Website Classification Dataset
This dataset includes extracted and cleaned texts from approximately 1,375 unique websites, classified into several categories. It is designed to train and evaluate automatic classification models for web content.
Approximately 1,375 websites, cleaned text files, CSV format
CC0: Public Domain
Description
The “Website Classification Dataset” contains textual data extracted from various websites, with each record associated with the original URL and cleaned text ready for analysis. The sites are divided into different categories according to their content.
What is this dataset for?
- Train web content classification models
- Develop filtering or recommendation systems based on site content
- Study the textual characteristics of websites in various sectors
Can it be enriched or improved?
The dataset can be enriched by adding new categories, updating texts or by including additional metadata (e.g.: HTML structure, associated images). Manual annotation can improve the quality of labels.
🔎 In summary
🧠 Recommended for
- NLP students
- Chatbot developers
- Text classification researchers
🔧 Compatible tools
- Scikit-learn
- Hugging Face Transformers
- SpacY
- Pandas
💡 Tip
Use a TF-IDF vectorizer combined with a linear classifier for fast and efficient results.
Frequently Asked Questions
Does this dataset contain non-text data?
No, it only focuses on textual content extracted and cleaned from websites.
Can additional websites be added to this dataset?
Yes, the dataset can be enriched with additional data and reclassified according to new categories.
Is it suitable for training multilingual models?
Mostly English speaking, but with an adaptation, it is possible to include other languages via pre-processing.




