1 of 31

Intro

Adventures with TEI:

adapting the EVT Viewer for use with Fedora

James Alexander

Digital Library Developer

Fedora Showcase

September 2023

2 of 31

Fun fact

I live in an old house near the middle of England’s newest city, Milton Keynes.

2

3 of 31

Open University

About The Open University

Distance-based learning institution

Largest university in UK

Established in 1969

208 000 students

6 400 tutors

> 17% students identify as having a disability

Over 1 000 free courses on Open Learn site

Produce (almost) all taught course materials

  • Campus in Milton Keynes
  • Library with physical Archive on Campus

3

4 of 31

Digital Archive

Open University Digital Archive

Collections

Diverse collections centred around teaching materials and focused on its large Audio Video Archive.

  • 10 000 video broadcast programmes (mostly BBC collaborations) & associated assets
  • 20 0000 radio programmes
  • Several hundred books and images
  • 3 000+ course websites
  • More…

  • Launched in 2014
  • Fedora 3.x backend (migration underway!)
  • PHP frontend and administrative interface
  • Includes online exhibitions of content

4

5 of 31

Sampson Low collection - Slide 4

Sampson Low collection

A collection of 19th Century letters

200+ letters collected by British publisher, Sampson Low. Includes letters from eminent 19th Century figures, includes politicians, social campaigners and writers.

The collected letters were pasted into two notebooks with annotations about the writer and/or subject. Focused on publishing, the letters include a wide range of subject matter.

Letter writers include – Dickens, Gladstone, Wellington, Robert E. Lee, Gustav Doré.

5

6 of 31

Collection challenge

The challenge: develop the collection for public and scholarly access

The ‘usual’ things (we’d anticipated) -

Digitise the letters and prepare derivatives

Transcribe the text

Give some historical and biographic context for the letters (including attribution where possible) as part of creating online exhibition

There were x2 critical decisions that shaped development in retrospect…

6

7 of 31

Key decision - description

Key decision 1

How to best describe the content? [Spoiler we used TEI]

The letters are quite messy!

Greatly varying standard of legibility of handwriting , but also the following issues:

  • Historic glue damage caused by adhesive used to fix letters into the two books
  • Areas of lost, damaged and illegible text for a variety of reasons including physical loss of paper
  • Text crossed out, underlined
  • Embossed/printed/franked text as well as hand-written text
  • Artwork/doodles embedded in text
  • Use of superscript and subscript writing
  • Pages where text has been written in lines horizontally and vertically on the same page

7

8 of 31

View of a letter

Typical Letter

  • A previous owner has written a brief note about the author (at the top)
  • You can see areas where the glue has seeped through the paper
  • The writing isn’t very easy to read
  • There are a few crossings out

8

9 of 31

A difficult to read letter

One of my favourites!

The writer has presumably decided to save paper or possibly postage(?) by writing first horizontally and then vertically across the same page

9

10 of 31

TEI - introduction

Solution 1 - use TEI markup

Text Encoding Initiative (TEI)

Xml markup for describing texts

© Creative Commons Attribution 3.0

  • See https://tei-c.org/

  • TEI has evolved as a markup language with ways to describe texts and in particular, ways to describe incomplete and illegible documents as well as their content, history etc

10

11 of 31

TEI – example missing text

Example –

how to describe missing text

11

12 of 31

TEI markup example

TEI markup for missing text

12

13 of 31

Problem 2

Problem 2

How to display content to make it as visually interesting and appealing as possible?

13

14 of 31

Parallel views of image and text

Key decision 2 – Parallel image + text view

Any solution must incorporate parallel aligned views of the photographic page and a version of the transcribed text displayed opposite each other in a “book like” view -

Image

Text

14

15 of 31

Leveraging available open source

How to leverage what is already available?

We have previously developed Fedora content for viewing with the Internet Archive’s BookReader, but needed a view with parallel display of image and text. Originally, we’d assumed that this would have to be further modified.

There are a several TEI Viewers, but most are designed to facilitate views of TEI as xml - not what was wanted.

Once we discovered it, there was only one candidate, EVT Viewer.

15

16 of 31

EVT Viewer view

EVT Viewer

16

17 of 31

Slide Title 8

EVT Viewer

Edition Visualization Technology

See http://evt.labcd.unipi.it/

Version 1.3 demo - http://evt.labcd.unipi.it/demo/evt_v1-3/dotr/

Continuously developed at the University of Padua since 2017 to view digital editions of manuscripts based on text encoded as TEI. The EVT Viewer was originally based upon the Internet Archive BookReader but has a much modified display and masses of advanced functionality based around its ability to display TEI markup as interface elements.

