bedead/website-industry-13m

Dataset Card for 13M+ Website Domains with Industry Labels

This dataset contains over 12 million website domain URLs mapped to their associated industry categories.
It is useful for domain classification, industry prediction, NLP preprocessing, clustering, and machine learning research.

This dataset card has been generated based on the Hugging Face dataset card template.

---

Dataset Details

The dataset has two columns:

• website: Domain URL (cleaned, no DNS tags)
• industry: Associated industry or category (with ~5% null values)

---

Dataset Sources

• Repository: Kaggle Dataset
• Paper [optional]: N/A
• Demo [optional]: N/A

---

Uses

Direct Use

This dataset can be used for:

• Training and evaluating domain → industry classifiers
• NLP preprocessing for URL/domain-related models
• Business intelligence (e.g., mapping company presence across industries)
• Clustering & categorization of domains

Out-of-Scope Use

• Misuse for phishing detection or malicious targeting of industries
• Commercial exploitation of the dataset beyond research/educational purposes
• Inference of sensitive or personal data (none exists in the dataset)

---

Dataset Structure

The dataset consists of a single table with 13,642,052 rows and two fields:

• website: Cleaned domain URL
• industry: Associated industry or category (nullable, ~5% missing values)

Example:

{
  "website": "example.com",
  "industry": "Technology"
}

Dataset Creation

Curation Rationale

The motivation was to create one of the largest open datasets mapping web domains to industries for advancing research in web classification, NLP, and ML-driven categorization.

Source Data

Data Collection and Processing

• Scraping — Website URLs were collected from publicly available internet sources.
• Cleaning — DNS tags and non-relevant parts were removed.
• Industry Mapping — Mapped via metadata, keywords, and context.
• Final Dataset — Consolidated into two columns (website, industry).

Who are the source data producers?

• The data comes from publicly available websites.
• No private or proprietary datasets were used.

Annotations [optional]

Annotation Process

• Industries were mapped using rule-based and contextual matching techniques.

Who are the annotators?

• The dataset was created and curated by the author (Satyam Mishra).

Personal and Sensitive Information

• No personal, sensitive, or private data is included.
• Dataset consists only of publicly available domains and general industry categories.

---

Bias, Risks, and Limitations

• ~5% missing values in the industry column.
• Some industries may be ambiguous or mislabeled due to heuristic mapping.
• Domains without clear metadata may not have precise industry labels.

Recommendations

• Users should clean and validate industries before downstream ML tasks.
• Consider re-labeling ambiguous or null industry entries with custom classification.
• Dataset is best suited for research and educational purposes.

Citation

If you use this dataset, please cite:

BibTeX:

@misc{satyam_mishra_2024,
  title={12M+ Website Domains with Industry Labels},
  url={https://www.kaggle.com/dsv/9734207},
  DOI={10.34740/KAGGLE/DSV/9734207},
  publisher={Kaggle},
  author={Satyam Mishra},
  year={2024}
}

APA:

Mishra, S. (2024). 12M+ Website Domains with Industry Labels. Kaggle. https://doi.org/10.34740/kaggle/dsv/9734207