sixfingerdev's picture
Update README.md
0024fad verified
metadata
license: apache-2.0

Turkish Wikipedia Topic-to-Summary Dataset

This dataset consists of title–summary pairs extracted from the Turkish Wikipedia XML dump. Each entry contains a topic title as the input and a cleaned, HTML-free summary generated from the first paragraph of the corresponding article as the output. It is suitable for training language models, retrieval systems, and knowledge extraction tasks.

Format

The dataset is provided in JSONL format. Each line represents a single record:

{
  "input": "Title",
  "output": "A short description or summary about the given title."
}

Cleaning Process

The following preprocessing steps were applied:

Removal of HTML tags

Removal of HTML entities such as  , &, "

Removal of template structures ({{ ... }})

Removal of Infobox fields and markup

Removal of wiki links ([[Page]], [[Page|Text]])

Normalization of whitespace

Removal of section headers (== Heading ==)

Extraction of the first paragraph or first few sentences

All summaries are plain text and designed to be model-friendly.

Possible Use Cases

Training retrieval and ranking models

Topic-to-text or title-to-abstract generation

Knowledge extraction and factual reasoning tasks

Pretraining or finetuning LLMs on structured encyclopedic data

Building question answering systems with short factual outputs

Source

The dataset is derived from the publicly available Turkish Wikipedia dump. It does not contain any proprietary content and follows Wikipedia's licensing terms.

Size

Format: JSONL

Each record: one topic title and its cleaned summary

Dataset size will depend on the specific Wikipedia dump version used

Notes

This dataset is not an official product of Wikipedia or the Wikimedia Foundation. It is a processed derivative created for research and machine learning purposes.