If you want to fetch a particular version, use the [`revision` argument](https://huggingface.co/docs/datasets/main/en/package_reference/loading_methods#datasets.load_dataset.revision).
For example, to get the data used to train [BERTu](https://huggingface.co/MLRS/BERTu), use the `4.0.0` tag:
```python
import datasets

dataset = datasets.load_dataset("MLRS/korpus_malti", revision="4.0.0")
```


## Configurations

### Shuffled data

The default configuration (`"shuffled"`) yields the entire corpus from all genres:
```python
import datasets

dataset = datasets.load_dataset("MLRS/korpus_malti")
```

All sentences are combined together and shuffled, without preserving the sentence order.
No other annotations are present, so an instance would be of the following form:
```json
{
  "text": "Din hija sentenza."
}
```

### Domain-split data

All other configurations contain a subset of the data.
The available data subsets are:
- `belles_lettres`: Literary texts, usually published and included in the corpus by permission of the copyright holder. Unfortunately these cannot be disseminated in their integral form.
- `blogs`: Online blog articles from specific blogs, identified in advance and known to contain text written (or human-translated into) Maltese.
- `comics`: A small set of online information about comic books in Maltese.
- `court`: Publicly available proceedings form the courts of Malta.
- `eu_docs`: Miscellaneous policy documents from the European Union institutions.
- `gov_docs`: Miscellaneous policy documents from the Government of Malta.
- `government_gazzette`: The official, publicly available gazette of the Government of Malta. The gazzette is bilingual; only the Maltese text is included.
- `law_eu`: Miscellaneous EU laws in their official Maltese translation, obtained via the Eur-Lex repository and including the segments of the Acquis Communautaire available in the DGT translation memory.
- `law_mt`: Maltese laws.
- `legal`: Miscellaneous legal text.
- `nonfiction`: Miscellaneous nonfiction, published or unpublished. Published texts are included with the permission of the copyright holder, where relevant.
- `parliament`: The officially released transcripts of parliamentary debates of the Maltese parliament.
- `press_eu`: Press releases in Maltese by the European Council of Ministers, European Parliament and European Commission.
- `press_mt`: Articles in the Maltese press, sourced primarily from the online portals of Maltese newspapers.
- `speeches`: Miscellaneous speeches in Maltese (pre-written).
- `theses`: Academic dissertations written in Maltese.
- `umlib_oar`: Very broad variety of nonfiction texts which are publicly available in the University of Malta Open Access Repository. Included with help and permission from the University of Malta library.
- `web_general`: Miscellaneous text scraped from pre-identified web pages in Maltese.
- `wiki`: The Maltese Wikipedia dump (downloaded 26th May, 2020).

For instance, this loads the Wikipedia portion:
```python
import datasets

dataset = datasets.load_dataset("MLRS/korpus_malti", "wiki")
```

For these configurations the data is not shuffled, so the sentence order on a document level is preserved.
An instance from these configurations would take the following form:
```json
{
  "text": ["Din hija sentenza.", "U hawn oħra!"],
  ...
}
```

The instances also contain additional metadata.
Their structure differs from one instance to another, depending on what's available from the source.
This information was typically scraped from the source itself & minimal processing is performed on such data.


## Additional Information

### Dataset Curators

The dataset was created by [Albert Gatt](https://albertgatt.github.io), [Kurt Micallef](https://www.kurtmica.com), [Marc Tanti](https://www.um.edu.mt/profile/marctanti), [Lonneke van der Plas](https://sites.google.com/site/lonnekenlp/) and [Claudia Borg](https://www.um.edu.mt/profile/claudiaborg).

### Licensing Information

This work is licensed under a [Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License][cc-by-nc-sa].
Permissions beyond the scope of this license may be available at [https://mlrs.research.um.edu.mt/](https://mlrs.research.um.edu.mt/).

[![CC BY-NC-SA 4.0][cc-by-nc-sa-image]][cc-by-nc-sa]

[cc-by-nc-sa]: http://creativecommons.org/licenses/by-nc-sa/4.0/
[cc-by-nc-sa-image]: https://licensebuttons.net/l/by-nc-sa/4.0/88x31.png

### Citation Information

This work was first presented in [Pre-training Data Quality and Quantity for a Low-Resource Language: New Corpus and BERT Models for Maltese](https://aclanthology.org/2022.deeplo-1.10/).
Cite it as follows: 

```bibtex
@inproceedings{BERTu,
    title = "Pre-training Data Quality and Quantity for a Low-Resource Language: New Corpus and {BERT} Models for {M}altese",
    author = "Micallef, Kurt  and
              Gatt, Albert  and
              Tanti, Marc  and
              van der Plas, Lonneke  and
              Borg, Claudia",
    booktitle = "Proceedings of the Third Workshop on Deep Learning for Low-Resource Natural Language Processing",
    month = jul,
    year = "2022",
    address = "Hybrid",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.deeplo-1.10",
    doi = "10.18653/v1/2022.deeplo-1.10",
    pages = "90--101",
}
```