How this index is built

A national search cannot be built by crawling a domain zone. Two facts make that impossible, and both are specific to Dominica.

A .dm domain does not mean Dominica. The zone is sold worldwide, and a good part of it belongs to companies that simply liked the two letters. And most of Dominica is not in .dm at all — the island’s hotels, shops, churches and associations sit on .com, .org and .net, where no domain sweep will ever find them.

So the collection runs in two directions at once: from organisations to their websites, and from websites back to the organisations behind them.

From organisations to websites

The starting registers are the ones the country keeps itself: the Government Directory with every ministry, division and agency; the Companies and Intellectual Property Office; village and town councils; the tourism authority’s listings of hotels, guest houses and tour operators; the sector lists of schools, banks, churches and media. To these we add two open registers that know the island well — OpenStreetMap, where local shops and clinics are already mapped with telephone numbers, and Wikidata, which holds schools, parties, villages and stadiums.

An organisation without a website is not a dead end. Its name is turned into domain guesses, each guess is checked, and a site is only accepted when the organisation is actually named on the page — otherwise a nature reserve would be matched to a German plant nursery.

From the web back to organisations

The .dm zone is swept through Common Crawl and the public certificate logs, which together show the hosts that exist without asking the registry for a zone file. Then the link graph does the rest of the work: on a small island, a government site links to its agencies, the chamber to its members, a newspaper to the organisations it writes about. Every new host found this way goes through the same verification as everything else.

Dominica Confidence Score

Each website carries a score from 0 to 100, assembled from evidence rather than from a single rule. The weights below are the working values; every signal that fired is recorded next to the site, so any verdict can be read and argued with.

The subtraction matters as much as the addition. Search engines and directories mix up Dominica and the Dominican Republic constantly; without an explicit penalty, a national index of one country fills up with the other.

What is stored

Not a list of links — a database. For every organisation: name, type, sector, register and register number, parish and town, address, telephones, e-mails, social accounts, website, how it was discovered and how confident we are. For every site: zone, status, language, feeds, page and picture counts, the evidence behind its score. For every page: title, lead, full text, publication date. For every picture: its caption, alt, credit and the text around it — because in an island archive the caption is often the only place a name is written down.

Everything is downloadable: organisations and websites as CSV.

Manners

The crawler obeys robots.txt, noindex and crawl-delay, pauses between requests to the same host, and every pass is bounded — see the crawler page. Adult sites, gambling and pirate storefronts are not indexed.