Digital Manuscript Interoperability Via Shared Canvas

Digital Manuscript (DMS) Interoperability
Via Shared Canvas
A Case Study in Open Annotation
OA Rollout * Stanford * April 9, 2013

Tom Cramer Borrowing liberally from… and-> Ben Albritton
Chief Technology Strategist DMS Program Manager
Stanford University Libraries Rob Sanderson Stanford University Libraries
tcramer@stanford.edu Los Alamos National Labs blalbrit@stanford.edu
rsanderson@lanl.gov

Walled
Gardens:

Custom,
Func<onal,
Isolated

Consider

…an
art
historian
who
would
like
to
compare
and
annotate

illustra<ons
from
manuscripts
at
two
different

repositories,
available
via
two
different
interfaces

…a
codicologist
who
would
like
to
study
the
original
state

of
the
Codex
Sinai<cus,
fragments
of
which
are
now

dispersed
across
four
different
sites

…a
linguist
who
would
like
to
cite
the
first
wriIen
use
of

old
English
words
in
manuscripts,
referring
to
both
digital

images
and
transcrip<ons

…a
paleographer
who
would
like
to
analyze
scribal
hands

using
a
custom-‐built,
image
analysis
tool

Digital
Medieval
Manuscripts
Today:

A
World
of
Silos
&
Duplica<on

Parker App Rose App Gallica St. Gall App App X

Parker Rose BnF St. Gall X
Data Store Data Store Data Store Data Store Data Store

•  Dis<nct
applica<on
func<onality

•  Separate
data
stores

•  Unique
data
models

•  Diﬀerent
(or
no)
API’s

Dis<nct
Concerns

Build
useful

tools
and
apps

Scholars
Want:
Users
&

resources

Find,
Use,

Tool

Analyze,
Annotate
Makers

Want:
Mix
&

Match,
Best
of

Breed

Repositories

Host,
Preserve
(and
Enrich)
Resources

Want:
Use
of
Resources,
Enriching

services,
Enriched
content

API’s
Enable
Reuse

Applications Dictionary of TPEN
Parker App
User interfaces Old English Transcription Tool
present DMS data
with specialized Image Des- Image Image Trans-
Des-
tools and contexts Delivery cription Delivery Search Delivery criptions
cription

Import & Export
DMS apps and tools can both
API's consume and contribute data
I's
AP
RESTful via common web services. E.g.,
Web Services transcriptions or annotations
can be deposited in the source
repository, enriching the corpus
Repository Parker Data
for future researchers.
Canonical Data Store
for Images, Description,
and TechMD
Images MD Annot.

API’s
-‐>
Framework
-‐>
Ecosystem

Annotation & Cross Repository Search
Transcription Tools & Delivery Environments
Parker App Gallica
Image Trans- Image
Delivery criptions Delivery Search
Image Des- Image Des-
Delivery Search cription Delivery Search cription
I's
I's

I's

I's
AP
AP

AP

AP
Parker Data Oxford Data Rose Data BnF Data

Images MD Annot. Images MD Annot. Images MD Annot. Images MD Annot.

Specific
Use
Cases
for
Interoperability

1.  Drive
a
page
turner
with
a
sequence
file

(harder
than
you
might
think!)

2.  Enable
3rd
party
transcrip<ons:
Relate

transcrip<ons
to
pages,
and
to
regions
of

interest
(ROI)
on
pages
(text
blocks)

3.  Enable
3rd
party
annota<on:
Relate

annota<ons
(scholarly
commentary)
to

specific
manuscript
elements
–
pages,

images,
text,
marginalia,
etc.

DM
–
Image
Annota<on
Tool

T-‐PEN:
Transcrip<on
Tool

DMS
Interop
Par<cipants

Universi<es
Libraries,
Archives
and
Labs

•  Stanford
*
•  Bri<sh
Library
*

•  Cambridge
•  BNF
*

•  Oxford
*
•  e-‐codices
*

•  Harvard
•  Los
Alamos
Na<onal
Lab
*

•  Yale
*

•  Johns
Hopkins
*

•  St
Louis
*

•  Drew
*
*
ac<vely
building
tools

•  Ghent

•  Oregon

with
support
from
the
Andrew
J.
Mellon
Founda7on

Technical
Approach
to
DMS
Interop

+

Shared
Canvas

Data
Model

IIIF
Image

Delivery

IIIF
Image
Delivery
API

hIp://library.stanford.edu/iiif/image-‐api

Requirements
for
Data
Model

1.  Reﬂect
the
physical
and
structural
complexi<es
of

medieval
manuscripts

–  Enable
ordered
delivery
of
(page)
images
and

associated
(transcrip<on)
texts

2.  Relate
mul<ple
(and
poten<ally
overlapping)

images,
texts
and
other
resources
to
a
single
page

3.  Reliably
relate
addi<onal
annota<ons
(scholarly

commentary)
to
the
manuscripts
(whole
item,

range
of
pages,
page-‐level,
or
speciﬁc
region)

4.  Express
via
an
easy
transforma<on
of
exis<ng

descrip<on
&
resources

Naïve Approach: Annotate an Image

CCC
26
f.
iiiR

SharedCanvas: Collaborative Manuscript Facsimiles
Open Annotation Collaboration, July 26-27, Chicago IL, USA

Naïve Approach: Annotate which Image?

