From collection search to collections as data
Tim Sherratt・@wragge・#HOTA2019
These slides:
seeing differently
search for 'radio' in Trove newspapers
comparing 'radio', 'telegraph', & 'wireless
search for 'radio' showing place of publication
words from newspapers describing origins of immigrants
objects in the NMA, by year of production
number of items described in NAA by top-level function
number of items digitised in NAA by top-level function
Australian aviators in Trove newspapers
White Australia policy records in the NAA
@TroveNewsBot
what is GLAM data?
- metadata (not content)
- structured text / data
- unstructured text
- images
- derived data
- user generated data
- activity data
- born digital data
some varieties of GLAM data
metadata
(data about collections)
seeing what we're not allowed to see
structured data
(think data with rows and columns)
GLAM CSV Explorer
unstructured text
(lots of words)
text from Trove journals
images
(of images and text)
3,471 Bulletin editorial cartoons
derived data
(data you extract from data)
redactions from ASIO files
user-generated data
(data added by the public)
where do you get GLAM data?
- harvest data from APIs...
- extract data from web pages by screen scraping...
but if it's not already packaged for download, you can...
🤯
it all seems too hard!
the GLAM Workbench is here to help!
Jupyter notebooks?
notebooks can be tools or hacks...
...or even simple apps
asking questions of data
what happened to radio in 1955?
exploring data
what can I do with the NMA API?
playing with data
what happens when you view interjections as tweets?
what becomes possible when I...?
one possible pathway...
- visualise searches in Trove newspapers
- find patterns, ask questions
- zoom in on points of interest
- harvest article text and metadata
- explore harvested data in detail