The Mechanical Curator

ELAG 2014
The Surprising Adventures of the Mechanical Curator, and Other Tales
Ben O’Steen, technical leader of  British Library Labs

(see description of the talk and the slides)

This project started last year, as an accident! Taking the stuff that's technically accessible... and making it accessible! Engaged with the researchers, formally and informally through yearly competitions. What they win is our time and effort! The unifying theme to (pretty much) all the requests is: Give us everything! But this is quite depressing: so librarians don't take part in research, they're only there to provide content? Another theme is to have tools to interpret the content, to be able to work on broad sweeps of content rather than one at a time.

The Sample Generator shows the chasm between the collection and the digitised material. Not only is the content not as much digitised but it is also not as accessible as it could be.

The challenge was that research didn't want to work with api's but access large amounts of data. Made an experiment: face detection on 19thC illustrations - it wasn't very successful. The depiction is usually "clean" and posed, males represented differently from females and therefore less often detected etc. But it gave the idea of the mechanical curator, who digs in the collection of digitised images and tries to find visually similar images, based on a calculated match. It has now been doing that for a couple of months (and tweets about it). An unguided way of discovering material.

Images published on Flickr, many views in 4 days. They are published as CC0 and there are already examples of creative re-uses, such as colouring-in for children, an artit's interpretation etc. But this doesn't bring money to the Library, which is always hard to justify. But this is encouraging creativity, it may not be research but it's not less imporatnt. An animation student used images to represent them in 3-D. Moments, by Joe Bell

So the impact is hard to measure. Accessible is great, can we make it more useful? A group of UCL big Data CS students will be given access to all the book data, cloud computing and will make an experiment for broader and more direct access to the collections.

Data visualisation

ELAG 2014
Data visualization as a library service? Examples from Chalmers library
Stina Johansson, Librarian/bibliometrician at Chalmers Library, Sweden

(see description of the talk)

Father of Data? William Playfair (1759-1823), Scottish engineer and political economist. Developed first data charts. "The king at once undrestood the charts and was hihgly pleased. He said they spoke all languages..."

Graphics can make data much more "readable". A lot of communication is done through images. Chalmers Library uses a lot of visualisation to communicate about its data. E.g. topical analysis through keywords or citations networks or geospatial visualisations. E.g. Author co-citation analysis: a map representing the most cited authors, with links between them meaning various things, size or centrality of "dots" providing visual information.

Some of the visualisations are interactive - clicking brings to additional or linked info. Also visualisations of country level collaboration with environmental department data, for example.

See: http://chalmeriana.lib.chalmers.se/visuals/journal_citation/

Tools
Gephi: open source visualisation tool
Raw: open web app to create vector-based visualisations (from spreadsheets, potentially?) used on top of D.js.library (java) through a simple interface
VOSviewer: can use with a raw text file, easy to use

Data has to be clean and structured though! Be creative and play!