---
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
dataset_info:
  features:
  - name: tokens
    sequence: string
  - name: ner_tags
    sequence: int64
  - name: dataset
    dtype: string
  splits:
  - name: test
    num_bytes: 16147720
    num_examples: 42144
  - name: train
    num_bytes: 161576681
    num_examples: 349195
  - name: validation
    num_bytes: 12398792
    num_examples: 33464
  download_size: 43074463
  dataset_size: 190123193
task_categories:
- token-classification
language:
- fr
size_categories:
- 100K<n<1M
license: cc-by-4.0
---

# Dataset information 
**Dataset concatenating NER datasets, available in French and open-source, for 3 entities (LOC, PER, ORG).**   
There are a total of **420,264** rows, of which 346,071 are for training, 32,951 for validation and 41,242 for testing.  
Our methodology is described in a blog post available in [English](https://blog.vaniila.ai/en/NER_en/) or [French](https://blog.vaniila.ai/NER/).


#  Usage
```
from datasets import load_dataset
dataset = load_dataset("CATIE-AQ/frenchNER_3entities")
```


# Dataset
## Details of rows
| Dataset Original    | Splits  | Note  |
| ----------- | ----------- | ----------- |
| [Multiconer](https://huggingface.co/datasets/aashsach/multiconer2)| 16,548 train / 857 validation / 0 test  | In practice, we use the original validation set as test set<br> and creat a new val set from 5% of train created, i.e.<br> 15,721 train / 827 validation / 857 test|       
| [Multinerd](https://huggingface.co/datasets/Babelscape/multinerd)| 140,880 train /  17,610 val / 17,695 test |   |       
| [Pii-masking-200k](https://huggingface.co/datasets/ai4privacy/pii-masking-200k)| 61,958 train / 0 validation / 0 test  | Only dataset without duplicate data or leaks |       
| [Wikiann](https://huggingface.co/datasets/wikiann)| 20,000 train / 10,000 val / 10,000 test  |   |       
| [Wikiner](https://huggingface.co/datasets/Jean-Baptiste/wikiner_fr)| 120,682 train / 0 validation / 13,410 test | In practice, 5% of val created from train set, i.e.<br> 113,296 train / 5,994 validation / 13,393 test |


## Removing duplicate data and leaks
The sum of the values of the datasets listed here gives the following result:

```
DatasetDict({
    train: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 351855
    })
    validation: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 34431
    })
    test: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 41945
    })
})
```

However, a data item in training split A may not be in A's test split, but may be present in B's test set, creating a leak when we create the A+B dataset.  
The same logic applies to duplicate data. So we need to make sure we remove them.  
After our clean-up, we finally have the following numbers:

```
DatasetDict({
    train: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 346071
    })
    validation: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 32951
    })
    test: Dataset({
        features: ['tokens', 'ner_tags', 'dataset'],
        num_rows: 41242
    })
})
```

Note: in practice, the test split contains 8 lines which we failed to deduplicate, i.e. 0.019%.


