tshasan/multi-label-web-categorization

Multi-Label Web Page Classification Dataset

Dataset Description

The Multi-Label Web Page Classification Dataset is a curated dataset containing 14,258 web page titles and snippets, extracted from the CC-Meta25-1M dataset. Each entry has been automatically categorized into multiple predefined categories using ChatGPT-4o-mini.

This dataset is designed for multi-label text classification tasks, making it ideal for training and evaluating machine learning models in web content classification and natural language processing (NLP).

Source & Processing

• Source: The dataset is derived from CC-Meta25-1M, a multilingual web dataset.
• Classification: ChatGPT-4o-mini was used to classify each web page into one or more categories.
• Categories: The dataset includes the following 11 categories:
  • News
  • Entertainment
  • Shop
  • Chat
  • Education
  • Government
  • Health
  • Technology
  • Work
  • Travel
  • Uncategorized (for ambiguous cases)

Each record includes metadata features, such as the number of assigned categories and the text lengths of the title and snippet.

Dataset Structure

Format: CSV

The dataset is provided as a single CSV file with the following columns:

Column Name	Type	Description
url	string	Web page URL
title	string	Page title
snippet	string	Short excerpt of webpage
language	string	Language code (e.g., en, cs, ru, de)
category	string	Comma-separated list of assigned categories
category_list	string	List representation of categories (e.g., ["Work", "Education"])
num_categories	int	Number of assigned categories
title_length	int	Character count of the title
snippet_length	int	Character count of the snippet

Usage & Applications

This dataset is well-suited for:

• Training multi-label text classification models.
• Developing content recommendation systems.
• Enhancing search engine categorization.
• Analyzing web content distributions across languages.

To load the dataset using Hugging Face’s datasets library:

from datasets import load_dataset

dataset = load_dataset("tshasan/multi-label-web-classification")
print(dataset[0])  # Display the first sample

License

This dataset is licensed under CC BY 4.0, allowing free use and modifications with proper attribution.

Acknowledgments

• tshasan – for creating the CC-Meta25-1M dataset.
• OpenAI – for providing ChatGPT-4o-mini, which powered the classification.
• Common Crawl: A large-scale web crawl dataset providing openly available web data.