Functionality in addition to the parallel display of photographic image and text. -

The viewer incorporates the ability to switch from a diplomatic (“what you can see”) view to a critical/editorial view (includes interpretative material).

It would take a day to demo this adequately: there are more than thirty actionable controls on the main viewer screen alone.

Parallel text display and much, much more!

  • Source code can be found at GitHub. https://github.com/evt-project/evt-builder
  • Contact Roberto Rosselli Del Turco for further information.

17

18 of 31

EVT functionality

EVT functionality

Diplomatic View

The text as it appears

Creates togglable multiple views of the same content - Diplomatic and Critical views of text

Critical View

Text that includes editorial additions

18

19 of 31

Customising the EVT Viewer

EVT Viewer customisation

Vanilla modifications

Rebadging for Open University’s style guidelines

Toggling on/off interface elements

Alterations to basic templates (to simplify)

Accessibility & javascript changes

Numerous changes for keyboard (and other) accessibility requirements

Amended some interface actions and elements

19

20 of 31

See:

20

21 of 31

EVT main feature - 2

Further EVT Viewer customisation and prep work

Creating multiple viewers

From one to many

EVT is designed to display a single manuscript but the requirement was to show each letter in a standalone viewer: so viewer creation was modified to facilitate creation of 200+ instances with distinct urls.

Also, we wanted to hold the master copy of the TEI xml as a Fedora datastream and not as a standalone xml document.

21

22 of 31

Collection preparation

Collection preparation work

Standardising size of image displayed

We created access versions of the digitized master images with, as far as possible, standard display sizes, based on an image height of 1200 pixels, using ImageMagick® to generate these and associated thumbnails.

Transcription of letter texts

Transcriptions of the letter text matching exactly the end-of-line returns in the original were prepared. This greatly facilitated in creating the TEI markup automatically (which is marked up line-by-line).

Creating 200+ TEI documents

This is quite an undertaking – so we tried as possible to streamline creation of TEI xml documents…

22

23 of 31

EVT workflow

Normal EVT workflow to create viewer

TEI

xml

xslt

Static HTML pages

+

Usually the xslt transformation is performed in an xml editor, such as Oxygen

and this generates a series of static page templates and page fragments for display.

23

24 of 31

EVT workflow

EVT workflow modified to simplify creation of TEI xml

TEI fedora

datastream

xslt

Static HTML pages

+

HTML Input form

xslt

+

24

25 of 31

Input form

Input form for Letter

Simplifying creation of TEI documents for archive staff without an XML editing background

An input form with x24 fields including the transcribed letter text was created within the existing custom Fedora administration interface.

On submission, this creates a LETTER datastream with the form data. Additionally, this datastream is used to generate a TEI datastream that is a complete well-formed TEI xml document.

25

26 of 31

Modify TEI

The generated TEI datastream

View of the markup of the TEI xml document as a datastream

A TEI document including the TEI header element and the marked-up pages of text is generated as a datastream.

The resulting document can still be edited as text using an xml editor, such as Oxygen, before being copied back into the form and re-submitted.

When the TEI datastream is submitted as a form, elements of any pre-existing version of the html for the EVT viewer are first removed and then an xslt transformation is triggered using Saxon Transform to create all the necessary html elements for the EVT viewer from the TEI.

26

27 of 31

View of same letter in EVT viewer

27

28 of 31

Take aways

Take away

We were able to develop a robust and user-friendly interface for Fedora content with relatively little effort, leveraging the work of the EVT Viewer developers.

  • Dual view of image and text
  • Ability to switch diplomatic and critical/interpretative views of text
  • Page turner

Some of these features were things we regarded as essential -

Some features were highly desirable -

  • Magnifier
  • Manuscript description

Some features we never used -

  • The ‘link Image and Text’ - really cool feature, but no resource to draw and label boxes around every line!
  • Search within document – most of the letters were a few pages long

28

29 of 31

Next steps

Next steps

Currently no plans to release modified EVT Viewer as open source but if there was some interest, we would reconsider this.

Migration!

Demo link

29

30 of 31

Questions and Contact

Questions?

Thank you

james.alexander@open.ac.uk

31 of 31

Slide Title 31