README.md

---
license: cc-by-sa-4.0
task_categories:
- question-answering
language:
- bem
- fon
- ha
- ig
- kin
- sw
- wo
- yo
- zu
- tw
pretty_name: AfriQA
size_categories:
- 10K<n<100K
multilinguality:
- multilingual
tags:
- cross-lingual
- question-answering
- qa
---

# Dataset Card for AfriQA

## Table of Contents
- [Table of Contents](#table-of-contents)
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
  - [Curation Rationale](#curation-rationale)
  - [Source Data](#source-data)
  - [Annotations](#annotations)
  - [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
  - [Social Impact of Dataset](#social-impact-of-dataset)
  - [Discussion of Biases](#discussion-of-biases)
  - [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
  - [Dataset Curators](#dataset-curators)
  - [Licensing Information](#licensing-information)
  - [Citation Information](#citation-information)
  - [Contributions](#contributions)

## Dataset Description

- **Homepage:** [homepage](https://github.com/masakhane-io/afriqa)
- **Repository:** [github](https://github.com/masakhane-io/afriqa)
- **Paper:** [paper]()
- **Point of Contact:** [Masakhane](https://www.masakhane.io/) or oogundep@uwaterloo.ca

### Dataset Summary

AfriQA is the first cross-lingual question answering (QA) dataset with a focus on African languages. The dataset includes over 12,000 XOR QA examples across 10 African languages, making it an invaluable resource for developing more equitable QA technology.

The train/validation/test sets are available for all the 10 languages.


### Supported Tasks and Leaderboards

- `question-answering`: The performance in this task is measured with [F1](https://huggingface.co/metrics/f1) (higher is better) and [Exact Match Accuracy](https://huggingface.co/spaces/evaluate-metric/exact_match).

### Languages

There are 20 languages available :
- Bemba (bem)
- Fon (fon)
- Hausa (hau)
- Igbo (ibo)
- Kinyarwanda (kin)
- Swahili (swą)
- Twi (twi)
- Wolof (wol)
- Yorùbá (yor)
- Zulu (zul)

## Dataset Structure

### Data Instances

- Data Format:
- id : Question ID
- question : Question in African Language
- translated_question : Question translated into a pivot language (English/French)
- answers : Answer in African Language
- lang : Datapoint Language (African Language) e.g `bem`
- split : Dataset Split
- translated_answer : Answer in Pivot Language
- translation_type : Translation type of question and answers


```bash
{   "id": 0, 
    "question": "Bushe icaalo ca Egypt caali tekwapo ne caalo cimbi?", 
    "translated_question": "Has the country of Egypt been colonized before?", 
    "answers": "['Emukwai']", 
    "lang": "bem", 
    "split": "dev", 
    "translated_answer": "['yes']", 
    "translation_type": "human_translation"
    }
```

### Data Splits

For all languages, there are three splits.

The original splits were named `train`, `dev` and `test` and they correspond to the `train`, `validation` and `test` splits.

The splits have the following sizes :

| Language        | train | dev | test |
|-----------------|------:|-----------:|-----:|
| Bemba         |  502 | 503 |  314 |
| Fon          |  427 | 428 |  386 |
| Hausa           |  435 | 436 |  300 |
| Igbo            |  417 | 418 |  409 |
| Kinyarwanda     	  |   407 |  409 |  347 |
| Swahili       |  415 |   417 |  302 |
| Twi          |  451 |   452 |  490 |
| Wolof        |  503 |    504 |  334 |
| Yoruba          |  360 |   361 |  332 |
| Zulu        |  387 |    388 |  325 |
| <b>Total</b>    |  <b>4333</b>  |  <b>4346</b>  |<b>3560</b>  |

## Dataset Creation

### Curation Rationale

The dataset was introduced to introduce question-answering resources to 10 languages that were under-served for natural language processing.

[More Information Needed]

### Source Data

...

#### Initial Data Collection and Normalization

...

#### Who are the source language producers?

...

### Annotations

#### Annotation process

Details can be found here ...

#### Who are the annotators?

Annotators were recruited from [Masakhane](https://www.masakhane.io/)

### Personal and Sensitive Information

...

## Considerations for Using the Data

### Social Impact of Dataset
[More Information Needed]


### Discussion of Biases
[More Information Needed]


### Other Known Limitations

Users should keep in mind that the dataset only contains news text, which might limit the applicability of the developed systems to other domains.

## Additional Information

### Dataset Curators


### Licensing Information

The licensing status of the data is CC 4.0 Non-Commercial

### Citation Information

Provide the [BibTex](http://www.bibtex.org/)-formatted reference for the dataset. For example:

```
Incoming ...
```

### Contributions

Thanks to [@ToluClassics](https://github.com/ToluClassics) for adding this dataset.