README.md

---
dataset_info:
  features:
  - name: docid
    dtype: string
  - name: title
    dtype: string
  - name: text
    dtype: string
  splits:
  - name: corpus
    num_bytes: 5473924474
    num_examples: 14047759
  download_size: 3101484658
  dataset_size: 5473924474
configs:
- config_name: default
  data_files:
  - split: corpus
    path: data/corpus-*
language:
- es
pretty_name: MessIRve (corpus)
---

# Dataset Card for MessIRve (corpus)

<!-- Provide a quick summary of the dataset. -->

This is the Spanish Wikipedia corpus for MessIRve. A document corresponds to a paragraph in Wikipedia. We used [WikiExtractor](https://github.com/attardi/wikiextractor) to process the Spanish Wikipedia dump of 2024-04-01. We provide the title of the Wikipedia articles to use as additional context when retrieving. 


## Dataset Details


### Dataset Sources

<!-- Provide the basic links for the dataset. -->

- **Repository:** TBA
- **Paper:** [MessIRve: A Large-Scale Spanish Information Retrieval Dataset](http://arxiv.org/abs/2409.05994)


## Dataset Structure

<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->

### Data Instances

A typical instance in the dataset looks like:

```json
{
    "docid": "17#9",
    "title": "Demografía de Andorra", 
    "text": "El mayor núcleo de población del Principado es la capital, Andorra la Vieja, con 24.574 habitantes (2008), lo que representa el 29,6% de la población andorrana. Las siguientes parroquias por población son Las Escaldas-Engordany (16.474 hab.) y Encamp (14.030hab.)."
}
```

Following [MIRACL](https://huggingface.co/datasets/miracl/miracl-corpus), `docid` has the pattern X#Y, where Y indicates the sequentially numbered passage inside the Wikipedia article, and all passages with the same X are from the same article. The article name from which the paragraph is taken is in the `title` field.


### Data Fields

- `docid`: document ID in the corpus
- `title`: article title
- `text`: document text (a passage)

<!-- Note that the descriptions can be initialized with the **Show Markdown Data Fields** output of the [Datasets Tagging app](https://huggingface.co/spaces/huggingface/datasets-tagging), you will then only need to refine the generated descriptions. -->


## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

```bibtex
@article{valentini2024messirve,
      title={MessIRve: A Large-Scale Spanish Information Retrieval Dataset}, 
      author={Francisco Valentini and Viviana Cotik and Damián Furman and Ivan Bercovich and Edgar Altszyler and Juan Manuel Pérez},
      year={2024},
      eprint={2409.05994},
      journal={arxiv:2409.05994},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2409.05994}, 
}
```

**APA:**

Francisco Valentini, Viviana Cotik, Damián Furman, Ivan Bercovich, Edgar Altszyler, & Juan Manuel Pérez (2024). MessIRve: A Large-Scale Spanish Information Retrieval Dataset. arxiv:2409.05994.