Changelog du schéma
Pour des raisons de simplicité, le contenu de cette page n'est pas disponible en français.
Version 2022-04-15 ⟹ Version 2023-06-27
Image-Text-Pairs
is now available forformat
objects.- A new
automatically-aligned
boolean is now available to flag datasets resulting automatic alignment. Default value isFalse
.
Hotfixes since release
- (26/10/2023) Corrected
transliteration
's type declaration (boolean
, notbool
). - (26/10/2023) Fixed syntax (missing
}
)
Version 2021-10-15 ⟹ Version 2022-04-15
- Added free-text field
transcription-software
. Recommended values:eScriptorium + Kraken
Transkribus
OCR For All
script
is now a list of objects.- Basically, all script ISO values should become
- iso: IsoValue
such as- lat
->- iso: lat
- Added the ability to specify a little more the script with a new free text field
qualify
- Basically, all script ISO values should become
- Added an
institutions
object which reuses the pattern ofauthors
, except that it forcesname
and does not supportfirstname
orcid
is now available forauthors
objects- A new
characters
field is now available, it supports three sub-properties- A unicode normalization mode called
mode
, which can beNone
,NFD
,NFKD
,NFC
,NFKC
- A
transliteration
boolean to indicated whether script is transliterated (such as hieroglyphs): Default value isFalse
. - A
members
property which accepts a list of characters and represent all glyphs in the ground truth.
- A unicode normalization mode called
Hotfixes since release
- (21/11/2023) Corrected
transliteration
's type declaration (boolean
, notbool
). - (09/09/2022) In
volume
,count
's minimum value is0
instead of1
. - (05/09/2022) Fixed pattern for ORCID check.
- (30/06/2022) Removed
required
status forroles
. - (17/06/2022) Added
data-provider
as a possible value inroles
. - (17/06/2022) Added
institutions
object as distinct fromauthors
instead of its subtype. - (17/06/2022) Added
characters object.