yasalma/tg-messages

Dataset Summary

This dataset contains structured textual content in Markdown format extracted from Telegram messages written in the Tatar language. The messages were collected from several public Tatar-speaking groups and channels. The dataset is intended to provide clean, structured, and semantically meaningful conversational content to support natural language processing tasks, social media analysis, and research in Tatar language technologies.

The extracted Markdown preserves key structural elements such as:

• Message text and replies (threading via parent relationships)
• Timestamps of messages
• Sender information (anonymized)
• Basic formatting such as bold, italics, and links
• Forwarded message indicators

Extraneous elements such as media files, system notifications, and irrelevant metadata have been removed to provide clean and usable text.

Dataset Structure

The dataset is provided in a single Parquet file with the following columns:

• id: Internal primary key of the message (autoincremented)
• telegram_id: Original Telegram message ID
• parent_id: Parent message ID within the dataset (for threading)
• reply_to_msg_id: Telegram reply target ID (may differ from parent_id)
• group_id: Identifier of the Telegram group or channel where the message was posted
• message: Message text content in Markdown format
• username: Anonymized username of the sender (if available, may be missing)
• url: Link to the original Telegram message
• date: Message timestamp (Unix time in seconds)
• reactions: Serialized blob of reactions to the message

Each row represents a single Telegram message.

Reactions format

The reactions column stores data as a UTF-8 encoded JSON array of objects. Each object includes:

• emoticonBlob: Encoded representation of the reaction emoticon
• count: Number of times this reaction was used

Example decoded JSON:

[
{ "emoticonBlob": "😀", "count": 5 },
{ "emoticonBlob": "🔥", "count": 2 }
]

Supported Tasks and Use Cases

This dataset supports various use cases, including but not limited to:

• Pretraining or fine-tuning language models for the Tatar language
• Building Markdown-aware content models
• Semantic document analysis
• Text segmentation, summarization, or classification
• Cross-lingual NLP (e.g., aligning Tatar with other languages)

Languages

• Tatar (tt, script: Cyrillic)

License

This dataset is released under the MIT License.

You are free to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the dataset, subject to the following condition:

• The original copyright notice and permission notice must be included in all copies or substantial portions of the dataset.

See the full license text here: https://opensource.org/license/mit/

Citation

If you use this dataset in your work, please cite it as:

@dataset {tt_structured_content, author = {yasalma}, title = {Tatar telegram messages}, year = {2025}, url = {https://huggingface.co/datasets/yasalma/tg-messages} }

Contributions

This dataset was created and maintained by yasalma. Contributions and suggestions are welcome.