Hugging Face's logo

Datasets:
Subayyal
/
Urdu-Low-Resource-Language-Dataset

Tasks:
Summarization
Feature Extraction
Text2Text Generation
Modalities:
Text
Languages:
Urdu
DOI:
License:
Dataset card Viewer Files Community
Urdu-Low-Resource-Language-Dataset
File size: 3,135 Bytes 
723b9f4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
 
---
license: apache-2.0
---

# Dataset Card for Low-Resource-Language-Dataset

This dataset is designed to aid Natural Language Processing (NLP) research on low-resource languages, particularly Urdu. It includes structured datasets and preprocessing tools curated from the BBC Urdu website.

## Dataset Details

### Dataset Description

This dataset contains articles, summaries, and topics scraped from [BBC Urdu](https://www.bbc.com/urdu). It is structured into training and testing datasets for machine learning applications. The data is preprocessed and tokenized to support text summarization and other NLP tasks.

- **Curated by:** Subayyal Sheikh and contributors.
- **Language(s):** Urdu.
- **License:** Apache 2.0 License.

### Dataset Sources

- **Repository:** [GitHub Repository](https://github.com/subayyal802/Low-Resource-Language-Dataset)
- **Paper:** *Breaking Language Barriers: Dataset Development for Low Resource Language*
- **Demo:** None (optional if available)

## Uses

### Direct Use

The dataset is ideal for:
- Text summarization
- Language modeling
- Topic classification

### Out-of-Scope Use

This dataset may not be suitable for tasks outside of Urdu language processing. Users are advised against malicious or inappropriate use of the data.

## Dataset Structure

The dataset is structured as follows:
- **JSONL Files**:
  - `Articles`, `Topics`, and `Summaries` (raw and processed).
  - Training and test splits (`Articles-train`, `Articles-test`, etc.).
  - Preprocessed articles (`BBCArticle512`, `BBCArticle512J`, etc.).
- **Text Files**:
  - Token lengths (`Length-Article-512`, `Length-Summary-512`, etc.).
  - Logs and ratio calculations (`Log`, `Ratios-512`, `TestRatio-512`).

## Dataset Creation

### Curation Rationale

The dataset was created to address the lack of resources for NLP research in the Urdu language. It facilitates diverse NLP applications.

### Source Data

#### Data Collection and Processing

Data was collected using scraping scripts and processed to normalize and tokenize text into usable formats.

#### Who are the source data producers?

The source data is curated from publicly available articles on [BBC Urdu](https://www.bbc.com/urdu).

### Annotations

The dataset does not include human annotations; all content is machine-extracted and processed.

#### Personal and Sensitive Information

The dataset contains publicly available data and does not include personal or sensitive information.

## Bias, Risks, and Limitations

- The dataset may contain biases inherent in its source, such as topic selection by the BBC Urdu editorial team.
- Summaries and topics may not align perfectly with the user's context.

### Recommendations

Users should consider dataset limitations and apply domain knowledge when using the data for specific tasks.

## Citation

If you use this dataset in your research, please cite:
```bibtex
@article{Subayyal2024,
  title={Breaking Language Barriers: Dataset Development for Low Resource Language},
  author={Subayyal Sheikh, Yasir Jan, Masab A. Javaid, Ammad Khalil, Jebran Khan},
  journal={PeerJ Computer Science},
  year={2024}
}