CCC
26
f.
iiiR
Fold
A
Open


Naïve Approach: Annotate which Images?

CCC
26
f.
iiiR
Fold
A
Open
Fold
A
and
B
Open


Naïve Approach: …?!

CCC
26
f.
iiiR
Fold
A
Open
Fold
A
and
B
Open
f.
iiiV


Naïve Assumption: Everything is Square

Non-‐rectangular
objects

•  Fashionable
heart
shaped

manuscripts

•  Fragments

•  Pages
with
foldouts

Facsimile
of
BNF
Rothschild
2973

hDp://www.omifacsimiles.com/brochures/montchen.html


Naïve Assumption: Image Equals Page
Parts
of
Pages:

•  Only
illumina<ons
digi<zed

•  Fragments
of
page

•  N.B.
Informa<on
about
non-‐
digi<zed
parts
may
s<ll
exist

Mul<ple
Pages:

•  Open
spread

•  Fragments
from
mul<ple

pages
in
single
image

Cod.
Sang.
1394:
10.5076/e-‐codices-‐csg-‐1394


Naïve Assumption: Image Exists

The
page
may
not
be
digi<zed
at

all

•  Not
"interes<ng"
enough
This
page
inten<onally,

•  Digi<za<on
destruc<ve
but
unfortunately,

lej
blank

•  Page
no
longer
exists

•  Page
only
hypothe<cal

•  Lack
of
money,
<me,
…

Countless
manuscripts,
all
around
the
world!


Naïve Assumption: Only One Image

Alignment
of
mul<ple
images

of
same
object

•  Mul<-‐spectral
imaging

•  Mul<ple
resolu<ons

•  Mul<ple
digi<za<ons

•  Image
<ling

Bonus
naïvete:

•  Page
has
one
orienta<on

•  Page
has
one
text

Archimedes
Palimpsest
Mul7-‐Spectral
Images

hDp://www.archimedespalimpsest.org/



Canvas Paradigm
• 
A
Canvas
is
an
empty
space
in
which
to
build
up
a
display

• 
Well-‐known
concept,
eg
HTML,
PDF


Canvas Paradigm
• 
A
Canvas
is
an
empty
space
in
which
to
build
up
a
display

• 
A
SharedCanvas's
top
lej
and
boIom
right
corners
correspond
to
the

equivalent
corners
of
a
[rectangular
bounding
box
around
a]
page


Canvas Paradigm: Implementation

Empty
space
is
not
hard
to
implement!


Painting the Canvas?
Requirements:

• 
Need
to
allow
distributed
associa<on
of
resources
with
the

Canvas,
or
part
of
the
Canvas

• 
Any
type
of
resource,
or
part
of
a
resource,
should
be
able
to
be

painted

• 
Need
to
allow
users
to
comment
about
the
Canvas,
or
part
of
it,

or
any
of
the
resources

Solu<on:

• 
Associate
resources
and
commentary
using
Open
Annota<ons!


Annotations to Paint Images
Open
Annota<on
with
single
Target
(Canvas)
and
Body
(Image)


Annotations to Paint Images
An
oa:Choice
with
a
default
and
1
or
more
op<ons
for
the
image.


Annotations to Paint Images: Implementation


Annotations to Paint Text
Speciﬁc
Resource
with

Selector
to
deﬁne
an

area
of
the
Canvas

And
also
need
Text

Selectors
to
describe

the
appropriate
part
of

the
text...


Annotations to Paint Text


Annotations to Paint Text: Implementation


Image Segments


Image Segments: Implementation


Audio Segments


Audio Segments: Implementation


[Scholarly] Commentary


[Scholarly] Commentary: Implementation


Ordering: Missing Pages


Ordering: Rebinding


Non Medieval Use Cases


SharedCanvas Data Model


In
Summary

•  DMS
Interoperability
requires
common

approaches
to
serving
data
and
images

•  Shared
Canvas
provides
an
open,
extensible

data
model
based
on
OAI-‐ORE
and
Open

Annota<on
constructs

–  Using
the
original,
physical
object
as
the
absolute

frame
of
reference

•  New
scholarly
work
produced
in
OA-‐compa<ble

tools
add
to
and
enrich
the
scholarly
record

–  Crea<ng
a
virtuous
circle

–  And
truly
transcending
silos!

Using
OA
Adds
to
the
Ecosystem

Annotation & Cross Repository Search
Transcription Tools & Delivery Environments
Parker App Gallica
Image Trans- Image
Delivery criptions Delivery Search
Image Des- Image Des-
Delivery Search cription Delivery Search cription
I's
I's

I's

I's
AP
AP

AP

AP
Parker Data Oxford Data Rose Data BnF Data

Images MD Annot. Images MD Annot. Images MD Annot. Images MD Annot.

Digital
Manuscript
Interoperability

hIp://lib.stanford.edu/dmm

Shared
Canvas

hIp://shared-‐canvas.org

IIIF
(Int’l
Image
Interop
Framework)

hIp://lib.stanford.edu/iiif

Digital Manuscript Interoperability Via Shared Canvas

More Related Content

Viewers also liked (20)

Similar to Digital Manuscript Interoperability Via Shared Canvas (20)

More from Tom-Cramer (8)

Digital Manuscript Interoperability Via Shared Canvas