open-index/umi-pages-2026w35-00

umi-pages

Part of umi, an open web crawl published as Parquet. Before you use any of this, read the exclusion list at open-index/umi-meta and filter the rows it names. Published files are never rewritten and never deleted, so the exclusion list is how a takedown reaches you, and applying it is a condition of using the data rather than a suggestion.

One row per page umi fetched. The extracted text as markdown, the title, the description, the headings, the outlinks with their anchor text, the sixteen response headers doc 11.5 keeps, and the duplicate detection sketches. There is no raw HTML in here and there never will be, because doc 10.2's arithmetic does not survive storing it and doc 07.8 caps how long we may hold it.

Columns

The schema is doc 10.5's, mapped one to one into Parquet with nothing renamed and nothing flattened.

column	type	null	what it is
url	string	no	The URL we asked for, canonicalised.
final_url	string	yes	Where the redirects ended, null when that is the same place.
url_key	binary(10)	no	The 10 byte fingerprint of url, which is what deduplication and the frontier key on.
pld_id	binary(8)	no	The 8 byte id of the pay level domain.
host	string	no	The host as text.
fetched_at_ms	uint64	no	When the fetch finished, Unix milliseconds.
status	uint16	no	The HTTP status, or zero if we never got one.
outcome	uint8	no	How the fetch ended: 0 ok, and the other codes cover not modified, gone, blocked, failed and off domain redirect.
tier_used	uint8	no	Which rung of the fetch ladder produced this row, 0 through 4.
tier_path	list of uint8	no	Every rung tried, cheapest first, so a reader can see what a page cost.
content_type	string	yes	Content-Type as sent.
content_length	uint32	no	Body length in bytes after transfer decoding, which is not always what the header said.
lang	binary(3)	yes	The BCP 47 primary subtag, padded to three bytes.
body_digest	binary(32)	no	blake3 over the body bytes.
chunk_root	binary(32)	no	blake3 tree root over 16 KiB leaves of the body, so a range can be checked without the whole file.
extract_digest	binary(32)	no	blake3 over the extraction, which is what makes an extraction reproducible.
markdown	string	yes	The extracted text. Null when there was no body, and null when the page asked not to be indexed.
title	string	yes	The title, after the usual boilerplate trimming.
description	string	yes	The meta description or its open graph equivalent.
headings	list of string	no	h1 through h3 in document order.
snippets	list of struct	no	The same strings the title, description and heading columns hold, tagged by kind, for a reader who wants one column instead of four.
links	list of struct	no	The outlinks: absolute href, anchor text, the rel bits packed, and what kind of link it was.
headers_kept	map of string to string	no	The sixteen response headers doc 11.5 allows, and nothing else. No cookies and no credentials, ever.
content_usage	string	yes	The site's AIPREF Content-Usage preference, verbatim. Read this one.
minhash	binary(256)	no	64 MinHash values over the text shingles, for near duplicate clustering.
simhash	uint64	no	A 64 bit simhash over the same shingles.
text_bytes	uint32	no	Plain text length, since the plain text itself is not stored.
link_count	uint32	no	How many outlinks, so a filter does not have to decode the list.
fetcher_id	binary(32)	no	Which fetcher delivered it. The matching receipt carries the signature.
verification	uint8	no	How far doc 06 got: arrived from a known fetcher, or corroborated by a second one.
robots_checked_ms	uint64	no	When the robots.txt behind this fetch was last read.
crawl_profile	uint32	no	Which crawl profile queued the URL.

How this was made

Extractor umi/0.0.1, canonicalisation canon/1, against the spec at docs/spec. Rows are written in fetch completion order, so the fetched_at_ms statistics on a row group are the ones worth pushing a filter down to. Files are around 128 MB and a day folder holds one day of them.

umi crawls to build a public, openly licensed web index, and that is the whole declaration. We do not train models on the corpus, we do not run an agent that browses for a user, and we do not resell access. The crawler identifies itself as umi and its bot page is https://umi.dev/bot. The corpus is open, so what anyone else does with it is out of our hands, which is a fact worth stating plainly rather than eliding.

Robots and Content-Usage

Every page in the corpus was fetched under RFC 9309: the host's robots.txt was read first, a Disallow that matched was obeyed, and a robots.txt we could not read at all disallowed the host rather than allowing it. On the pages schema robots_checked_ms says when the file behind that decision was last read.

content_usage carries the site's AIPREF Content-Usage preference, from the response header or the robots.txt, verbatim and unparsed. We record it and we do not act on it, because it is the publisher's statement to whoever reads the corpus and not ours to interpret on their behalf. If you are training on this data, that column is the one to read.

A robots snapshot is only good for a day. RFC 9309 caps a cached robots.txt at 24 hours and umi honours that, so the robots corpus is a record of what a host said when we asked, not a standing permission you can crawl against today. Use it to plan, to see which hosts have rules at all, to pick up Crawl-delay and Sitemap before you queue anything, and then ask the origin yourself.

Corrections

We never rewrite a published file and we never delete one, because both break every checksum anyone recorded and make old work unreproducible. Corrections go on an append only exclusion list in open-index/umi-meta under blocks/, naming a repository and a file and either a row predicate or a set of url_key values, with a reason and a date. Apply it as a filter when you read.

Every day folder has a manifest under _manifest/ listing each file with its digests, and a detached Ed25519 signature next to it. The signing keys are in the same meta repository, so a manifest can be checked without asking us for anything.

Licence

The split is real and worth stating plainly. Everything umi created is CC0: the annotations, the quality signals, the duplicate sketches, the link structure, the manifests and the schemas. The extracted page content is third party material that we did not create and cannot license. It is published on the same basis Common Crawl has published on for over a decade, with per row provenance, the publisher's own stated preferences carried in the row next to it, and a takedown process that works. Putting a single licence tag on the repository and hoping nobody looked closely would be the dishonest version.

Takedown

Mail tamnd87@gmail.com with the URLs or the host. A request is honoured by adding the rows to the exclusion list, which takes minutes rather than a release cycle, and by adding the host to the crawler's block list so it is not fetched again. You do not need to explain why.