Discovery and canonicalization
Domains enter through public datasets, editorial research, and community submissions. We normalize internationalized domains, follow redirects within strict limits, group regional domains, and retain evidence for material facts.
Classification and translation
Automated models may propose categories, descriptions, and translations. Structured validation, confidence thresholds, and human review gates prevent automated publication of high-risk or ambiguous records. A site has one primary and at most three secondary categories.
Ranking
Exact domain and verified name matches lead. Relevance, evidence quality, freshness, availability, and aggregate popularity follow. A diversity pass reduces repeated organizations, countries, languages, and site types. Payment is not an organic ranking signal.
Freshness
Published sites are scheduled for checks at least every 30 days. Material content changes invalidate affected translations and return uncertain records to review.