### Details of entities (after cleaning)

<table>
<thead>
  <tr>
    <th><br>Datasets</th>
    <th><br>Splits</th>
    <th><br>O</th>
    <th><br>PER</th>
    <th><br>LOC</th>  
    <th><br>ORG</th>
  </tr>
</thead>
<tbody>
    <tr>
    <td rowspan="3"><br>Multiconer</td>
    <td><br>train</td>
    <td><br>200,093</td>
    <td><br>18,060</td>
    <td><br>7,165</td>
    <td><br>6,967</td>
  </tr>
  <tr>
    <td><br>validation</td>
    <td><br>10,900</td>
    <td><br>1,069</td>
    <td><br>389</td>
    <td><br>328</td>
  </tr>
  <tr>
    <td><br>test</td>
    <td><br>11,287</td>
    <td><br>979</td>
    <td><br>387</td>
    <td><br>381</td>
  </tr>
    <tr>
    <td rowspan="3"><br>Multinerd</td>
    <td><br>train</td>
    <td><br>3,041,998</td>
    <td><br>149,128</td>
    <td><br>105,531</td>
    <td><br>68,796</td>
  </tr>
  <tr>
    <td><br>validation</td>
    <td><br>410,934</td>
    <td><br>17,479</td>
    <td><br>13,988</td>
    <td><br>3,475</td>
  </tr>
  <tr>
    <td><br>test</td>
    <td><br>417,886</td>
    <td><br>18,567</td>
    <td><br>14,083</td>
    <td><br>3,636</td>
  </tr>
    <tr>
    <td rowspan="1"><br>Pii-masking-200k</td>
    <td><br>train</td>
    <td><br>2,405,215</td>
    <td><br>29,838</td>
    <td><br>42,154</td>
    <td><br>12,310</td>
  </tr>
    <tr>
    <td rowspan="3"><br>Wikiann</td>
    <td><br>train</td>
    <td><br>60,165</td>
    <td><br>20,288</td>
    <td><br>17,033</td>
    <td><br>24,429</td>
  </tr>
  <tr>
    <td><br>validation</td>
    <td><br>30,046</td>
    <td><br>10,098</td>
    <td><br>8,698</td>
    <td><br>12,819</td>
  </tr>
  <tr>
    <td><br>test</td>
    <td><br>31,488</td>
    <td><br>10,764</td>
    <td><br>9,512</td>
    <td><br>13,480</td>
  </tr>
  <tr>
    <td rowspan="3"><br>Wikiner</td>
    <td><br>train</td>
    <td><br>2,691,294</td>
    <td><br>110,079</td>
    <td><br>131,839</td>
    <td><br>38,988</td>
  </tr>
  <tr>
    <td><br>validation</td>
    <td><br>140,935</td>
    <td><br>5,481</td>
    <td><br>7,204</td>
    <td><br>2,121</td>
  </tr>
  <tr>
    <td><br>test</td>
    <td><br>313,210</td>
    <td><br>13,324</td>
    <td><br>15,213</td>
    <td><br>3,894</td>
  </tr>
  <tr>
    <td rowspan="3"><br>Total</td>
    <td><br>train</td>
    <td><br><b>8,398,765</b></td>
    <td><br><b>327,393</b></td>
    <td><br><b>303,722</b></td>
    <td><br><b>151,490</b></td>
  </tr>
  <tr>
    <td><br>validation</td>
    <td><br><b>592,815</b></td>
    <td><br><b>34,127</b></td>
    <td><br><b>30,279</b></td>
    <td><br><b>18,743</b></td>
  </tr>
  <tr>
    <td><br>test</td>
    <td><br><b>773,871</b></td>
    <td><br><b>43,634</b></td>
    <td><br><b>39,195</b></td>
    <td><br><b>21,391</b></td>
  </tr>
</tbody>
</table>


## Columns
```
dataset_train = dataset['train'].to_pandas()
dataset_train.head()

 	tokens 	                                            ner_tags 	                                        dataset
0 	[On, a, souvent, voulu, faire, de, La, Bruyère... 	[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, ... 	wikiner
1 	[Les, améliorations, apportées, par, rapport, ... 	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 2, ... 	wikiner
2 	[Cette, assemblée, de, notables, ,, réunie, en... 	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, ... 	wikiner
3 	[Wittgenstein, projetait, en, effet, d', élabo... 	[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ... 	wikiner
4 	[Le, premier, écrivain, à, écrire, des, fictio... 	[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, ... 	wikiner

```

- the `tokens` column contains the tokens
- the `ner_tags` column contains the NER tags (IOB format with 0="O", 1="PER", 2="ORG" and 3="LOC")
- the `dataset` column identifies the row's original dataset (if you wish to apply filters to it)


