---
configs:
- config_name: hau
  data_files:
  - split: train
    path: hausa/hau_dictionary.csv
- config_name: ibo
  data_files:
  - split: train
    path: igbo/ibo_dictionary.csv
- config_name: swh
  data_files:
  - split: train
    path: swahili/swa_dictionary.csv
- config_name: yor
  data_files:
  - split: train
    path: yoruba/yor_dictionary.csv
- config_name: yor_v1
  data_files:
  - split: train
    path: yoruba/v1/entries.csv.gz
- config_name: en_yor_v1
  data_files:
  - split: train
    path: yoruba/reverse_v1/english-to-yoruba-v1.csv.gz
- config_name: hau_v1
  data_files:
  - split: train
    path: hausa/v1/entries.csv.gz
- config_name: en_hau_v1
  data_files:
  - split: train
    path: hausa/reverse_v1/english-to-hausa-v1.csv.gz
- config_name: ibo_v1
  data_files:
  - split: train
    path: igbo/v1/entries.csv.gz
- config_name: en_ibo_v1
  data_files:
  - split: train
    path: igbo/reverse_v1/english-to-igbo-v1.csv.gz
- config_name: swh_v1
  data_files:
  - split: train
    path: swahili/v1/entries.csv.gz
- config_name: en_swh_v1
  data_files:
  - split: train
    path: swahili/reverse_v1/english-to-swahili-v1.csv.gz
license: mit
language:
- ha
- ig
- sw
- yo
task_categories:
- translation
pretty_name: Afri-Dict
tags:
- african-languages
- dictionary
- nlp
- lexicon
dataset_info:
  features:
  - name: word
    dtype: string
  - name: pos
    dtype: string
  - name: definition
    dtype: string
---

# Afri-Dict

## Dataset Summary
`afri-dict` is a bilingual dictionary dataset for four major African languages: Hausa, Igbo, Swahili, and Yoruba. 
Entries include a headword, part-of-speech tag, and definition in English or the target African language.
This dataset can serve as a foundational resource for machine translation systems, language learning tools, spell checkers, cross-lingual search, and other NLP applications for African languages.

## Languages
The dataset is organized into subsets by language. Note that the headword language in the `word` field varies by subset.

| ISO Code | Language | Size (Entries) | Mapping Direction |
| :--- | :--- | :--- | :--- |
| `hau` | Hausa | 4,920 | English ↔ Hausa |
| `ibo` | Igbo |  | Igbo ↔ English |
| `swh` | Swahili | 9,760 | English ↔ Swahili |
| `yor` | Yoruba | 12,200 | Yoruba ↔ English |


## Dataset Structure

### Data Fields
* **`word`**: This is the  headword or source term.
* **`pos`**: Part of speech for the headword.
* **`definition`**: The equivalent word or an explanatory definition in the target language.  In some cases, multiple synonyms are provided, separated by commas.


## Usage
```python
from datasets import load_dataset

# Load specific language
hausa = load_dataset("taresco/afri-dict", "hau")
swahili = load_dataset("taresco/afri-dict", "swh")

# Access the first entry
print(hausa['train'][0])
print(swahili['train'][0])
```

## Citation Information
If you use this dataset in your research or project, please cite it as follows:

```
@misc{taresco2025afri-dict,
  author = {The African Research Collective},
  title = {Afri-Dict},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{[https://huggingface.co/datasets/taresco/afri-dict](https://huggingface.co/datasets/taresco/afri-dict)}}
}
```