Sign in
The code expires in a few minutes and can be used once.
Claims
OCR
Service
- API
- —
- PostgreSQL
- —
- Typesense
- —
- OCR engine
- —
Latest sync
- Indexed documents
- —
- Last updated
- —
- Next sync fetches
- —
- Centre changes
- —
Claims
—
A download applies the filters above and covers every matching row, not just the page on screen.
—
| Cover |
|---|
Release Workflow
For a digitized book with no claim. A claimed book is entered from its claim instead: Claims › Ready for Release.
| Book | Centre | Location | Pages | Added | Assigned to | Released |
|---|
Steps 1 and 2 are for reference: Release Management fills them in. Your part is Step 3, Abstract & Blog.
The URL slug from Step 1: the folder of page images.
Which page to use as the cover.
Reads every page with Sarvam once the viewer is attached; the text joins the viewer as it is read. Unticked, no Sarvam credits are spent.
Metadata Master
| Name | Value used | Status |
|---|
Settings
Settings will appear here.
Users
Sign-in is an emailed code, so the address is the account: a user without one cannot sign in at all.
The first one ticked is the default.
| Name | Role | Centre | Status |
|---|
Generate a Tify viewer
The folder name is the book's dcterms:identifier in the catalogue and
the folder of page images in storage — they are the same string.
Which page to use as the cover. Defaults to the first.
Replaces any Tify viewer already on the item, because the catalogue caches a copy of the manifest when the viewer is attached. Skipped when no item matches the folder name — the manifest is still generated.
Reads every page with Sarvam in the background once the viewer is published, and adds each page's text to the viewer. Unchecked, no OCR runs and no Sarvam credits are spent; any text an earlier OCR published for this book is kept.
OCR
Recent generations
| Folder | Status | Pages | Started by | Manifest | OCR |
|---|
Bulk generation
One folder per line. To choose a cover other than the first page, put the page number after the folder name — separated by a comma, a tab or a space, so a two-column paste from a spreadsheet works as it is.
Attaching replaces the item's existing viewer and deletes its old HTML viewer. That cannot be undone, so run it as a dry run first and read the report.
—
| # | Folder | Status | Pages | Note |
|---|
Recent batches
| Started | By | Folders | Mode | Status |
|---|
Re-generate IIIF manifests
Each book is rebuilt and its viewer re-attached, whatever its timestamps say and whether or not it already has a viewer.
Both days included. Only the title, summary, metadata and homepage change; books that already match the catalogue are skipped. A refreshed viewer is re-attached, so from then on it counts as created on the day of the refresh.
| Include | Folder | Title | Created |
|---|
How many manifests one run may rebuild — mostly ones generated before Thulyam. The scan stops at this number, so its summary describes exactly what the run would do.
—
| # | Folder | Status | Pages | Note |
|---|
Recent regenerations
| Started | Type | By | Items | Status |
|---|
Search index
- Private items
- —
- Last updated
- —
- Next sync fetches
- —
- Omeka
- —
- Centre changes
- —
Recent syncs
| Mode | Status | Fetched | Indexed | Took | By |
|---|
Thumbnail hashing
A perceptual hash of each catalogue thumbnail, used to compare a photographed cover with the scanned one. Nothing ranks by these yet: the plan gates that on image matching lifting Recall@5 by at least two points, measured on a gold set.
- Hashed
- —
- Used for ranking
- —
- Last run
- —
Hashing fetches every thumbnail over the network, so a first run across the whole catalogue takes a while. It is resumable: run it again and it picks up what is left.