sphita/Multi-Domain-Tech-Scrape-Preview

This is a 750-row high-fidelity sample of our premium, multi-domain web extraction dataset. It is engineered specifically for AI teams building Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) pipelines.

🛒 Get the Full Dataset

Ready to scale your AI models with production-grade data?

👉 Buy the Full Dataset Here

---

📊 Dataset Overview

This sample demonstrates the schema, parsing quality, and metadata richness you can expect from the full commercial dataset. The data is scraped from high-value technical domains (e.g., Python documentation, Meta engineering, Intel blogs).

Key Features:

• Clean Markdown Extraction: HTML noise is stripped. Code blocks, lists, and text formatting are preserved as pristine Markdown.
• Rich Metadata: Includes Open Graph (og_*) tags, canonical URLs, language detection, and exact word counts.
• Ready for RAG & Fine-Tuning: Standardized and immediately usable for vector embeddings and contextual retrieval.

🛠️ Schema

Field	Type	Description
url	string	The crawled URL
final_url	string	The resolved URL after redirects
status	int	HTTP status code
tier	string	Domain classification tier
fetch_time_ms	int	Timestamp of extraction
bytes	int	Size of the payload
content_hash	string	Hash of the content for deduplication
links	list[string]	Array of outgoing URLs found on the page
title	string	Extracted <title> tag
description	string	Meta description
og_title	string	Open Graph title
og_description	string	Open Graph description
og_image	string	Open Graph image URL
canonical	string	Canonical URL for the page
lang	string	Detected ISO language code (e.g., en)
word_count	int	Total word count of the extracted text
html	string	Raw HTML payload
markdown	string	High-fidelity Markdown conversion of the page content

⚠️ Limitations

This is a restricted subset (750 rows) provided strictly for schema validation, pipeline testing, and parsing evaluation.

Unlock the full, multi-terabyte dataset today.