## Split
- `train` corresponds to the concatenation of `multiconer` + `multinerd` + `pii-masking-200k` + `wikiann` + `wikiner` 
- `validation` corresponds to the concatenation of `multiconer` + `multinerd` + `wikiann` + `wikiner` 
- `test` corresponds to the concatenation of `multiconer` + `multinerd` + `wikiann` + `wikiner` 



# Citations

### multiconer
```
@inproceedings{multiconer2-report,  
    title={{SemEval-2023 Task 2: Fine-grained Multilingual Named Entity Recognition (MultiCoNER 2)}},  
    author={Fetahu, Besnik and Kar, Sudipta and Chen, Zhiyu and Rokhlenko, Oleg and Malmasi, Shervin},  
    booktitle={Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)},  
    year={2023},  
    publisher={Association for Computational Linguistics}}


@article{multiconer2-data,  
    title={{MultiCoNER v2: a Large Multilingual dataset for Fine-grained and Noisy Named Entity Recognition}},  
    author={Fetahu, Besnik and Chen, Zhiyu and Kar, Sudipta and Rokhlenko, Oleg and Malmasi, Shervin},  
    year={2023}}
```

### multinerd
```
@inproceedings{tedeschi-navigli-2022-multinerd,  
    title = "{M}ulti{NERD}: A Multilingual, Multi-Genre and Fine-Grained Dataset for Named Entity Recognition (and Disambiguation)",  
    author = "Tedeschi, Simone and  Navigli, Roberto",  
    booktitle = "Findings of the Association for Computational Linguistics: NAACL 2022",  
    month = jul,  
    year = "2022",  
    address = "Seattle, United States",  
    publisher = "Association for Computational Linguistics",  
    url = "https://aclanthology.org/2022.findings-naacl.60",  
    doi = "10.18653/v1/2022.findings-naacl.60",  
    pages = "801--812"}
```

### pii-masking-200k
```
@misc {ai4privacy_2023,  
    author = { {ai4Privacy} },  
    title = { pii-masking-200k (Revision 1d4c0a1) },  
    year = 2023,  
    url = { https://huggingface.co/datasets/ai4privacy/pii-masking-200k },  
    doi = { 10.57967/hf/1532 },  
    publisher = { Hugging Face }}
```

### wikiann
```
@inproceedings{rahimi-etal-2019-massively,  
    title = "Massively Multilingual Transfer for {NER}",  
    author = "Rahimi, Afshin and Li, Yuan and Cohn, Trevor",  
    booktitle = "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics",  
    month = jul,  
    year = "2019",  
    address = "Florence, Italy",  
    publisher = "Association for Computational Linguistics",  
    url = "https://www.aclweb.org/anthology/P19-1015",  
    pages = "151--164"}
```

### wikiner
```
 @article{NOTHMAN2013151,  
    title = {Learning multilingual named entity recognition from Wikipedia},  
    journal = {Artificial Intelligence},  
    volume = {194},  
    pages = {151-175},  
    year = {2013},  
    note = {Artificial Intelligence, Wikipedia and Semi-Structured Resources},  
    issn = {0004-3702},  
    doi = {https://doi.org/10.1016/j.artint.2012.03.006},  
    url = {https://www.sciencedirect.com/science/article/pii/S0004370212000276},  
    author = {Joel Nothman and Nicky Ringland and Will Radford and Tara Murphy and James R. Curran}}
```

### frenchNER_3entities
```
@misc {frenchNER2024,  
    author       = { {BOURDOIS, Loïck} },  
    organization  = { {Centre Aquitain des Technologies de l'Information et Electroniques} },  
    title        = { frenchNER_3entities },  
    year         = 2024,  
    url          = { https://huggingface.co/CATIE-AQ/frenchNER_3entities },  
    doi          = { 10.57967/hf/1751 },  
    publisher    = { Hugging Face }  
}
```

# License
[cc-by-4.0](https://creativecommons.org/licenses/by/4.0/deed.en)