Setting up a
GLAM Workbench
in your library
Image: State Library of Victoria, http://handle.slv.vic.gov.au/10381/342096
Tim Sherratt
@wragge
these slides
- Collections as data?
- The GLAM Workbench
- What is Jupyter?
- Types of Jupyter notebooks
- Notebook interfaces
- Running notebooks
- Your own GLAM Workbench
- Homework
This workshop
Collections as data?
See also: https://collectionsasdata.github.io/
SEARCH IS FAMILIAR
BUT What about the other 3m?
SAME search
DIFFERENT VIEW
MORE searches
different questions!
APIs
data dumps
CSVs
full text
images
collections
as
data
?
APIs
data dumps
CSVs
full text
images
collections
as
data
+
GLAM Workbench
Not just how, but why...
- possibilities – why should I be interested?
- starting points – can you give me an example I can use?
- pathways – where do I go next?
628
CSV files from Australian GLAM organisations
harvested from government data portals
CSV FILES
GLAM CSV Explorer
MY GLAM WORKBENCH
GLAM Workbench
- tools, tutorials, examples, hacks
- live code
- editable, reusable, hackable
- openly licensed
is
GLAM Workbench
- coding 101
- finished or perfect
is NOT
GLAM Workbench
Powered
by
JUPYTER
GLAM Workbench
Jupyter
- Computing in your browser
- A computational narrative – combine text, images, code & more
- A standard format – use on different platforms
- See Introduction to Jupyter Notebooks
Jupyter Notebooks
JUPYTER
LIVE CODE
in your browser
JUPYTER
Click to edit
SHIFT+CLICK
to run
JUPYTER
Gather data
Visualise results
JUPYTER
TRY IT YOURSELF
JUPYTER
Clicking the button will:
- create a customised computing environment in the cloud
- load the Jupyter notebook containing these examples – LIVE!
Examples
NOT all notebooks are the same!
Use Jupyter notebooks to...
- create tutorials
- develop tools
- harvest data
- share code snippets
- document an API
- visualise a dataset
- hack an interface
- move data around
visualise searches over time
A Tutorial
(asking historical questions with search facets)
A Tutorial
(asking historical questions with search facets)
visualise searches over time
A Tutorial
(asking historical questions with search facets)
visualise searches over time
A SIMPLE tool
(providing lots of nicely-packaged data)
Trove newspaper harvester
National Museum of Australia Collection
EXPLORE AN API
(collections in time & space)
EXPLORE AN API
(collections in time & space)
National Museum of Australia Collection
EXPLORE AN API
(collections in time & space)
National Museum of Australia Collection
National archives of australia
Harvest data
(no API? you can always try screen-scraping!)
Harvest data
(no API? you can always try screen-scraping!)
National archives of australia
White Australia policy in NAA
Harvest data
(no API? you can always try screen-scraping!)
save a newspaper article as an image
Hack Interfaces
(add a new download option to Trove)
save a newspaper article as an image
Hack Interfaces
(the same notebook as an app)
NOTEBOOK INTERFACES
Notebook interfaces
STATIC or LIVE?
- static interfaces only render the content of the notebook
- live interfaces also let you edit and run the code
GitHub
GITHUB provides a static view
NBVIEWER
NBVIEWER is static, but...
Jupyter can refer to either:
- the notebook standard (which is really just a JSON object)
- an interface used to create, edit, and run notebooks
- but Jupyter notebooks can be used in interfaces other than Jupyter!
two jupyterS?
The Jupyter interface comes in two flavours:
- the classic Jupyter notebook
- the new Jupyter Lab
Three jupyterS?
One notebook per window
Classic jupyter
Multiple tabs in a window & more!
jupyter LAB
ChangE the Jupyter interface
Extending jupyter
- iPyWidgets – adds interactive widgets
- Appmode (Classic only) – hides the code & runs all cells
- RISE – turns your notebook into a slideshow
- Voila – turns your notebook into a dashboard or standalone app
One notebook, Multiple views!
In the GLAM Workbench
One notebook, Multiple views!
In the GLAM Workbench
} STATIC
} LIVE
One notebook, Multiple views!
POSSIBLE PATHWAYS
} STATIC
} LIVE
Novice
One notebook, Multiple views!
POSSIBLE PATHWAYS
} STATIC
} LIVE
Learning
One notebook, Multiple views!
POSSIBLE PATHWAYS
} STATIC
} LIVE
Experienced
OtheR interfaces
RUNNING NOTEBOOKS
Running notebooks
You can run Jupyter notebooks on your own computer, or by using a cloud service such as:
- Binder
- CoLab
- SWAN (from AARNET)
- see this handy comparison for more options
run notebooks from GitHub
Binder
Open notebooks LIVE in binder
BINDER
Binder In GLAM WorKbench
The magic of Binder
Binder In GLAM WorKbench
- A single click to start
- Batteries included (no software for the user to install)
- Encourages experimentation
- Just try it...
Limitations of BINDER
Binder In GLAM WorKbench
- Limited resources
- Notebooks close when inactive
- Changes not saved
- Where do I go next?
Create your own GLAM Workbench
The BASIC Recipe
- an environment to develop notebooks
- a public GitHub (or similar) repository
- a configuration file (specifying the software your notebooks need)
- a Binder link
Your own GLAM Workbench
Developing notebooks
- set up a development environment on your own computer
- install Python (or Anaconda) to run Jupyter & create notebooks
- notebooks can use a variety of programming languages
- copy, edit, & reuse my GLAM Workbench!
Your own GLAM Workbench
PUBLISH Notebooks
- create a repository on GitHub (or similar) to save & share your notebooks
- create a README file
- add an open licence (to encourage re-use)
- add a configuration file
Your own GLAM Workbench
LINK to Binder
- copy the url of your GitHub repository
- go to Binder and paste the url in the box
Your own GLAM Workbench
LINK to Binder
- copy the Binder link or button code
- add the link to your GitHub README
Your own GLAM Workbench
LINK to Binder
Your own GLAM Workbench
REPEAT! EXPAND!
Your own GLAM Workbench
- create more notebooks, and more repositories
- create a documentation site to bring together the links
- share, share, share!
Try This!
HOMEWORK
- explore the GLAM Workbench, try the different links
- open the GLAM CSV Explorer & upload your own CSV
- fork the GLAM CSV repository & modify it to display your own datasets