README.md

---
license: cc-by-nc-sa-4.0
source_datasets:
- extended
language_creators:
- found
multilinguality:
- multilingual
language:
- bg
- cs
- da
- de
- el
- en
- es
- et
- fi
- fr
- hr
- hu
- it
- lt
- lv
- mt
- nl
- pl
- pt
- ro
- sk
- sl
- sv
tags:
- politics
size_categories:
- 10K<n<100K
pretty_name: EU Debates
---


# Dataset Description

EU Debates is a corpus of parliamentary proceedings (debates) from the European parliament released by [Chalkidis and Brandl (2024)](https://arxiv.org/abs/2403.13592). The corpus consists of approx. 87k individual speeches in the period 2009-2023. 
We exhaustively scrape the data from the official European Parliament Plenary website ([Link](https://www.europarl.europa.eu/)). All speeches are time-stamped, thematically organized on debates, 
and include metadata relevant to the speaker's identity (full name, euro-party affiliation, speaker role), and the debate (date and title). 
Older debate speeches are originally in English, while newer ones are linguistically diverse across the 23 official EU languages, thus we also provide machine-translated 
versions in English, when official translations are missing, using the EasyNMT framework with the [M2M2-100 (418M)](https://huggingface.co/facebook/m2m100_418M) model (Fan et al., 2020).

# Data Fields

- `speaker_name: a `string` with the full name of the speaker.
- `speaker_party`: a `string` with the name of the euro-party (group) that the MEP is affiliated with.
- `speaker_role`:  a `string` with the role of the speaker (Member of the European Parliament (MEP), EUROPARL President, etc.)
- `debate_title`:  a `string` with the title of the debate in the European Parliament.
- `date`:  a `string` with the full date (YYYY-MM-DD) of the speech.
- `year` a `string` with the year (YYYY).
- `text`: a `string` with the full speech of the speaker.
- `translated_text`: a `string` with the translation of the speech in English, if the original is not.


# Data Instances

Example of a data instance from the EU Debates dataset:

```
{
'speaker_name': 'Michèle Striffler'
'speaker_party': 'PPE',
'speaker_role': 'MEP',
'debate_title': 'Famine in East Africa (debate)',
'date': '2011-09-15',
'year': '2011'
'text': "Monsieur le Président, Madame le Commissaire, chers collègues, la situation humanitaire sans précédent que connaît la Corne de l'Afrique continue [...]",
'translated_text': 'Mr. President, Mr. Commissioner, dear colleagues, the unprecedented humanitarian situation of the Horn of Africa continues [...]'}
}
```

# How to use

```python

from datasets import load_dataset
eu_debates_dataset = load_dataset('coastalcph/eu_debates', split='train')

```

# Dataset Statistics

Distribution of speeches across euro-parties:

<table>
<tr><td> <b>Euro-party</b> </td> <td> <b>No. of Speeches</b> </td> </tr>
<tr><td>EPP  </td> <td> 25,455 (29%)</td> </tr>
<tr><td>S&D </td> <td> 20,042 (23%)</td> </tr>
<tr><td>ALDE </td> <td> 8,946 (10%)</td> </tr>
<tr><td>ECR </td> <td> 7,493 (9%)</td> </tr>
<tr><td>ID </td> <td> 6,970 (8%) </td> </tr>
<tr><td>GUE/NGL </td> <td>6,780 (8%)</td> </tr>
<tr><td>Greens/EFA </td> <td> 6,398 (7%)</td> </tr>
<tr><td>NI </td> <td> 5,127 (6%)</td> </tr>
<tr><td>Total </td> <td> 87,221 </td> </tr>
</table>


Distribution of speeches across years and euro-parties:

