|
|
--- |
|
|
license: cc-by-4.0 |
|
|
task_categories: |
|
|
- summarization |
|
|
- text-classification |
|
|
- text-generation |
|
|
language: |
|
|
- tr |
|
|
pretty_name: htv-news |
|
|
size_categories: |
|
|
- 100K<n<1M |
|
|
--- |
|
|
|
|
|
# HalkTV-News Dataset |
|
|
|
|
|
## Dataset Summary |
|
|
HalkTV-News is a Turkish news dataset containing 173,179 articles collected from Halk TV's news portal. Each article includes a title, summary, full content, and a source URL. The dataset is designed for tasks such as text classification, summarization, and topic modeling in Turkish. It provides a rich resource for understanding and modeling Turkish news content. |
|
|
|
|
|
## Supported Tasks and Leaderboards |
|
|
- **Text Summarization**: With included summaries, the dataset can be used for abstractive or extractive summarization tasks. |
|
|
- **Text Classification**: The dataset can be used to classify news articles into categories or topics. |
|
|
- **Sentiment Analysis**: The content can be analyzed for sentiments in Turkish. |
|
|
- **Topic Modeling**: Suitable for discovering hidden themes in news content. |
|
|
|
|
|
## Languages |
|
|
The dataset is entirely in Turkish (`tr`). |
|
|
|
|
|
## Dataset Structure |
|
|
The dataset contains the following columns: |
|
|
- `Başlık`: The title of the article. |
|
|
- `Özet`: A brief summary of the article. |
|
|
- `İçerik`: The full content of the article. |
|
|
- `Link`: The URL linking to the full article on the Halk TV website. |
|
|
|
|
|
#### Example Row: |
|
|
| Başlık | Özet | İçerik | Link | |
|
|
|------------------------------------------|---------------------------------------|--------------------------|------------------------------------------------| |
|
|
| Kepçe operatörüne yol gösterirken... | Artvin’in Şavşat ilçesinde... | İlçeye bağlı Maden... | https://halktv.com.tr/gundem/kepce-operator...| |
|
|
|
|
|
## Data Splits |
|
|
The dataset is provided as a single split with 173,180 examples. Users can split the dataset into train, validation, and test sets as needed. |
|
|
|
|
|
## Source Data |
|
|
The dataset was collected from Halk TV's publicly available news portal. The articles were programmatically scraped and processed to extract relevant metadata such as title, summary, and content. |
|
|
|
|
|
## Licensing Information |
|
|
This dataset is available for research purposes only. Please ensure compliance with the original data sources' terms of use. |
|
|
|
|
|
## Citation |
|
|
If you use this dataset, please cite the following: |
|
|
- N. Z. Kayalı and S. I. Omurca, "HTV-News: A New Dataset with High Novelty Rate for Turkish Text Summarization," 2024 9th International Conference on Computer Science and Engineering (UBMK), Antalya, Turkiye, 2024, pp. 1-6, doi: 10.1109/UBMK63289.2024.10773431. https://ieeexplore.ieee.org/document/10773431 |
