Mehriddin1997/lex-uzbek-laws

Lex.uz — “Xavfsizlik va huquq-tartibot muhofazasi” (Partial) Dataset (Uzbek)

Dataset Summary

This dataset contains Uzbek legal texts collected from the lex.uz portal, specifically from the “Xavfsizlik va huquq-tartibot muhofazasi” (Security and law enforcement protection) section.
The collection is partial: the full section has not been fully scraped yet, and the dataset may be expanded in future updates.

The dataset is intended for:

• domain adaptation / fine-tuning of Uzbek language models on legal text
• retrieval-augmented generation (RAG) corpora
• legal NLP experiments (classification, summarization, QA, etc.)

Data Source

• Primary source: lex.uz (Uzbek legal documents portal)
• Section: Xavfsizlik va huquq-tartibot muhofazasi
• Note: This dataset includes only the subset of documents successfully collected during scraping runs.

Collection Method

Documents were collected by crawling lex.uz pages and extracting the main legal document content.
Some pages may contain template/UI fragments (e.g., navigation labels). A cleaning step is applied to reduce such noise.

Dataset Structure

Each record is a JSON object (JSON Lines / .jsonl) representing one document.

Typical fields:

• id: document identifier extracted from the lex.uz /docs/ URL (string)
• url: source URL on lex.uz
• lang: language code (expected: uz)
• title: document title (may be missing for some items if the page layout did not expose it)
• text: cleaned main text content of the document
• source: constant string "lex.uz"
• Optional fields (if present):
  • okoz: OKOZ category string extracted from the page (if available)
  • tsz: TSZ category string extracted from the page (if available)

Example

{
  "id": "6588801",
  "url": "https://lex.uz/uz/docs/-6588801",
  "lang": "uz",
  "title": null,
  "text": "Oʻzbekiston Respublikasining Qonuni ... (cleaned text)",
  "source": "lex.uz",
  "okoz": "[OKOZ: ...]",
  "tsz": "[TSZ: ...]"
}