<table>
<tr><td><b>Year</b></td><td><b>EPP</b></td><td><b>S&D</b></td><td><b>ALDE</b></td><td><b>ECR</b></td><td><b>ID</b></td><td><b>GUE/NGL</b></td><td><b>Greens/EFA</b></td><td><b>NI</b></td><td><b>Total</b></td></tr>
<tr><td> 2009 </td><td> 748 </td><td> 456 </td><td> 180 </td><td> 138 </td><td> 72 </td><td> 174 </td><td> 113 </td><td> 163 </td><td> 2044 </td></tr>
<tr><td> 2010 </td><td> 3205 </td><td> 1623 </td><td> 616 </td><td> 340 </td><td> 341 </td><td> 529 </td><td> 427 </td><td> 546 </td><td> 7627 </td></tr>
<tr><td> 2011 </td><td> 4479 </td><td> 2509 </td><td> 817 </td><td> 418 </td><td> 761 </td><td> 792 </td><td> 490 </td><td> 614 </td><td> 10880 </td></tr>
<tr><td> 2012 </td><td> 3366 </td><td> 1892 </td><td> 583 </td><td> 419 </td><td> 560 </td><td> 486 </td><td> 351 </td><td> 347 </td><td> 8004 </td></tr>
<tr><td> 2013 </td><td> 724 </td><td> 636 </td><td> 240 </td><td> 175 </td><td> 152 </td><td> 155 </td><td> 170 </td><td> 154 </td><td> 2406 </td></tr>
<tr><td> 2014 </td><td> 578 </td><td> 555 </td><td> 184 </td><td> 180 </td><td> 131 </td><td> 160 </td><td> 144 </td><td> 180 </td><td> 2112 </td></tr>
<tr><td> 2015 </td><td> 978 </td><td> 1029 </td><td> 337 </td><td> 405 </td><td> 398 </td><td> 325 </td><td> 246 </td><td> 240 </td><td> 3958 </td></tr>
<tr><td> 2016 </td><td> 919 </td><td> 972 </td><td> 309 </td><td> 387 </td><td> 457 </td><td> 317 </td><td> 225 </td><td> 151 </td><td> 3737 </td></tr>
<tr><td> 2017 </td><td> 649 </td><td> 766 </td><td> 181 </td><td> 288 </td><td> 321 </td><td> 229 </td><td> 162 </td><td> 135 </td><td> 2731 </td></tr>
<tr><td> 2018 </td><td> 554 </td><td> 611 </td><td> 161 </td><td> 242 </td><td> 248 </td><td> 175 </td><td> 160 </td><td> 133 </td><td> 2284 </td></tr>
<tr><td> 2019 </td><td> 1296 </td><td> 1339 </td><td> 719 </td><td> 556 </td><td> 513 </td><td> 463 </td><td> 490 </td><td> 353 </td><td> 5729 </td></tr>
<tr><td> 2020 </td><td> 1660 </td><td> 1564 </td><td> 823 </td><td> 828 </td><td> 661 </td><td> 526 </td><td> 604 </td><td> 346 </td><td> 7012 </td></tr>
<tr><td> 2021 </td><td> 2147 </td><td> 2189 </td><td> 1290 </td><td> 1062 </td><td> 909 </td><td> 708 </td><td> 990 </td><td> 625 </td><td> 9920 </td></tr>
<tr><td> 2022 </td><td> 2436 </td><td> 2273 </td><td> 1466 </td><td> 1177 </td><td> 827 </td><td> 962 </td><td> 1031 </td><td> 641 </td><td> 10813 </td></tr>
<tr><td> 2023 </td><td> 1716 </td><td> 1628 </td><td> 1040 </td><td> 878 </td><td> 619 </td><td> 779 </td><td> 795 </td><td> 499 </td><td> 7954 </td></tr>
</table>

Distribution of speeches across the 23 EU official languages:

| Language    | Examples |
| ----------- | -------- | 
| en | 40736  (46.7%) |
| de | 6497   (7.5%) |
| fr | 6024   (6.9%) |
| es | 5172   (5.9%) |
| it | 4506   (5.2%) |
| pl | 3792   (4.4%) |
| pt | 2713   (3.1%) |
| ro | 2308   (2.7%) |
| el | 2290   (2.6%) |
| nl | 2286   (2.6%) |
| hu | 1661   (1.9%) |
| hr | 1509   (1.7%) |
| cs | 1428   (1.6%) |
| sv | 1210   (1.4%) |
| bg | 928    (1.1%) |
| sk | 916    (1.1%) |
| sl | 753    (0.9%) |
| fi | 693    (0.8%) |
| lt | 618    (0.7%) |
| da | 578    (0.7%) |
| et | 342    (0.4%) |
| lv | 184    (0.2%) |
| mt |   0    (0.0%) |



# Citation Information

*[Llama meets EU: Investigating the European political spectrum through the lens of LLMs. 
Ilias Chalkidis and Stephanie Brandl. 
In the Proceedings of the Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 
Mexico City, Mexico, June 16–21, 2024.](https://arxiv.org/abs/2403.13592)*

```
@inproceedings{chalkidis-and-brandl-eu-llama-2024,
    title = "Llama meets EU: Investigating the European political spectrum through the lens of LLMs",
    author = "Chalkidis, Ilias  and Brandl, Stephanie",
    booktitle = "Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics",
    month = jun,
    year = "2024",
    address = "Mexico City, Mexico",
    publisher = "Association for Computational Linguistics",
}

```