selimfirat/bilkent-turkish-writings-dataset

Compilation of Bilkent Turkish Writings Dataset

Dataset Description

This is a comprehensive compilation of Turkish creative writings from Bilkent University's Turkish 101 and Turkish 102 courses (2014-2025). The dataset contains 9119 student writings originally created by students and instructors, focusing on creativity, content, composition, grammar, spelling, and punctuation development.

Note: This dataset is a compilation and digitization of publicly available writings from Bilkent University. The original content was created by students and instructors of the Turkish Department. The dataset compilation, processing, and distribution tools were developed by Selim F. Yilmaz.

Dataset Information

• Version: v2
• Date Created: 2025-05-24
• Number of Entries: 9119
• Language: Turkish
• License: Academic Use Only
• Original Source: Bilkent University Turkish Department

Dataset Structure

Data Fields

• text: The full text content of the writing

Data Splits

This dataset is provided as a single dataset without predefined splits. You can create your own splits as needed:

# Create custom splits if needed
dataset = dataset.train_test_split(test_size=0.2)
train_data = dataset["train"]
test_data = dataset["test"]

Usage

from datasets import load_dataset

# Load the latest version (v2) - default configuration
dataset = load_dataset("selimfirat/bilkent-turkish-writings-dataset")

# Access the data
for item in dataset['train'].take(5):
    print(f"Text: {item['text'][:100]}...")
    print("---")

# Create custom splits if needed
split_dataset = dataset['train'].train_test_split(test_size=0.2)
train_data = split_dataset["train"]
test_data = split_dataset["test"]

Version Information

This dataset provides multiple configurations:

• Default (v2): Latest dataset with 9,119 entries (2014-2025) - Recommended
• v1: Original dataset with 6,844 entries (2014-2018)
• v2: Same as default, explicitly named configuration

Accessing Different Versions

from datasets import load_dataset

# Method 1: Load default version (v2 - recommended)
dataset = load_dataset("selimfirat/bilkent-turkish-writings-dataset")

# Method 2: Explicitly load v2 configuration
dataset_v2 = load_dataset("selimfirat/bilkent-turkish-writings-dataset", "v2")

# Method 3: Load v1 configuration (original version)
dataset_v1 = load_dataset("selimfirat/bilkent-turkish-writings-dataset", "v1")

# All methods above return the same structure
print(f"Number of entries: {len(dataset['train'])}")
print(f"First text preview: {dataset['train'][0]['text'][:100]}...")

Citation

If you use this dataset in your research, please cite:

@dataset{yilmaz_bilkent_turkish_writings_2025,
  title={Compilation of Bilkent Turkish Writings Dataset},
  author={Yilmaz, Selim F.},
  year={2025},
  publisher={Zenodo},
  doi={10.5281/zenodo.15498155},
  url={https://doi.org/10.5281/zenodo.15498155},
  version={v2},
  note={Compilation of Turkish creative writings from Bilkent University Turkish 101 and 102 courses (2014-2025). Original content by Bilkent University students and instructors.}
}

License

This dataset is released under an Academic Use License. Commercial use is prohibited. See the LICENSE file for full details.

Acknowledgments

• Original Content Creators: Students and instructors of Turkish 101 and 102 courses at Bilkent University (2014-2025)
• Bilkent University Turkish Department for creating and publishing the original writings
• Dataset Compiler: Selim F. Yilmaz for collecting, processing, and structuring the dataset

Contact

For questions or issues, please visit the GitHub repository.